Want to see the answers to previous questions?  Here’s the blog archive; the pre-blog newsletter archives are here.

Q: One of the first things I am doing in my new job is to identify the Customer Lifecycle pattern – how many periods (month, year) will it be before a customer is likely buy again.  In enterprise software industry, where software cost easily 6 figures, # of years is a reasonable time frame.

A: Yes, one would assume this.  But these notions would most likely be based on a feeling of the “average” behavior, and on average, it probably does take a long time.

What is not known is this:  if the “average” is composed of short-cycle and long-cycle buyers, who are the short cycle buyers, and what are they like?  What industry SIC code, for example?  And can we get more of them, or at least focus more resources on them, if they are the most profitable?  So the challenge is not only to look for the “average”, but then understand how this average is composed.  If you can break down the average by industry, or by salesperson, for example, this might be highly directional information.

Q: From my internal analysis, however, I discerned from the sales figures something quite counterintuitive – the period between first and next sale is much shorter than I would have thought for the SW industry in general.

A: Pleasant surprise, eh?  I don’t know what kind of figures you are looking at, but make sure the data is in fact what you think it is.  For example, if you want to study software purchase itself, do the “sales” figures you are looking at include transactions involving not the sale of software, but also service, like installation or modification fees?  It would make sense that a “software sale” would be followed pretty quickly by an “installation” sale, for example.  You need to know this to properly segment.

Q: Would you be able to point me to some studies on how often customers wait after the first purchase before contemplating an upgrade of software or something you personally have done in consulting projects for SW companies?  This industry benchmark will then shed some light on whether this trend is something peculiar to our company or not…

A: I am not aware of any published study of this type.  And as you might imagine, these numbers would vary quite widely in the industry and the nature of the information would be a highly guarded corporate secret.  So I don’t think you will find any “benchmark” studies of this type, and sorry, I can’t share client data with you!

Q: The next step for me then is to map out the drivers for this behaviour and then calculate the LTV (LifeTime Value) and take a look at the actual LifeCycle events creating this LTV.

A: Yes, but to be precise, LTV is typically a forecast when working with current customers; it’s not known until the customer actually defects, marking end of  “Life”.  What you are probably looking for is more accurately called “Life to Date” (LTD), the actual sales of a customer from start of relationship until present.

Also, when segmenting customers by LTD, of course look for temporal bias – a 10-year customer is likely to have higher LTD than a 2-year customer.  If there are enough customers, it might be a good idea to first segment by start year, then LTD.  This way you have cohorts of customers who are going through the same experiences together and differences in LTD will be more significant in terms of predictive power – you don’t have to hunt around for external bias (e.g. competitive changes) that might affect LTD.

Think about what I said above about breaking the “average” down into different groups, because this will likely provide the Eureaka! moment and turn the data you are looking at into information.  For example, if you find the LTD of the “average customer”, this is very interesting information indeed, but not highly actionable – what “action” do you take knowing this information?  Can you point to or predict which customers are “average”?

However, if you were to find out the LTD differed dramatically by industry, by salesperson, by country, by time of year, by type of software module installed first – this is highly actionable information, because it provides very direct instruction on where the most profitable areas of business are.

If you lack thoughts in this area, try segmenting by variables that directly affect the experience of becoming a new customer.  At least one of these is most always predictive of LTD, when directly tied to the acquisition of a new customer:

1.  Campaign media (e.g. trade show versus magazine, online versus offline)

2.  Campaign content / offer

3.  Salesperson / Service teams

4.  Product or Category of first purchase

5.  SIC code / Industry (proxy for Product suitability to customer needs)

For example, if you find LTD differs by salesperson, you will find salespeople who create high LTD customers and salespeople who create low LTD customers; the company should study how each salesperson sells and teach the others based on results.

Or perhaps likelihood to purchase again is determined by which customer interface team installed the software – one team does such a good job the customer re-orders the next module very quickly, as compared with other teams.

Knowledge of this type would be extremely valuable to the company – you can use LTD to discover “best practices” hidden within the “average” data, and by spreading those best practices throughout the company, create enormous benefit and increase in profitability.

The secret to creating meaningful customer analysis that delivers high impact is this: always think about what you would **do** with the information you uncover.  If you can’t **do something** with the numbers you evolve, you probably need to drill down a little further and uncover the true meaning of the underlying data.

Hope this helps.  My personal guess based on what I know about behavior would be this: the type of software installed first and the sales / installation / after the sale care team are the two most likely variables predicting speed to next purchase, followed by the industry the buyer comes from.

This is a very interesting project; please keep me informed of your progress and I will help you in any way I can.

Jim

Have a question on Customer Valuation, Retention, Loyalty, or Defection? Go ahead and send it to me here. If on the topic above, you can leave a comment on the post:

Segmentation by LTD & LifeCycle

Want to see the answers to previous questions?  Here’s the blog archive; the pre-blog newsletter archives are here.

Q: I visited your website because I am trying to understand how to develop a customer LifeTime Value model for the company that I work at.  The reason is we are looking at LTV as a way to standardize the ROI measurement of different customer programs.

Not all of these programs are Marketing, some are Service, and some could be considered “Operations”.  But they all touch the customer, so we were thinking changes in customer value might be a common way to measure and compare the success of these programs.

A: Absolutely!  I just answered a question very much like this the other day, it’s great that people are becoming interested in customer value as the cross-enterprise common denominator for understanding success in any customer program!

If I am the CEO, I control dollars I can invest.  How do I decide where budget is best invested if every silo uses different metrics to prove success?  And even worse, different metrics for success within the same silo?

By establishing changes in customer value as the platform for all customer-related programs to be measured against, everyone is on an equal footing and can “fight” fairly for their share of the budget (or testing?) pie.  By using controlled testing, customers can be exposed to different treatments and lift in value can be compared on an apples to apples basis – even if you are comparing the effect of a Marketing Campaign to changes in the Service Center.

But are you sure you want to use LifeTime Value for this application?

Q: From what you stated on your website, I will not be able to develop a LifeTime Value model unless I understand the customer Lifecycle.  The customer lifecycle is something that I could get a good understanding from using doing a RFM analysis.

My question is, once I complete the RFM analysis, what would be my next steps in developing a customer LifeTime Value model?   At this point in time, the hardest thing that I am trying to wrap my head around are the variables to include in the model.  I visited Arthur Middleton Hughes’ website:

http://www.dbmarketing.com

and he suggests the following variables (download spreadsheet, if interested):

http://www.dbmarketing.com/special_ltv.htm

Jim, could I simply use those variables going forward to calculate the LifeTime Value of a customer at my company?  I would appreciate any assistance you may be able to provide to me on this matter.  Thanks.

A: Well, that’s a big tangle of related issues!    Let’s unpack first, then answer the question.  First, the relationships between these ideas:

Lifetime Value versus Lifecycle – LTV is a number, LifeCycle is a trend over time that contains trigger events.  You don’t need the LifeCycle to develop (calculate) LTV, you need the LifeCycle to most efficiently and profitably act on and manage LTV issues.

RFM versus Lifecycle – RFM is a tactical model that is a “snapshot” of customer state at a point in time, the customer’s likelihood to respond.  Frequently used names for these customer states include active, lapsing, lapsed, defected.   Lifecycle is the “movie” one might put together from these snapshots of RFM states; the migration from one customer state to the next are the Lifecycle trigger points.

Now, let’s make sure we understand each one of the ideas:

LifeTime Value

Strictly speaking, LTV is not a very flexible concept and is best used for determining how much you can spend to acquire a customer and still make a profit.  This is the equation that Mr. Hughes has provided, a man by the way that I have a lot of respect for.  His model is quite detailed and useful for the purpose of finding break-even cost to acquire a customer.

To use Arthur’s LTV model, you have to find historical values and plug them in.  You could assume nothing will change and the LTV of certain segments of past customers will be the same; this is great for “benchmarking”, for example.  However, this approach is not measuring LTV, it’s predicting LTV based on historical data.  This is fine, and a valid method for certain types of analysis.

But, the premise of your question is you will be testing, and testing implies something new will occur.  So while you could use LTV to estimate results, you’d have to wait quite a while to prove the results one way or another.  LTV is really “forensic” in this way – you won’t know the final answer until the customers defect.

You could certainly go back 2 – 5 years after the tests, and prove one group had higher LTV than another, but that’s not typically a very useful approach when doing testing.

RFM (Recency, Frequency, Monetary)

RFM is a predictive model that takes a “snapshot” of the customer base and gives you a score for each customer, a prediction of likelihood to respond relative to all customers.

By itself, RFM doesn’t tell you if you are making money or not.  It is used to classify the “state” of customers at a point in time, usually for targeting purposes – are they active, lapsing, lapsed, defected?  In other words, it’s a customer segmentation tool.

For example, RFM could be used to choose your test and control groups for a campaign using Lift measurement – you would want test and control to have the same range and balance of scores.  In fact, one of the tragic campaign measurement mistakes people often make is not taking into account the likelihood to respond when selecting test and control groups, resulting in biased test results.

Customer LifeCycles

One of the great features of RFM is the idea of “ranking” customers relative to each other; this gives allocation of budget and success measurement a standard to follow.  A single  customer can have many different scores over the course of their LifeTime, with the likelihood to respond the score at a specific time.  In fact, if you looked at RFM scores over time for a single customer, you would have a clear understanding of the LifeCycle of a customer – the most powerful segmentation available in terms of message and offer targeting.

The problem with looking at RFM scores over time is complexity; the beauty of individual customer scores at a single point in time becomes unbearable when you are talking 125 different scores on 50,000 customers over 6 months.  That’s the internal or analytical problem.  Externally, this kind of information is extremely gnarly to present and explain to senior managers, it’s presentation hell.

The way I solve this problem is with a tool I call LifeCycle Grids.  The Grids takes the same fundamental drivers used in the RFM model and instead of ranking, uses thresholds or “hurdles” to classify customer states.  This creates a standardized customer LifeCycle “dashboard” so comparisons of customer value between different segments can be made more easily.  It works for both short and long term observations and is easy to represent either numerically or graphically.  And because it uses finite thresholds for activity rather than ranking, the same calculations that create the dashboard can be used to actually drive or trigger actions.

So the dashboard is actually the controller as well.  This is extremely beneficial in terms of linking presentations, plans, and results. People can literally point to a segment on the LifeCycle framework and say, “Let’s deliver message X to each person from segment Y who enters this cell” and see the results right where they pointed when the dashboard is updated.

Once you test some ideas and find out which approach generates incremental profits for a cell in the Grid, you can automate delivery of the program as customers enter that cell of the Grid.  This is the classic “sense & respond” approach to marketing communication – right message, right person, right time.

The LifeCycle Grids are demonstrated in a lot of detail for different applications in the series here, but probably of most interest to you as it relates to customer analysis, see here.

And now, to answer your question:

Which approach above, if any of these, would be best for standardizing measurement of ROI in widely diverse customer programs?

LTV would be appropriate if what you want to know is breakeven cost to acquire.  Since we are talking about customer programs, I doubt that’s what you want to use.  Plus, if you want a hard number rather than a prediction, you could be waiting a long time for the answer.

RFM is a “snapshot” model and so not really suited to long-term studies of customer value.

Customer Lifecycle models are more likely to be involved in the execution of a program, not the success measurement.  LifeCycle tracking could be (and often is) used to predict the financial success of campaigns before they have run their course, but you’re only predicting success, not delivering numbers into an ROI model the CFO would accept as “fact”.

Answer: None of the above.

What you need is an approach designed for the task, which in this case, is:

Lift Measurement or Near-Term Value

Lift is a measure of the performance of a test group of customers compared with a control group of similar customers who are not exposed to the test.  You can read more about control groups here.  In the analysis of value contributed by each group, many of the same values from Arthur’s LTV model are used – product margin, costs of program, fulfillment costs, payment parameters, etc.  However, if you are talking about a program to existing customers, cost to acquire is probably not relevant, though you might use source (campaign) to segment your test approach.

Lift is typically measured at intervals, say every 30 or 60 days, to see how test versus control populations are tracking, and can continue after the test is over to pick up residual value created in the customer.  However, this is not a Lifetime Value measurement, Lift models measure incremental contribution to LTV created by the Marketing, Service, or Operations program execution.

This means if you get lift from program test versus control, when you go back 2 – 5 years later and measure true rather than predicted LTV – after the customer has defected – you should in fact see the LTV in the test group higher than in the control group, barring any radical downstream difference in customer experience between test and control.  In this way, Lift models are actually predictive of changes in LTV.  That’s why the output of Lift models is sometimes referred to as the measurement of “Near-Term Value” and used much more often than the forensic approach of waiting for customers to defect.

Summary

All the above are core concepts in customer value measurement and management.

LTV is a measurement of net financial value contributed by a customer, and Lift measures  are like a “time slice” of the overall LTV curve.

LifeCycles are a management framework for programs designed to affect LTV, and models using Recency, Frequency, and Monetary are used to look at a “time slice” of the LifeCycle.

LTV can generally be increased in two ways: by creating more value during the existing LifeCycle, or by extending the LifeCycle.  Marketing (including Product) is typically used when doing the first, Service and Operations – customer experience and satisfaction – are largely what affects the second.

So it is completely appropriate to establish a unified approach to the measurement of customer programs intended to increase the value of a customer across all these disciplines, in order to ensure the allocation of  scarce resources to highest and best use.

A great question, and for a great cause!

Jim

Update:

Listrak asked me to do a podcast with them on these and related topics, check it out (MP3 link) here, or see list of all their Email Marketing Today podcasts here (I’m on Episode 42).

Have a question on Customer Valuation, Retention, Loyalty, or Defection? Go ahead and send it to me here. If on the topic above, you can leave a comment on the post:

LTV, RFM, LifeCycles – the Framework

Want to see the answers to previous questions?  Here’s the blog archive; the pre-blog newsletter archives are here.

Q: Thank you for your recent article about Control Groups.  Our organization launched an online distance learning program this past August, and I’ve just completed some student behavior analysis for this past semester.

Using weekly RF-Scores based on Recently and Frequently they’ve logged in to courses within the previous three weeks, I’m able to assess their “Risk Level”– how likely they are to stop using the program.  We had a percentage who discontinued the program, but in retrospect, their login behavior and changes in their login behavior gave strong indication they were having trouble before they completely stopped using it.

A: Fantastic!  I have spoken with numerous online educators about this application of Recency – Frequency modeling, as well online research subscriptions, a similar behavioral model.  All reported great results predicting student / subscriber defection rates.

Q: I’m preparing to propose a program for the upcoming semester where we contact students by email and / or phone when their login behavior gives indication that they’re having trouble.  My hope is that by proactively contacting these students, we can resolve issues or provide assistance before things escalate to the point they defect completely.

A: Absolutely, the yield (% students / revenue retained) on a project like this should be excellent.  Plus, you will end up learning a lot about “why”, which will lead to better executions of the “potential dropout” program the more you test it.

Q: However, in light of your newsletter, I realized that we should probably have a control group with whom we do NOTHING (just as we did this past semester) in order to prove the effectiveness (or not) of the program.

A: Correct.  Otherwise, you won’t be able to make a valid claim to the “saved students”. People can always argue a variety of other factors were in play – seasonality, topic, course sequence, etc.

Q: Since the actual number of students is confidential, can you please tell me what percentage you would use for a control group if we had 400, 800, 1200, 1600, 2000, 3500, or 5000 students?  You mentioned 10% in your newsletter, but the population you were referring to exceeded millions.

A: Well, there are online calculators you can use confidentially, example right here.

If you don’t understand the variables they are asking for, explanations at bottom of page, though this is very simple – what is confidence level and interval plus population size.

Q: Our population is MUCH smaller, and each customer is therefore even more critical.  I don’t want to recommend an unnecessarily large control group that would prevent us from retaining future students when we could see they were having trouble.

I suspect that our defection rates will be lower 2nd semester than 1st since students should be beyond the “learning curve,” so I don’t think we can justly say that the program alone is the reason for lower defection rates if we don’t use a control group.

A: Yes, well, this desire to “get as much test as we can” was the main point discussed in the newsletter.  And that’s the challenge with very small populations – to hit statistical confidence levels at say population = 500, you need over 300 or so in control.

Not so great.

So we go back to the question of company culture and how intuitively confident people will be with the results.  Do they in fact need true statistical significance for a program like this?

There is a way around the significance issue – repetition. The stats part of this is all about the “likelihood you get the same results again” – real important for drug testing, not so much for 500 folks in a marketing program.

The question you need to ask: do you really need “prediction”?  Or does prediction just make the whole test more complex and expensive than it’s worth?  What if you repeated the test a couple of times and got roughly the same results, is that “proof”?

Here is what I might do.  I would ask whoever needs to believe in the results of this test a question like this:

“Let’s say we took a random 20% sample of the students and excluded them from the marketing.  We apply the marketing to the other 80% and their retention rate is 15% higher than the 20% who had no marketing. We do this test 2 more times and the retention rate of students in the test is 13% and 17% higher than the students in the 20% who do not receive the marketing.  Would you at that point believe that without question, the marketing drives at least a 13% improvement in retention among students?”

Do you see where I’m headed with this?  The more times you repeat the test, the more confident you will be in the results – regardless of sample sizes and statistical mumbo jumbo. At some point, the reality of the differences between test and control performance has to be accepted.  It may help to define up front how many repetitions the “boss” needs.

There are two clues to help you evaluate the validity of your results / how many times you need to repeat the test to be “confident”.

One clue is the variability of the results – the more inconsistent the results are, the more likely the data is “noisy” and the more times you need to repeat the test to be confident.

If the spreads between test and control for the first 3 tests are 20%, 5%, and 10%, then you’ll need more repetitions of the test to get a good feeling for the actual impact.  If the results tend to cluster as in the example above (15%, 13%, 17%) then you can be more confident earlier in the test series the actual impact is somewhere around 15%.

The other clue is in the “spread” between test and control.  If the spread is consistently  ”wide”, say +10% (or more), this provides additional confidence a positive impact is being made.  The result over a series of tests may not actually be +10% (confirm by repeating the test), but it’s more likely to be positive.  If you consistently get a spread more like 1% or 2%, it’s more likely the actual result could be zero or negative and you need to keep repeating the test to gain confidence you have a positive result.

In the end, you may not want or be able to repeat the test enough times to know with statistical confidence what the result is.  But if the spread between test and control is wide and consistent, and the cost relative to the benefit is small, then does it really matter if there is statistical confidence?

For example, if you can make the statement you’re confident the program generates at least $10 in profit for each $1 invested, does it really matter if the statistically confident  number is $11 or $12 profit for $1 in cost?  We’re doing Marketing here, not drug testing.  There is an opportunity cost (profit left on the table) to not rolling out a program based on a test with results like this; rather than repeat the test to death just to be more confident I’d roll it out and continue to monitor the results.

One more tip, on this idea of sequencing / semesters / experience with the program.

There is no doubt in my mind that 2nd semester students would have what is called a “survivor bias” and be less likely to drop out; you will get the best performance in a program like this with 1st semester students.  So if at all possible, run the test / control on only 1st semester students , or segment by semester.

But, just because you run it on only 1st semester students does not mean you don’t have an effect in 2nd semester.  Continue to follow test and control into 2nd, 3rd, 4th semesters and you may see the dropout rate of the original 1st semester group continue to widen versus control.

This is not only great for the profitability of the initial 1st semester program but also provides you the baseline you have to beat (control) for those 2nd, 3rd, 4th semesters.  When you decide to see if you can have an additional effect by intervening in those periods, you’ll have 2 groups: those affected by Marketing in the 1st semester, and those new to any Marketing intervention.

My guess: a 1st semester intervention will have tremendous impact, both then and throughout the 4th.  The impact of intervention at each subsequent semester will diminish compared with acting in 1st semester, as will the “tail” value created over the student life, since the number of months left in the student life is shrinking each semester.

Hope that helps!

Jim

Have a question on Customer Valuation, Retention, Loyalty, or Defection? Go ahead and send it to me here. If on the topic above, you can leave a comment on the post:

Control Groups in Small Populations

Want to see the answers to previous questions?  Here’s the blog archive; the pre-blog newsletter archives are here.

 Q:  I am a big fan of your web site and read your Drilling Down book. Great work!

A:  Thanks for the kind words!

Q:  I was wondering if you could help me picking the right control group size for a project of ours?  The population is 25 million telco customers that for which we want to do a long term impact analysis (month by month) in regards to revenue increase versus control group.  The marketing initiatives are mix of retention, lifecycle and tactical/seasonal activities.  We want to measure revenue increase through any of the marketing activities compared to control group.

A:   Great project, this is the kind of idea that can really improve margins if you can find out which specific tactics drop the most profit to the bottom line.

Q:   I have searched the web for some help and found calculators that say: On 25 million and smallest expected uplift of 0.1% and highest likely rate of > 5% the calculator gives 250k (1%).  Is that sufficient to calculate the net impact on the remaining base?  Would be very grateful if you could give me your thoughts.

A:  Well, it could be and might not be…

If you were manufacturing widgets, where the outcomes are clear (unit is defective or not defective), you might use this approach to the question.  But in Marketing we’re talking about human behavior, and there is quite a lot more variability in outcomes and more room for interpretation.  You can encounter a number of problems down the road by running a control so “tight” to the statistically correct size.

From a practical perspective, when you do a test of this magnitude (and I assume strategic importance), you don’t want test to just “beat control”, you want to beat control beyond a shadow of any executive’s possible doubt.

From personal experience, I can tell you that executives tend to be non-believers with a 1% control versus a 5% control or a 10% control. So some of this control size choice is culture-based – if the exec team is a bunch of engineers that understand / believe in statistical sampling methods, then 1% is probably OK in terms of believing the results are predictive of future events.

But if you need to convince a CFO or somebody who will be working from gut or risk management rather than “science” then 1% may not be enough, there is too much perceived “room for error” with a 1% sample (even with the science).

This is in effect a “perceived confidence interval” argument – the difference between 95% confidence and 99.999% confidence. Engineers may be OK with 95% because they intimately understand the derivation of it; CFO’s not so much.  CFO’s may not even understand the math behind confidence but intuitively, they perceive that 10% control is “more likely to be accurate” than 1%.

Said another way, do you want people to argue about the math and stats and waver on their belief in the outcome, or do you want them to just look at a simple chart of test versus control numbers and say, “Congratulations, that’s a tremendous success!”.  A 10% control gets you complete agreement on the results without any quibbling.  At 1%, you may get “what about the chance we are wrong” arguments.

Now, there are financial implications to using very large controls – some positive (reduced expense), and some negative (potential revenue foregone).  So choosing control group size can be impacted by these other issues.  In small population tests these financial impacts are usually quite small, so negligible and I always go for large controls.

But in a population of 25 million, maybe not so.

Which brings us to the second consideration -  segmentation or “drill down” after the test.

Nothing is quite so painful as gearing up for a test of this magnitude, producing a stunning positive result on a “macro” basis across all initiatives, and then having the execs ask, “What is the driving force behind this increased profitability in the test group?  Is it retention, lifecycle or tactical / seasonal?”  Or as often happens in telco (usually from an ops GM or VP), “What was the result of this test in my region or on my platform?”

Uggghh…

With a 1% control across the entire population, you frequently are “boxed in” when it comes to sub-populations because you lose significance (both perceived and scientific) as you drill in.  You may be OK on a couple of large scale events on large populations, but as we know, every answer begs another question and you can run out of statistically significant answers pretty quickly.  If you use a large control at the macro level, you are (as a rough example) 99% confident at the macro level, 98% confident one segment down, 97% confident two segments down, 95% confident three segments down, etc.

One way to handle this is to build the test from subsegments up to the macro level.  Let’s say at a minimum you want 3 subsegments in the test – retention, lifecycle or tactical / seasonal – and each of these you want to be 95% confident in.  Since some of these programs are triggered by behavior (lifecycle) and some by calendar (seasonal) I’d guess the sizes of the populations and number of executions could be vastly different.  Meaning, you may only need 1% control on the seasonal promotions but more like 5% or 10% control for some of the lifecycle stuff to be 95% confident on the outcomes of those.

When you sum all these segments up, you often end up with more like 2% or 3% of the entire population in control groups to always be at least 95% confident at all the desired subsegments, which means you end up with even higher confidence at the macro “all campaigns” level – a very good thing.

And much better than trying to explain why you can’t answer a subsegment question because you used 250K instead of 400K or 600K in the control group, if you know what I mean!  That’s when people forget the arguments about foregone revenue and start saying stuff like “Why did you not use a larger control group for this test?”

In the end, you will thank yourself again and again for using a larger than minimum required control at the macro level because you WILL come up with that unexpected “must know” question and be thrilled to find out you actually can answer it at a decent level of confidence.

Good luck with it, let me know what you learn!

Jim

Have a question on Customer Valuation, Retention, Loyalty, or Defection? Go ahead and send it to me here. If on the topic above, you can leave a comment on the post:

Choosing the Size of Control Groups

Want to see the answers to previous questions?  Here’s the blog archive; the pre-blog newsletter archives are here.

Q: You kindly clarified a few issues when I was reading Drilling Down earlier this year – so I hope you don’t mind the direct email.

A: Yes, I remember!

I am working for www.XYZ.com, a social networking / virtual world site based abroad but visitors are 85% US.

Our growth up to now has been mainly viral and in the summer we hit 1.2M UVs operating on the Freemium model with only 5% of our registered users converting to paying customers and a significant portion of our revenue coming from ads.  On average our customers are active on the site for something like 4 months making their first purchase around day 28. 

But to take us to the next stage we are embarking on some marketing for the first time using AdWords and various revenue share campaigns, and of course to do this sensibly we need to arrive at a reasonable estimate of LTV.

A: Makes sense!

Q: To calculate an adjusted LTV I removed all customers with a lifetime of less than 4 months but this gives a low estimate as this calculation ignores the bumper summer months and the extra paid for features put in place earlier this year.  Calculating LTV using ARPU and monthly churn (not sure how to calculate this in our environment) gives another different estimate.  Is there any help or advice you could perhaps give us?  If not in the US then perhaps you could recommend somebody abroad – can’t find anything in the literature relevant for start-up like us.

A:  It sounds to me like you’re trying to make this too complicated, at least for the place you are at this time.  Monthly churn and the “28 day” threshold are nice to know on a tactical level, but LTV is more of a Strategic idea that does not necessarily benefit from analysis at that level.  And you may not really want LTV, but a derivative that might be more helpful.

Let’s say the average user sticks around 4 months.  Say also that you generate revenues of $1 million over that period, and 1,000,000 users had some level of activity.  So your revs per active user are $1.  In terms of generating net revenue, you want to acquire  users for less than $1.

Now, we know that number is topline, and obviously there are expenses.  Companies like yours do not have very straightforward financial models because of the large amount of R & D that may be capitalized rather than expensed.  So you need to go to someone in Finance and determine what the right number is to use for looking at ROI.  Ask them, what percent of our revenues are left over to pay bills?  Or, what number would you like to see increased through a Marketing program?

Is it cash flow? Earnings before Interest, Taxes, Depreciation, Amortization?  Gross Margin?  Some other?  Then, what percent of sales does this number generally run at?

Let’s say it’s 40%.  In other words, 40% of revenues is actually available to pay bills and so forth.  So in the example above, .4 x $1 = 40 cents, which is the max you can pay to acquire a user, and anything less than that generates money to pay bills.

This method of course looks at all revenues. Not sure why you would want to look at it any differently, since even users that don’t “purchase” still generate ad revenues.

But let’s say you want to be more specific, you care only about buyers and only want to run campaigns that generate buyers.  In other words, the advertising revenue is “nice to have” but you want to build out the paid marketing model based on the acquisition of visitors likely to purchase.  You can run the same model above, but only look at the known the buyer group.

Take any 4 month period, find revenues from purchases and divide by number of people purchasing (not purchases, but individuals who purchase, revenue / user).  Then apply the same 40% flow through from the model above, and that’s the max you can pay to acquire a buyer.

When you aggregate known buyers, segment by source and you will find different campaigns generate different kinds of buyers; some will stick around longer than 4 months, some less.

Segment these folks by campaign and run the same model as above, purchase revenue for the period they stick around divided by purchases times the 40% flow through.  That’s the max you can pay to acquire a buyer in that segment.  So you end up with (just guessing) being able to pay 5 cents for campaigns that generate people who stick around 2 months, 15 cents for people who stick for 3 months, and 40 cents for people who stick 4 months.

You can make LTV equations very complex, but often the point of the exercise is not really “what is the customer LTV?” it’s “how much should we spend to improve cash flow?” or something similar.  This is a much easier question to answer and often what the company really wants to know.

Said another way, it’s very difficult and often dangerous to peg an LTV number in a dynamic environment because there are so many potential changes that will impact it; LTV is a number you may fully understand 2 – 5 years from now.  Until then, you need something “close” that drives the same kind of thinking and action, and the approach above will get this done for you.

As things evolve, this number (called “flow through”, recently have heard it called “Near Term Value”) will basically approach true LTV as you extend the number of months in the measurement period.

At the point where your business is somewhat static at an operational level, you can then look for true LTV by examining the revenue of actual defectors.  This is the only way to really peg LTV – after users have left and sufficient time has elapsed where you do not believe they will come back by themselves.

Until then, you are better off trying to figure out how much you can pay to attract high  quality user / buyer segments.

Jim

Have a question on Customer Valuation, Retention, Loyalty, or Defection? Go ahead and send it to me here. If on the topic above, you can leave a comment on the post:

Customer Value in the Freemium Model

Want to see the answers to previous questions?  Here’s the blog archive; the pre-blog newsletter archives are here.

Q: I have recently purchased your book Drilling Down and going through the many interesting concepts.

A: Thanks for that!

Q:  I work for a membership Organization and we would like to conduct some analysis into who we may lose and approach them even before their membership lapses.  But the only problem here is that we carry data only on the purchases made (though many of our members do not purchase our products and stay a member) and web site visits.

A:  Are you *sure* that’s all the data you collect?  I once worked with a professional membership org that thought they only had one data source, but turns out they had 8 – from 8 different areas of the org – that nobody really knew about.

Q:  How do I know if a particular member is going to resign and lapse soon with this limited amount of behavioral data.  Recently it’s been a concern that we are losing members who have been with us for more than 10 years and who are in their mid career profession (aged between 30 to 45) and indicated no specific reason for resignation. 

This has been going on for the last few months and now we would like to strategically target these customers and approach them even before they react negative.  What concepts could help me to do this? Your guidance would be much appreciated.

A:  OK, my answer will be in two sections: if you (hopefully) find you have more data than you think, and if you really don’t have any other data to fall back on.

The first thing you should do is make sure you don’t have any other data sources.  Purchases and web sites visits might be the most obvious, but do you have:

Membership Campaigns
Conference registrations
Customer Service incidents
Local or regional meetings
Training sessions
Speakers
Orders for training materials

and so on.  In some orgs a lot of this is siloed  or even outsourced, but the data is still there – it’s a question of whether you can access it.   Think through the business model, and think about any possible interaction points with a member.  Then ask, Where would this data be?  You might be surprised at what you have.

Once you have more data, subscription relationship analysis of this type and predicting churn come down to three main thrusts:

1. Number of activities – similar to banking customer analysis, we often find that the number of different areas a member has tangible contact with is predictive of retention or defection.  For example, some members attend the annual conference and others do not.  Some members participate in local meetings and others do not.  Some buy training materials and others do not.  Some members do all three, some two, some one.

On average, the members engaged in all three activities are the most likely to remain members relative to those engaged in only two.  And members engaged in only two activities are the most likely to remain members relative to those engaged in only one.  This is a “run rate” kind of retention, an expectation.  From a Marketing perspective, this means you want to always be adding to the number of activities a member engages.

2. Change in number of activities – when you see a member drop from 3 to 2 activities, this is a clear signal that there is a problem of some kind, it’s a dis-engagement from the org.  You want to take action when you see these events, find out what is happening and if there is anything you can do to correct this.

The reason driving this downgrade may be a soft incident, say a content problem, or may be a hard incident, like payment problems. Either way, the org needs to find out if the issue can be addressed.  If you start to see “clustering” of these kinds of downgrades in relationship quality, it’s likely something more systemic is going on.  Often you will see certain segments who exhibit similar problems.

For example, members acquired through a certain publication may exhibit similar downgrade behavior at roughly the same interval from joining.  This is evidence of a systemic problem – something about folks from this source is unique, and for whatever reason, the org is not satisfying their needs.

3. Predicting change in activities – if you want to go further down this road and actually *predict* a change in the number of activities before it happens, you can look at the Recency within that activity.  A member who goes to conferences on a regular basis who then skips one is in danger of defecting from that activity – for some reason, the conferences are not providing the value they used to.

Or, something has changed with the member, their position in the LifeCycle has moved.  The conferences still provide the same value as they did before, but this value is shrinking for the member who (perhaps) needs more challenging or different content.

The same could be said for attending local meetings or buying training materials, etc.  For a known user of a specific activity, how long has it been since they used it?  Does this non-usage break an established pattern of usage?  If so, you have a triggering event for a marketing / membership intervention.

OK, so what if you really don’t have any more data?  You can create it.  One of the easiest and least intrusive ways to do this in a member org is with surveys.

Why?  Membership orgs have embedded permission to interact with members; it’s the nature of being part of such a group. What I mean by this is asking members to take part in a survey is not only quite natural, it’s often expected and perhaps even appreciated.  After all, what could be more aligned with a membership org than asking the members where the org should be headed and where they would like to see it go?

By thinking through and properly crafting such a survey, one should not only get a sense of potential friction points in the org overall but also get a sense at the individual level of which members are becoming dissatisfied and more likely to defect.  Implementation of a program like this means, of course, that the survey responses are tracked at the individual level so that action can be taken at the individual level. You can’t just do a random popup on the web site to make this work.

Moreover, in the out years, one can reverse engineer the reasons for defection.  Each year, analyze the population of members who defected in the previous year, and ask yourself, How are these people similar?  Do they come from similar backgrounds or industries?  Did they join in the same year? Come from the same marketing source?  You may already have such survey data and simply had not thought of using it for analyzing and predicting which members would defect.

Finally, I am fully aware that often, bringing these kinds of issues to the table can be painful for a membership org.  In my experience, many of these orgs are highly politicized and there is a certain “we’ve done it this way for 100 years” attitude.  When you present data that contradicts long-held beliefs there can be tension.  And this is fine - when the org is on board with the project.

So my final advice would be this – whether you have the data or generate it, the success of a project like this can revolve around the strong commitment of the org to actually do something about member defection. Ask this question of management first:

If we find members are leaving because of something we are doing or not doing, will we change this, even if the problem contradicts long-held beliefs?  The answer will show you how committed the org is to fixing member retention.

“It depends” is not the answer you want to hear from these senior players, because this means they don’t have a retention problem, they have an acquisition problem.

Unless they are willing to change, they need to recruit more members that are not like the members they have if they want to decrease defection in the membership.

Hope that helps!

Jim

Have a question on Customer Valuation, Retention, Loyalty, or Defection? Go ahead and send it to me here. If on the topic above, you can leave a comment on the post:

Member Retention in Professional Orgs

Want to see the answers to previous questions?  Here’s the blog archive; the pre-blog newsletter archives are here.

Q:  I am hoping you can help answer a question for our team.  By way of introduction, I am the CEO of XXXX.  We are a specialty retailer / restaurant of gourmet pizza, salads and sandwiches.  We would like to know  restaurant industry averages (pizza industry if possible) for customer retention – What percentage of customers that have ordered once from a particular restaurant order from them a second time?  I am hoping with your years of expertise and harnessing data you may be able to assist us with this question.  Look forward to hearing from you.

A:  Unfortunately, in those said years of experience, I have found little hard information on customer retention rates in QSR and restaurants in general (if anyone has data, please leave in Comments).  It’s just the nature of the business that little hard data, if collected, is stored in such a way that one can aggregate at the customer level.  The high percentage of cash transactions doesn’t help matters much; there’s a lot of data missing.

Over the years, sometimes you see data leak out for tests of loyalty programs, and of course clients sometimes have anecdotal or survey data, but this is not much help in getting to a “true” retention rate.  More often than not you discover serious biases in the way the data was collected so at best, you have a biased view of a narrow segment.  Often what you get is a notion of retention among best customers, or customers willing to sign up for a loyalty card, but not all customers.  And the large “middle” group of customers is where all the Marketing leverage is.

What to do about this predicament?  

There are really two issues in your question; the idea of using industry benchmarks when analyzing customer performance, and the measurement of retention in restaurants.

As far as industry benchmarking, two things:

1.  Annual reports for publicly traded eateries may be of help.  Customer loyalty info may be disclosed in these documents or conference calls with Wall Street.  Still, it will probably be of the quality referenced above – narrow in scope or behaviorally biased.

Sometimes you can put snippets of different conversations into an equation that allows you to guess at repeat purchase rate; hospitality analysts often want to understand repeat behavior and do this kind of forecasting.

2.  Ignore the industry benchmarks.  If you have the capability to track repeat rates, simply establish what they are now and use them as internal benchmarks to not fall below or create programs to improve against them.  

Frankly, I tend to discourage using “industry benchmarks” because the kinds of businesses that can really leverage repeat behavior and retention (customer-centric model) are usually *different* from the industry, so using a benchmark (say, from Domino’s) is probably low-balling your potential.  

Not that Domino’s is a “bad” operation, mind you, but they are what they are, they tend to be more on the operational excellence side of the game than customer intimacy (that’s what we called the customer-centric / social approach back in the early 90’s). 

Product leadership, the 3rd value discipline, is pretty much table stakes for anyone in the restaurant biz, and I assume from your business description you just might consider this a primary focus which you then leverage to create power in the intimacy area.  This is essentially the Apple Strategic model.  If the product is not great, the love will not come.

My point is this: without understanding the value discipline and Strategy of a competitor, you can’t know if any benchmark is something you want to compare to, because the business may have a completely different focus than yours.  Worse, using industry averages simply hides any real information you might gain that is actionable for your business.

For example, even though Walmart and Nieman Marcus are in the same business, I don’t think anyone would say they have the same Marketing Strategy or core value proposition.  Walmart is of course the poster child for operational excellence with the end result being value pricing, which flows to the advertising content.  There’s nothing “wrong” with this approach, it simply is what it is, and customer intimacy / relational / social marketing simply doesn’t really fit here.  You certainly can try to be as intimate as possible; but it must be done within the constraints of the model and not reduce operational excellence.  Importantly, this is a “mass” concept, so Push media is the most effective.

Sam’s Club is an example of how one might accomplish this mix.  A “membership” is certainly more customer intimate and allows customized communication, a key component of customer intimate execution.  Again, this flows into the advertising content.  Sam’s gets to leverage the Walmart infra, so they can at the same time maintain a decent level of operational excellence.  Remains to be seen if they could do so without Walmart.

Nieman Marcus on the other hand uses a customer intimate value proposition, and their execution reflects that.  Value pricing is traded off for a high level of customization and personal service, where repeat business is very important since the number of customers this proposition attracts is smaller than the “mass” approach;  you have fewer, but each more valuable, customers.  In this model, mass media is not very effective because the audience is not mass; instead, you rely on the intimacy to Pull customers in, and much more of the Marketing budget is invested not in Advertising, but on in-store (employees, fixtures, locations) and individual communication. 

This relational or customer intimate model is the root of  ”social marketing” and why any attempt to turn online social activity into some kind of mass media advertising opportunity is a complete Paradox.  A step by step example of optimizing the relationship marketing / social model is here: Marketing Bands Series.  To optimize the social model, you divert Marketing budgets away from Mass Advertising and Push into Pull areas like Usability / Store / Interfaces / Packaging, Customer Service, and Customer Retention.

Given the above, would Nieman Marcus ever consider using Walmart’s customer retention rate as a benchmark?  I think not; this approach would make no sense at all.  The mass model can’t leverage customer retention because it’s not intimate; if you can’t act on the metric, why measure it?  This is not to say Walmart “doesn’t care” about repeat business, of course they do.  But they can’t really lever it because it’s more operationally efficient for them to use the mass approach.

That’s a very long explanation for why I dislike using industry benchmarks but many, many people don’t realize how important this idea is; it’s why on a core business model basis some companies will not be able to realize significant benefits from “going “social”.  So on the whole, I would much rather use internal benchmarks that I can improve on that are aligned with the business drivers and are controllable through my own execution.

From looking at your web site, I’d judge you a Nieman as opposed to a Walmart, so customer retention can be a powerful tool for you.  So let’s talk about measuring retention.

“Retention” is a very time-specific concept – over the course of 3 months?  A year?  Five years?  A 20% retention rate over a 5 year period and a 60% retention rate over a 3 month period might both be stunning achievements, if you know what I mean.

So, if you are able to do the analysis, I would pick some marks – 3 month, 6 month, 1 year, etc. – and see what you get for repeat buyer or retention rates.  The slope of that curve will determine where any danger points are that you might take action on.  

For example, if retention falls dramatically from 3 to 6 months, then you know that you should be watching for people who have not transacted in over 3 months, and for  those people you should craft mail / e-mail promotions designed to bring them back.

As often happens with restaurants, there’s probably a good chance that if the person is still living in the area (more on this below), the reason they are not coming back is probably  controllable – they had a bad experience.  A promotion like “We’ve missed you” or “Give us another chance” that is tightly targeted to known defectors will usually pay back quite handsomely in both the short and long term. Defected customers not only visit once on the promo but also (hopefully) have a better experience and re-engage as a repeat visitor.  If your value prop is customer intimate / social, you absolutely must invest in superior customer experience so repeat experiences are rewarding.

If you see some success with this approach, you could then fine tune the analysis to find out if the dropout has a peak in month 3, 4, or 5.  This fine tunes timing of your drop; the closer you can get to the behavior with the message the more effective the campaign will  be.  There is a “peak profitability” timing in one of these months.  

Then the program can be automated, for example: if we don’t see a transaction from this person for 120 days, drop the message.  This way, you end up mailing every month but the audience is completely different and very highly targeted each and every time.  You will find this “right message, to the right person, at the right time” approach is much more profitable than mailing all customers because it directly leverages the customer intimate value prop.

Speaking of mailing all customers, the people who are still active within this 4 month time frame are probably still loyal and you can improve overall margin by not sending these special promotions to those people until they “slip” out of the 4 month window.  There’s no reason to discount to people who are highly likely to purchase anyway.  This is the Pull part of a relationship or social  execution.  What you should be really concerned about are the people who are dis-engaging, where there has been product or service failure.

In fact, in a relational marketing scenario, there is no real need to market to these people at all, you’re basically “preaching to the choir” (example) and doing so is a waste of resources (and often margin).  You will be far better off taking the money you used to spend marketing to the choir and allocating it to in-store, core value proposition ideas.

Many marketing people (especially of the Push variety) find this difficult to understand, but there no more powerful Marketing tool than your value proposition when communicating to the active customer base.  It’s why they are coming back, your Pull is already strong with them.  Why beat them over the head with messages when they are telling you by continued transacting that they like what you are doing?  Wasteful.  (more detailed example)

Finally, in a location-based scenario such as restaurants (and since you are the CEO and not running a single store), you might consider factoring in local uncontrollable churn into any metrics you create as internal benchmarks.  

Households in different areas have different natural churn (move) rates.  Since you have stores in different states, for example, one would expect a lower retention rate from stores that have a higher natural household churn rate.  These stores might be doing very well with controllable churn (product, service) but without the household churn adjustment, they could be unfairly benchmarked “bad”.  HH churn numbers are generally available free from city / state government or the Census.

Hope that helps!

Jim

Note to blog readers: Do you see the parallels above to a lot of what is going on in online publishing / advertising / marketing?  If not, see Jonathan Mendez’s Reaping the Ads You Sow for a more direct analysis of the same concept online.  The strength of the web is in Pull, in converting demand, not Push or creating it.  Use offline for Push; that’s what it’s good at, and synch the two to optimize the entire Marketing ecosystem.

Have a question on Customer Valuation, Retention, Loyalty, or Defection? Go ahead and send it to me here. If on the topic above, you can leave a comment on the post:

Relational vs. Transactional

Want to see the answers to previous questions?  Here’s the blog archive; the pre-blog newsletter archives are here.

Q:  First of all, thank you for the excellent book!  I’m really excited about digging into our own customer data to see what we’ll learn.

A:  Thank you for the kind words!

Q:  However, when you’re creating the RF Scores, what is the standard timeframe you should use?  I have access to about 5 years worth of purchase data – should I create RF scores based on the last 5 years, 3 years, 2 years, 6 months?

Our sales are quite cyclical, so I think the baseline should probably be at least a year, and I’m considering doing two years.  It seems as though if I get too much larger than that, my results will be too watered down. 

I’m also planning on generating “historical” RF scores by filtering my data to reflect the purchases only up to a certain point.  So, to generate a Q1-09 score, I’d create it from sales data of Q1-07 through Q1-09.  The Q2-09 score would be from Q2-07 through Q2-09, etc.  Does this make sense?  It will allow us to see the changes that have been happening in our company even though we’re only just now looking at the data.  It will give me a picture of what it would have looked like, had I looked at it back then.

A:  I think you have accurately understood the situation and have the right approach!  This type of analysis is very sensitive to time frame.

There are really 2 broad types of customer analysis.  There is analysis for action in the present, a Tactical approach driving towards a “we should do this now” result, and the more Strategic analysis, which is informational and says “this is what we should have done then” and / or “this is why we should make these business changes”.  The shorter time frame is Tactical, the longer timeframe Strategic.

So, for example, a 2 year timeframe could give you the answer to this question: which of our best customers are becoming unlikely to buy from us again?  This leads to immediate activation of some kind of marketing outreach or discount / incentive program to get another purchase from this group.

Add a timeframe that ends 4 years ago, then one ending 3 years ago, then one ending 2 years ago could highlight changes in the business over time, for example, best customers with high intent to purchase 3 years ago clustered in certain segments or SIC codes; now customers with this same definition are clustering in different segments or SIC codes. You will see migration of segment focus, if any.

Another way to think about this is time frame for the RF analysis determines sensitivity to new customers.  Long time frames tend to rank customers who have been with you a long time higher than new customers; this is just a function of how the ranking methodology works – these long-term customers have had more time to increase the Frequency or Monetary component.  This can mask important rankings in Frequency with newer customers, what you might call “future best customers” or “up-and-comers” who are accelerating their purchase behavior.  These folks are ideal targets for soft recognition-style rewards (not discounts) – VIP treatment, bonuses, etc.

You could even use this kind of analysis to prove the strengths (or weaknesses) of the RFM methodology for your business: given an RFM score of XXX 3 years ago, what behavior did the customer engage in during the following years?  Does the score in one year predict behavior the next year?

Or, perhaps rather than a ranking approach, the fixed activity threshold approach (like LifeCycle Grids) is more appropriate to our business.  LifeCycle Grids are basically the same idea as RFM, only sometimes more accurate for businesses with known cyclicality; it’s easier to build that cyclicality into the model if you abandon “ranking” and use thresholds.

In fact, this idea was born from an exercise like the one you propose: let’s re-score and re-rank customers each quarter, and track the RFM score over time.  Nothing wrong with this really, except there is the fundamental problem of scores changing due to outside influences, for example, a large new customer campaign.

When such a campaign is executed and then the database is re-scored, the RFM scores of customers can change even if their behavior has not because you are re-ranking a customer file that has changed in composition. Due to the new customer campaign, it is now “heavier” with Recency = 5 customers, which can push down the other customer scores even though behavior has not changed.

This is the primary reason I invented the LifeCycle Grid idea.  If you use thresholds or Hurdles for behavioral segments rather than ranking, the “score” of someone does not change when the database composition changes.  Someone deemed “best” and likely to buy if R = 30 days and F >= 25 purchases is still “best”, no matter how many records you add to the database.  These thresholds define the customer status by putting them in a fixed position box on the Grid, not a ranking.

And that is why RFM tends to be used as the Tactical, “we are doing a campaign right now” valuation method, and LifeCycle Grids tend to be used for the more Strategic analytical exercises.  However, the Grids can also be used for Tactical execution.

For example, any customer with F >= 25 over past 2 year period, who drops in R past 90 days, automatically should receive a call from their salesperson.  These reports could get run on a weekly basis, and of course can be segmented many different ways depending on the population you run through the Grid.  Because you’re using thresholds rather than “ranking”, a customer will appear in the Grid at the same location no matter what the size or segment of the input population.

So for example, you can run only customers  who responded to a campaign and see where they end up in terms of Recency and Frequency over time.  With a series of such runs, say monthly, you can create a “movie” that shows the evolution of the customers over a time frame and begin to judge the long-term effects of certain campaigns.  An  example of this approach is here.

Overall, I like the Grid approach much better.  Not only do you avoid the “population problem” of ranking when using RFM, but you can use the same approach over and over (good for management understanding) for many different kind of analysis, both Strategic and Tactical depending on needs.  You can use all kinds of visual aids such as color in the grid to represent different segments or campaigns, making presentations much easier for management to understand.  Decision making with execs can be much more of a challenge when all you have is RFM scores.

All that said, RFM is still probably the easiest  approach for specific, usually campaign-related tasks such as predicting campaign response or profitability.  Same data but a different, more short-term oriented way to look at the world probably best kept out of the boardroom but still has a place in the analyst’s toolbox.

Hope that helps!

Jim

Questions?  Does anyone think there is value in predicting which customers will become best customers and which customers are defecting, by campaign source or product line?  If you knew this information, could you act on it?  Would management care about this?

Have a question on Customer Valuation, Retention, Loyalty, or Defection? Go ahead and send it to me here. If on the topic above, you can leave a comment on the post:

RFM versus LifeCycle Grids

Want to see the answers to previous questions?  Here’s the blog archive; the pre-blog newsletter archives are here.

Q:  I’m involved in a loyalty program analytics project.  This client is a local pharmacy.  All sales are done directly in store, the web site is just for communication purposes.  The general problem we are trying to solve is the manager doesn’t have any detailed ideas about shoppers behavior apart from human observation. 

The idea is to launch a card-based loyalty program which will track sales activity and give insight into customer behavior.  The program will be points-based calculated on amount spent.  Points can be redeemed as rebates, coupons, gift certificates, or use points to buy items in loyalty program catalog.

The task is to segment customers according to their recent purchase behavior and determine the customer lifecycle.  I’ve been able to do some basic analysis using the R package and MySQL database, but am unable to detect customer lifecycle. 

Can you please give me guidance on this?

A:  What is the Objective of detecting the LifeCycle, to create a more “active” customer retention program?  Loyalty programs can be quite “passive” and often benefit from a more active overlay.  But there can be many reasons to want to understand the LifeCycle…

Q:  My 2nd task is to use the behavioral data with demographics to  build a direct marketing strategy and provide management with insight into the customer base, for example: percent new customers, % of Gold customers who passed to Silver in last quarter.

A:  Again, it would be helpful to understand how management would take action on this data.  But I suppose you are in the common position of not knowing the tactical approach, and nobody will lay it out for you (a.k.a. they are clueless)…and you don’t know the right questions to ask or how to ask them.

Let’s infer that the Objective for the loyalty program is to increase customer retention, the Strategy is to customize communications based on behavior to increase relevance and response.  This is typical for a loyalty program.

So, the metrics required for the Objective would show which customers are active and which are in the process of defecting.  To execute the Strategy, you need segments such as New, Silver, and Gold customers.

I use LifeCycle Grids for this purpose.  See this post on Measuring Customer Engagement for examples of what this approach looks like.  You can use % instead of counts in the grid segments if you wish.

Customers tend to move from right to left in the grid over time if they are inactive and defecting.  If active, they stay over on the left. Generally, the customers of most concern are those in the orange upper left Quadrant (Q3), best customers who are becoming former best customers.  These are the folks who should be targeted with double-points offers and so forth to reactivate them.

The most profitable grid cells to target for  reactivation can only be found by testing, but I suspect for a pharmacy it’s generally in the middle of the Recency scale – 60 days no activity, 90 days no activity, 120 days no activity.  You can’t wait too long, but you don’t want to do expensive promotions to customers who would buy anyway.  Wait for them to show signs of lapsing, then promote.

As far as “how many Silver turn into Gold”, if you line up the Frequency parameters in the grid for that boundary, you can use the grid for that as well.  Or, create a custom grid to reflect just movements between levels by using a selective population input rather than “all customers” in the grid.

The great thing about using the LifeCycle Grid method is it’s the same analysis for any population or segment, so management gets used to it and this drives consistency in judging the value of Marketing programs.  So you could run the same grid for Men, or people 50+, or people within 5 miles of the store. 

Comparing what the grids look like for different populations produces insight you can use to further sharpen your targeting ability.

Q:  How do I determine the value of a point for the different segments, say Gold versus Silver?  How do I optimize the ROI in each different segment given differing point values, what is the formula I should use?

A:  Hmm, you’re really out in the cold on this, aren’t you?  Are there any Marketing people involved in this, or are you supposed to use the “magic of analytics” to tell the Marketers what they should be doing?  Don’t tell me, let me guess – this is a “MarCom” project and the person who normally buys online banners and sends out press releases will be in charge of loyalty communications…

Here’s the problem.  The question you are asking is a Marketing question, not an Analytical one; there is no “formula”.

First, a good loyalty program budgets 3 – 5% of sales to points / overhead, meaning a point could be worth in real money between 2 cents and 4 cents, or you are planning to award double / triple points in certain segments. 

By a “good program” I mean a proactive effort that changes behavior, as opposed to running as a faith-based initiative, the old “we have a loyalty program so people are more loyal” approach.

Companies try to do loyalty programs in the 1 – 2% range but it’s rare they actually work because there’s not enough leverage.  You can’t really provide decent rewards that truly motivate behavior in that range, unless you can provide a lot of intangible benefits though other arrangements outside the point formula.

For example, you have a movie star under contract and part of that contract is to attend company functions.  You then arrange for such a function and customers bid points for one of 10 available slots.  That’s an intangible benefit; there’s not a 1:1 financial correlation between the value to the customer and cost of points.  It’s “insider access”, recognition that is  emotionally rather than financially driven.  Inviting best customers to sporting events has long been used the same way.

If your budget is limited to a 1 cent a point, you really need to think through how you will create excitement for the program by adding these intangible benefits.  Not that you have access to movie stars, but how about unique discount programs?  Exclusive product Previews?  Paid services provided free?  And so forth.

Second, you really can’t change point values across program segments without getting into a lot of trouble on the finance side, it becomes too difficult to audit and you lock yourself into all kinds of execution problems.  For example, what happens when a Gold Member drops to Silver if the point values are different?  Further, what if an item is bought when the member is Gold and returned when the member is Silver?  How will the accounting work when the net differential in point value creates a value gap?

If you want to boost offer value for certain segments, you award more points - points are still worth the same amount, but you give double or triple points, not change the value of a point.  Points end up on the balance sheet as a liability so this is deadly serious stuff, you do not want the value of a point to fluctuate, trust me.

Ask any CFO.

What level is selected generally depends on the margins in the business, and how the program will be paid for.  Example: you decide to kill all promotional coupons and use the savings from those discounts to fund points, which would tend to improve your margins, because point discounts are targeted, coupons are not.

But there is no way to determine a point value that is optimum for ROI, it’s how the points are applied and used in the program, the Marketing Tactics – the Targeting, Messaging, and Timing driving behavioral change – that determine what the program ROI will be.

The best advice I can give you on this is to TEST before you decide on anything – point levels, budgets, segments, etc.  Use the LifeCycle Grid to find out what works in terms of offers and behaviors.  Can you induce loyalty and repeat purchase from lapsed buyers?  What level of discounting or special services is required to change behavior?  Then translate these results into budgets, point values, and programs. 

I realize you probably don’t control this decision, but I would raise the question if you can: what if we don’t really need a Loyalty Program?  What if a less complex and less expensive Retention Program (based on testing results with the LifeCycle Grids) would be more appropriate for our business? 

For example, what if we could get best customers to raise annual spend 10%, and at the same time induce just one more shopping trip before a customer defects?  How much would that be worth on annual basis?  How much can we afford to pay for these revenue increases?

The LifeCycle Grid gives you a “roadmap” for testing your Targeting, Messaging, and Timing  consistently across different segments as your customer population churns over time.  You can use the results to develop either a simple Retention Program or a more complex Loyalty Program, whichever the test results seem to indicate is the best direction to go.

Loyalty programs do not create “Loyalty” – your products and service create Loyalty.  Loyalty programs are an incentive to change behavior, and if analyzed and executed that way, can be very profitable, as can simple ongoing Customer Retention campaigns.

Jim

Readers: Additional questions or thoughts?

Have a question on Customer Valuation, Retention, Loyalty, or Defection? Go ahead and send it to me here. If on the topic above, you can leave a comment on the post:

Loyalty Program Structure & Tracking

Got a question about Customer Measurement, Management, Valuation, Retention, Loyalty, Defection?  Just ask your question.  Also, Feel free to leave a comment.  Want to see the answers to previous questions?  Here’s the blog archive; the pre-blog newsletter archives are here.

Q: I received this article (Norms of Reciprocity, measuring value of Social Marketing) via a friend’s Twitter account.  Very interesting.

A:  Glad you enjoyed it!

Q:  It has made open up my ACT! database, and my Outlook databases and add the metric of Growing / Strong / Weakening / Failed to my normal Sales and Business progress metrics.  If I group those categories and correlate to traditional metrics, it’s impressive how they reflect each other.

A:  Yes, most people are surprised.  It’s a very, very simple idea that seems to work across just about any human activity including crime, attendance, and so forth.  

The more Recently someone has done something, the more likely they are to do it again.  Conversely, the longer since an activity last took place, the less likely the person will do it again.  Often called Recency in Psychology and studied quite a bit.

Q:  Now I have to think about how I really use and apply this. : )

A:  Well, if I can guess you are in Sales from your title, typically one of the best applications is in what Strategic Marketing folks might call “allocation of resources”, which probably translates into “lead nurturing” for you.

Most experienced people in Sales have a sort of “sixth sense” when it comes to thinking about the likelihood of a close happening.  They worry about certain prospects more than others, and a sort of “ranking” or “scoring” happens in their mind.  One of the triggers that frequently comes up in this is “how long” it has been since there was any contact activity with the prospect, and the feeling the longer it has been without sales activity, the less likely the sale is to close.  Sales Managers will often allocate resources based on these kinds of “feelings” they or salespeople have.

The problem with all this “gut feel” is, newer sales people don’t have it, and so probably are not as productive as they could be.  The other is since a lot of this is not tracked in any way, there aren’t any firm “guideposts” and it may be that sales are lost that otherwise could have been made due to a lack of urgency or misdirection.

So, given limited resources, a sales force would generally like to focus on the leads most likely to close, and not work on the less likely leads until the most likely leads have been addressed.  This is the idea of scoring, let’s rank all of our prospects by likelihood to close.

Now, as far as what you might do in ACT! or similar (and knowing nothing about your business), here is what I would do.  Just start informally comparing prospects that close and those that don’t close in terms of these timing issues, “how long since contact” or “how long between contacts” for each case.

Typically you will start to see patterns of some kind, for example:

1. “Prospects who have not made it to 2nd sales appointment within 30 days of 1st contact are less likely to close”

2. “Prospects who take longer than 25 days to respond to proposal are less likely to close; prospects who take less than 10 days to respond to proposal are very likely to close”

and so forth.  Look at important events in the sales process and note the “time since” or “time between” and look for such patterns.

Now, as I said, many salespeople, especially experienced ones, have some sense of these ideas, but they have never been quantified. The advantage to quantifying them like this is you can move to a “triggered contact system” based on them, which I think you can do in ACT! if you have the data.  This conserves salesperson resources and helps them always be focused on where they are most likely to close the business.

So, for example, salespeople (sales managers, if more appropriate) receive a communication each day about any prospects who are coming close to any of these triggers above.

In scenario 1 above, a counter starts on 1st contact and if another sales call has not been scheduled within 20 days of 1st sales call, a reminder goes out saying “you have 10 days to get a 2nd appointment or you may lose this sale”.  In scenario 2 above, sending the proposal triggers the counter, and a sales contact is suggested at 7 days later and 15 days after that.

The optimal timing of these contacts is something discovered over time, and of course depends on the business. But having these triggered messages available to guide salespeople towards which contacts they should be most focused on that day or week is a lot better than nothing.

So instead of a salesperson thinking this:

“Gee, it’s ‘been awhile’ since I talked to prospect George. Maybe I should call him”.

you get this thought:

“I sent the proposal to prospect George 7 days ago, and I need to close him in 3 days, or he becomes less likely to close at all.”

The difference in those two thoughts and the action taken can be a lot of sales – especially with newer sales people, who don’t have enough experience to understand the “rhythm of the sale” in this specific business yet.  If you’d like a more detailed example, there’s one here: B2B Software – Latency Tripwire.

Spreadsheets are usually a great tool for this kind of discovery work.

Hope that helps!

Jim

——————————-
If you are a consultant, agency, or software developer with clients needing action-oriented customer intelligence or High ROI Customer Marketing program designs, click here
——————————-

Have a question on Customer Valuation, Retention, Loyalty, or Defection? Go ahead and send it to me here. If on the topic above, you can leave a comment on the post:

Lead Scoring and Nurturing