# How Long is a Customer LifeTime?

Jim answers questions from fellow Drillers
(More questions with answers here, Work Overview here, Index of concepts here)

Topic Overview

Hi again folks, Jim Novo here.

There’s always two questions about the topic of Lifetime Value – how do you quanitify value, and how long is / how do you measure / decide what a Lifetime is? For now we’ll leave the value question unanswered, because a lot of that depends on company culture and what question you are trying to answer. Plus, it depends on how you measuree a Lifetime.

So let’s do the Lifetime thing first, shall we? To the Drillin’…

Q:  First of all thanks for an excellent web site – I often visit it to learn and / or get inspiration in my work.

A:  Thanks for the kind words!

Q:  Anyway, I work in a telco retention department and I’m trying to calculate a true and fair value for customer life time answering the question : “How long do we on average have a customer?”.

A:  A both noble and useful pursuit!

Q:  I have data on when customers signed up and when they left (or of course whether they are still here). My first problem is whether to include both lost and existing customers in the calculation.  If you only include the customers you lost you are only able to answer the question for those.  If you include existing customers you don’t know what life time to use for them.

A:  Well, yes, that’s correct.  But you’re really trying to accomplish several things at the same time, so you can break the analysis into different parts and then apply some business logic to get your answers.

Let’s assume that for whatever reason, your company attracts two kinds of customers in terms of their profitability – really profitable ones and really unprofitable ones.  Whether the customer is profitable or not (as is often the case) is a direct function of how long they remain a customer.  So all your disconnected customers are “bad” and all your current customers are “good”. If you analyze the average life of disconnected customers, you will come up with the average life of a “bad” customer.  If you analyze the average life of all current customers, you will find out the average life of a “good” customer.

These lives could, and perhaps should, be different.  Or should they?

What you may find is the average life of a bad customers is actually longer than the average life of a good customer.  That’s an interesting idea, isn’t it?  What it means is the quality of your customers has been falling over time; that for some reason you are attracting more “short cycle” customers than you used to.

Conversely, if you find that the average life of current customers is longer than that of disconnected customers, the quality of customers has been rising over time.

You should also look into the active customer base and do an analysis of customers by life, for example, % 1 year, % 2 years, % 3 years, etc.  Here is what this typically looks like:

1 year 20%
2 years 40%
3 years 20 %
4 years 15%
5 + years 5%

This is a “snapshot” of the customer base, what it looks like today.

What you will probably find (depending, of course, on how long you have been in business) is that, for example, only 5% of active customers have been customers for 5 years or more. Now, just because there are customers who last for over 5 years, that doesn’t mean the “average” life of active customers is 5 years, does it?

Of course not.  Somewhere in your analysis of active customers you will find a “bulge” (in this case above, at 2 years) where the highest % of active customers are, and this is probably a good number to use for the average life.  It may also surprise you that this active customer life is remarkably similar to the average life of a disconnected customer.  If that happens, I think you have the answer to your question, right?  Despite the fact you may have active customers with longer lives, the disconnected customers provide a good view into the “average life”.

To drive further down this path and provide more information on “expected” life with active customers, you can do a “Longitudinal” study. Instead of a “snapshot”, you get something more like a “movie“, which will give you deeper insight into the LifeCycle and thus LTV.

Start with customers who all became customers in the same quarter or year (say 5 years ago), and look at the % of these who are still  active over the years, for example:

Active at 1 year – 100%
Active at 2 years – 60%
Active at 3 years – 30 %
Active at 4 years – 20%
Active at 5 years – 5%

There’s a serious drop off between years 2 and 3.  So if you’re talking “average life”, it’s probably somewhere in there.

Now, the average life is not a particularly useful number, by itself.  Telco customers have different lives depending on how they were acquired (the offer), what services they take, and how old their hardware is.  If you have collected “source” information on customers, for example, which campaigns they responded to when they became customers, then you can look at the average life in a more actionable way. I n other words, the question is not just the life, but the life in relation to how the customer was acquired.

Here is the same Longitudinal study from above with Source information:

1 year 100% Source: 60% discount mailer
2 years 60% Source: 65% TV / Radio
3 years 30 % Source: 80% Newspaper
4 years 20% Source: 75% Internet
5 + years 5% Source: 70% “tell-a-friend”

So over the years, as customers cycle out, we find that the ones with the longest life came from the campaign “tell-a-friend”.  In other words, the “average life” of a customer differs depending on source, and the average life generated by the campaign “tell-a-friend” is longer than average, and much longer than the average life of a customer generated by campaign “discount mailer”.

Frankly, I’m not sure where people got the idea that LifeTime Value is a static number.  If that were true, why bother with retention programs?  If your retention program is a success, LTV should increase.  If you do something that causes decreased satisfaction, LTV will decrease.  Expected LTV is dynamic based on what your company is doing at any one time.  The only time you get “final LTV” is when the customer actually defects.  When you go through a cycle where you do new kinds of advertising / change product offerings the “expected LTV” will always change.

Managing customer value is not really about the absolute LifeTime Value of the customer, but the expected LifeTime Value based on what kinds of changes you are making in marketing, service, technology, etc.  You can track expected LifeTime Value and the relative LifeTime Value between customer segments in the present, so what difference does it make if you don’t really know the “end game” on value until the customer terminates?

Tracking changes in relative and expected LTV LTV when comparing customer segments is the most powerful tool a marketer can possess.  If you can predict changes in customer value before they happen, you can take action in a timely way if you need to.

Q:  Second, I think it is relevant to limit myself to look back a specified amount of time, say 2 years worth of data.

A:  I’m not sure why you would do this or what is driving this thought…is it because your contracts are typically 2 years long?  Then you’re looking at more like a “churn” type of thing, what % renew the contract…

Q:  As an example, consider a company that has been in business for 10 years and only ever had 3 customers, one leaving after 1 day (ie life time 0), the second left after 5 years and the third is still a customer.  What is the customer life time in this case ?  It is 2.5 years for the lost customers but that does not say much about the actual customer base today.  You could say it is 5 years as the average between the three.  Or you could say that it is 10 years (and counting) based on your existing customer base.

A:  Well, that’s not a very realistic example, but I think I get your point.  If you have been in business 25 years, the average life of a customer “today” in relative terms is of course more useful, since so much has changed in the telco business over the years.

If you want to try to find a good cut-off, do the same kind of analysis referred to above – analysis of customers by life, for example, % 1 year, % 2 years, % 3 years, etc. and just keep going.  You will see the percentage of customers in a given year approach zero.  Where you cut off (1%, 5%, 10%?) would be arbitrary, but remember, LTV analysis has to be actionable to be useful.  If you can’t see the company ever designing a campaign that would reach under 5% of customers, cut it off there.  I think 2 years is a very short window for an LTV analysis, though.

Q:  I hope you can give me a little guidance.

A:  Should be enough here to get you started!

Jim

# Using Multiple, Related Customer Models Across the LifeCycle

Jim answers questions from fellow Drillers
(More questions with answers here, Work Overview here, Index of concepts here)

Topic Overview

Hi again folks, Jim Novo here.

So you have all these simple but powerful customer models – Recency alone, Latency,  RFM, or LifeCycle Grids – how do you know which one (or ones) are best to use for your business? Guess what – it depends on the specific features of your business and also how you run the business. Now, while that might sound a bit scary, it’s really not that big a deal and in the end, the great news is you’ll end up with an approach customized to your business. So how do you accomplish this? Just segment and analyze your customers; they will tell you, my fellow Driller, which direction is the best to follow. You dig? Let’s go ahead and see what that looks like …

Q:  We recently purchased your book

A:  Thanks for that!

Q: and we are ready to start building some RFM analysis.  We are a search marketing business – we have a large customer /prospect base.  We have limited knowledge about them and we are keen to start on the journey.

A:  OK…let’s see what you’ve got.

Q:  We are hoping to extract database (approximately 25k names) of the last 6 months records and do some RFM analysis on key customer groups.   Specifically:

TEST GROUP A – people who initially purchased one of our trial products – we want to know what is their RFM score.

TEST GROUP B – subscribers to our “tool kit” product at \$50 / month – we want to know what is their RFM score.

Q:  What kind of data are we talking about?  Is it web site visits, clicks on emails, transactional / subscription data, all of the above?

A:  Before setting up the model we have a couple of questions we hope you can shed some light on:

1.  How do we treat subscription – our business has a mix of one-off and subscription business – if someone “buys” every month with a subscription, is that included in Recency & Frequency?  Any insight you can provide us would be great – we found some info on this in the book but unsure given ours is a mix of subscription and one off.

A:  Well, let me start off by saying that your primary goal is to understand the customer LifeCycle so you can use it to increase your profits.  If you can understand the LifeCycle, you can discover the most profitable ways to acquire customers and stimulate demand from them.  Which behavioral model (RFMRecency alone, Latency, or LifeCycle Grids) is best suited for this task is something you will have to discover.  You’re probably jumping the gun a bit (I’m guessing) to hop right on to using the RFM model.  Before thinking about the models, you should probably do some thinking about segmentation – what logical groups of customers do you have, and what is likely the best model to score each segment with?

I’ll walk you through some of the thinking…

First, it is perfectly acceptable to use different models on different segments, in fact, often more desirable.  The one-off business – if you mean the ongoing purchase of products sequentially over time – is suitable for RFM analysis, it’s a “retail” business segment.  The most Recent and Frequent buyers of these one-off products are the most likely to purchase another one.

As you map promotional response to RFM score, you will see that there is a score that represents customer defection – response falls off rapidly below this score.  This score is your “trip wire”, it basically tells you that unless you get people with this score to purchase again, you have likely lost them.

Further, if you include the original source of the customer in the scoring process, you will find certain campaign sources tend to create customers with consistently high RFM scores and others tend to create customers with consistently low RFM scores.  This fact should allow you to better balance marketing spend, shifting spend from low score to high score customer sources.

The subscription business is something different, and probably requires a different approach.  For example, you might look at “average subscription length” to determine a “trip wire” for customer retention efforts.  If the average subscription length is 8 months, you know that in the eighth month of any sub’s tenure, you should be proactive in retaining them.  So for example, subscribers in their 8th month should be offered a renewal “bundle”:  “Renew for another 3 months and get 20% off”.  This will in effect drag a bunch of potential 8 month defectors into 11 months, where they have a better chance of “sticking” with subscription.

Two models for two distinct segments.

But, if I understand your business model, you have a fairly typical direct marketing of services approach – you acquire customers through free or low priced offers (these are also the one-off products?) and once customers gain confidence, you attempt to upgrade them (these are the subscription products?).  So what you really should be paying attention to is the “flow” of the customer through this expected one-off to subscription LifeCycle, then use the models to alert you to opportunities to increase profits along the way.

One way to do this is to deconstruct the LifeCycle of best customers.  Extract your best customers and look at the event sequences.  What series of events occurred, and what was the timing of these events, that lead to the creation of these best customers?

Then, use this real world LifeCycle pattern to improve the profitability of your marketing efforts.  This approach is the basis of my LifeCycle Grid model, and it’s extremely powerful because you are using the two rules of High ROI Customer Marketing to your highest and fullest advantage:

1. Don’t spend until you have to

2. When you do spend, spend at the point of maximum impact

So, to follow through with the example above:  Let’s say you are running your two models on the two segments – the RFM model / one-off segment and the Average Subscription Length / subscription segment.  If the LifeCycle is as described above, best customers probably come in through the one-offs, migrate to being high RFM score customers in the one-off segment, then convert to a subscription and end up in the subscription segment. This conversion to subscription is clearly a “tipping point” of some kind – the point of maximum impact.  Well then, when does this occur?

Further analysis of best customers shows that the conversion to subscription typically occurs when a one-off customer achieves a RFM score of 444 or higher, and the conversion to subscription takes place on average within 30 days of a customer achieving this 444 RFM score.  Knowing this, what can you do?

First, when a customer achieves a 444 RFM score, you know this customer has high potential to become a best / subscription customer.  So it would be worth it to you to focus on these customers, do something special for them, stroke them in a special way beyond the regular e-mail newsletter.  How about calling them and thanking them for their business?  Sending them a special e-book?  Adding a new software service free?  Divert resources from low ROI marketing activities to pay for this High ROI marketing activity.  Same budget, higher profits.

Second, you know there is a clock ticking.  You have 30 days to convert this customer to a subscription customer.  After 30 days, they will start becoming less and less likely to become a subscription customer.  It’s now or never.

You now have all you need to fulfill the two rules of High ROI Customer Marketing:

1. Don’t spend until you have to

2. When you do spend, spend at the point of maximum impact

You have just created a dynamic, action-oriented, custom-built, LifeCycle-based behavioral model for a particular segment.  A model customized to your business and one that self-adjusts to customer behavior dynamically.  What does that all mean for you?

It means that if it takes somebody a year to hit 444, that’s fine, you don’t waste a lot of effort marketing subscriptions to them – they are not ready anyway.  If someone hits 444 in 2 months, you are tipped off to act on subscription marketing in a big way.  Instead of treating every customer the same, regardless of their behavior and potential value to the company, you are allocating resources to customers based on a model that guarantees to increase your profitability.

And because it is all driven by simple numbers and time stamps, you can completely automate not only the scoring and “trip wires” but the marketing as well.  Same marketing resources, deployed in a smarter, more profitable way.

Then you take the next segment, then the next, then the next.  Let’s say you find that if a one-off buyer has not bought a subscription within 1 year of first one-off purchase, they are highly likely not to *ever* buy a subscription.  Well, stop marketing subscriptions to them at the 1 year trip wire.  Market one-offs, or create a new product that would appeal to this group.  Develop a model for each segment, and automate.

Q:  2. Do you capture RFM score monthly and track – if so does that mean you need to apply a date stamp for e.g. to the score?

A:  You can look at RFM scores on a sequential basis if you have a plan of action to take advantage of that knowledge.  Rising scores indicate rising potential value to the company – these are best buyers in the making.  Falling scores indicate the beginnings of customer defection.  This is the customer LifeCycle playing out in front of your eyes.

You could use this data to manage marketing activity at the individual level, though it’s pretty granular and would take a lot of resources; pretty advanced stuff and in the beginning, better to stick with segments and learn.

However, in the aggregate, you could create reports that look at the “delta” from month to month, basically predicting the future value of the business – is it rising or falling?  With this, you could answer this question: Is our marketing creating customers with high future value?  The approach is similar to the “delta grids” in the LifeCycle Grid part of the book, it’s a “migration” report showing you how customers are moving though the LifeCycle.

For example, you could create a report that answered this question:

Of all customers with a score of over 55X (rocket fuel customers) 6 months ago, where are they now?

35% 55X
25% 4XX
15% 3XX
15% 2XX
10% 1XX

This result isn’t particularly significant **by itself**.  However, when tracked on a monthly basis over time, if you saw increasing defection in the high end and growth in the low end, trending towards something like this (compare with above):

15% 55X  (down 20 % points)
15% 4XX  (down 10 % points)
20% 3XX  (up 5 % points)
25% 2XX  (up 10 % points)
25% 1XX  (up 15 % points)

that would literally mean the future sales (potential value) of your customers is falling; something you are doing (probably in marketing or service) is working against you.  Put another way, sales in the future will be lower on a per customer basis than they are today.  However, if you saw retention in the high end and defection in the low end, trending towards something like this (compare with first chart above):

40% 55X  (up 5 % points)
30% 4XX  (up 5 % points)
10% 3XX  (up 5 % points)
10% 2XX  (up 5 % points)
10% 1XX  (up 0 % points)

that would literally mean the potential value of your customer base is growing – you are acquiring higher value customers and  / or retaining a higher percentage of high value customers.  Sales in the future will be higher on a per customer basis than they are today.

That kind of crystal ball on future sales and profit can be very valuable indeed!

Jim

# Segment to Best Determine LifeTime Value (LTV)

Jim answers questions from fellow Drillers
(More questions with answers here, Work Overview here, Index of concepts here)

Topic Overview

Hi again folks, Jim Novo here.

LTV has to be actionable.  If  you can’t take action on the information, it’s not relevant anyway.

There you go, the most universally true rule when attempting calculation of LTV.

And the best / easiest way to accomplish this is to identify similar customer behaviors and segment the customers by these behaviors – THEN figure out LTV by segment.

If you can’t actually take action on the information, then why spend countless \$\$ and hours fussing over all the reasons the number you come up with might be wrong and trying to solve unsolveable data or corporate issues? The best idea to implement when developing / using LTV is consistency – let’s get the team to agree on what LTV is and how to measure it, stick with those ideas for at least several years, test and take action on the results to uncover value, THEN (perhaps) discuss improvements!

Q:  I have just been reading your series on Comparing the Potential Value of Customer Groups. I am having trouble calculating the lifetime value of our customers.

A:  Yes, well, everybody does for some reason!  Often the problem is too much
focus on trying to look at the “average customer” as opposed to segmenting
customers.  By segmenting first, it’s both easier to get to LTV *and* more useful since it’s easier to take action on  a segment than the “average customer”.

Q:  Our company provide accounting software solutions to small to medium sized owner operated  businesses.  Because of what we sell and who we sell to, a lot of our customers are most likely to just buy one or two of our software products and unless they sign up for support (only around 15% do), we may never here from them again.  It is therefore very difficult to determine an average / standard lifetime that customers use our product.

A:  Sure.  First, the 15% segment that does sign up for support sound like good customers to me.  So that’s one segment.  How long do they typically stay signed up?  That’s the average life for this segment.

Then there are probably people who upgrade over time, right?  I can’t imagine an accounting product that people would not upgrade – perhaps not every cycle, but every 2nd or 3rd cycle.  That’s another segment.  Then there are probably some who both follow the upgrade cycle and pay for support.  These are probably the “best customers” and they are a unique segment as well.

And finally, you have the buyer who makes one purchase and you never see again.  These people are also a segment.

Q:  What should I base it on, how long our customers use our products (which would be almost impossible to determine), or how long they spend money with us?  So I measure on average the time between the first and last transaction of customers who have the highest Recency???