Or Behavioral Messaging, as we used to call it.
Much has been written about Measuring Engagement, but once you measure it, then what do you do with this information? Most folks know the idea driving the Engagement Movement is to make your messaging more Relevant, but how do you implement? Perhaps you can find the triggers with a behavioral measurement, but then what do you say?
This is the part Marketing folks typically get wrong on the execution side. They might have a nice behavioral segmentation, but then crush the value of that hard analytical work by sending a demographically-oriented message, often because that is really all they know how to do. So as an analyst, how to you raise this issue or effect change?
Marketing messaging can be a complex topic, but there are some baseline ideas you can use. Start here, then do what you do best – analyze the results, test, repeat.
You want to think of customers as being in different “states” or “stages” along an engagement continuum. For example:
- Engaged – highly positive on company, very willing to interact – Highest Potential Value
- Apathetic – don’t really care one way or the other, will interact when prompted – Medium Potential Value
- Detached – not really interested, don’t think they need product or service anymore – Lowest Potential Value
Please note that none of these states have anything to do with demographics – they are about emotions. The messaging should relate to visitor / customer experience as expressed through behavior, not age and income.
These states are in flux and you can affect state by using the appropriate message based on the behavioral analysis. Customers generally all start out being Engaged (which is why a New Customer Kit works so well), then drop down through the stages. The rate of this drop generally depends on the product / service experience – the Customer LifeCycle.
Generically, this approach sets up what is known as “right message, to the right person, at the right time” or trigger-based messaging. Just think about your own experience interacting with different companies; for each company, you could probably select the state you are in right now!
OK, so for each state there is an appropriate message approach:
Engaged – Kiss Messaging: We think you are the best. Really. We’d like to do something special for you – give you higher levels of service, create a special club for you, thank you profusely with free gifts. Marketing Note: be creative, and avoid discounting to this group. Save the discounts for the next two stages.
Apathetic – Date Messaging: We’re not real clear where we stand with you, so we’re going to be exploratory, test different ideas and see where the relationship stands. Perhaps we can get you to be Engaged again? In terms of ROI, this group has the highest incremental potential. Example: this is where loyalty programs derive the most payback.
Detached – Bribe Messaging: You’re not really into this relationship, and we know that. So we are simply going to make very strong offers to you and try to get you to respond. A few of you might even become Engaged again.
Can you see how sending a generic message to all of these groups is sub-optimal? Can you see how sending an Engaged message to the Detached group would probably generate a belly laugh as opposed to a response? You’ve received this mis-messaged before stuff, right? You basically hate the company for screwing you and then they send you a lovey-dovey Kiss message. Makes you want to scream, you think, “Man, they are clueless!” and now you dislike the company even more.
Combine this messaging approach with a classic behavioral analysis, and you now have a strategy and tactic map. For example, you know the longer it has been since someone purchased, clicked, opened, visited etc, the less likely they are to engage in that activity again. Here’s the behavioral analysis with the messaging overlay:
Click image to enlarge…
Please note “Months Since Last Contact” means the customer taking action and contacting you in some way (purchase, click) not the fact that you have tried to contact them!
So does this make sense? Those most likely to respond are messaged as Engaged – as is proper in terms of the relationship (left side of chart). As they become less likely to respond, you should change the tone of your communication to fit the relationship up to a point, where quite frankly you should take a clue from the eMetrics Summit and not message them any more at all (right side of chart).
Example Campaign for the Engaged: At HSN, I came up with the idea of creating some kind of “Holiday Ornament” we could send to Engaged customers. If the idea worked (meaning it generated incremental profit), we could do it as an annual thing; we could put the year on the ornament and create a “collectible” feel, which is the right idea for this audience. No discount – just a “Thank You” message “for one of our best customers” and “Here’s a gift for you”.
These snowflake ornaments were about $1.20 in the mail (laser cut card stock) and generated about $5 in 90-day incremental profit per household with the Engaged, test versus control. Why? Good ‘ol Surprise and Delight, I would bet.
We had some test cells running to see how far we could take this, and as expected, the profitability dropped off dramatically based on how Engaged the customer was. If the customer was even minimally dis-engaged – no purchase for over 120 days – there was very little effect.
Interactivity cuts both ways; it’s great when customers are Engaged, but once the relationship starts to degrade, folks can move on very quickly emotionally. That’s why it is so important to track this stuff – so you can predict when your audience is dis-engaging and do something about it.
Ah! Now THIS is the sort of thing I wanted to know about when I put that request in on the WA Forum for Marketing Books suggestions way back when.
I’ve learnt a lot from what I have read to date – but not the above.
My Thanks Jim. A very informative posting!
One question tho: You mentioned response rates dropped off after 120 days. Is there scope to see if folk are buying based on the Xmas season vs rest of year type of thing?
My understanding being that HSN is very much a luxury market – so you may be conflicting with empty budgets from buying pressies???
Would it be wise to perhaps try again around March-May?
Cheers!
– Steve
Good question Steve, probably should have explained the idea of a “controlled test” in more detail. A random sample of the Engaged did not receive the ornament and performance was measured against this control group. The $5 in incremental profit per household was over and above the normal spend rise for the 90-day period as represented by the control group. Make sense?
So for example, let’s say the control group spends $100 during the 90-day period; at 40% margin that’s $40 variable profit. The test (received ornament) spends $115.50, at 40% margin is $46.20 variable profit. That’s an increase of $6.20 in variable profit, minus cost of $1.20 ornament mailer = $5 incremental profit due to promotion.
To answer the question behind the question, we tested the same idea in different forms at different times of the year and it always worked versus control, as long as the customer was highly Engaged. Generally, as days since last purchase increased, profitability decreased.
By the way, since it seems like you are trying to “study up” on Marketing, please be aware that for any “Holiday” promotion, you should be inclusive in terms of religious beliefs. The copy on this card was positioned around “Best Wishes for the Coming Year”, and even though you could punch out the laser-cut design and hang it on a tree, it worked as a free-standing “Best Wishes” card.
I’ve always thought that the reason marketers default to “demographic messages” isn’t simply that “that’s all they know”, but that in true behavioral messaging, some contacts aren’t designed to produce a purchase response. As a result, marketers are unwilling to “artificially depress” response rates.
Of course, as you well know, that logic is ridiculous, since “response rate” should be measured off a denominator of just the response (purchase)-seeking messages.
But try telling that to the CFO.
Ron, I’m starting to believe the biggest secret in web marketing – if not all marketing – is the control group. When I spoke of being able to measure the profitability of “non-response campaigns” at the eMetrics Summit, a lot of people were kind of shocked. At the DMA, the same topic was a bit better known (as might be expected) but still seemed to draw a lot of interest. Repeat after me:
YOU CAN MEASURE THE PROFITABILITY OF CAMPAIGNS WITHOUT A RESPONSE ELEMENT.
In fact (as you know) you should almost always use a control since you’re not measuring a lot of the true response by “measuring response” – if you know what I mean. There is plenty of action outside people using the response device that simply doesn’t get measured – especially online.
So the absolutely ironic lesson here is this: by using a response or tracking device instead of a control group, most online marketers are underestimating their true response rate.
Kevin provided more examples of control group thinking here.
Jim,
I agree it’s possible to measure the profitability of campaigns without a response element but don’t you think the measurement will be more accurate based on response. Of course, a lot of factors need to be considered from the campaign itself to the targeted market, but numbers – real numbers are the bottom line.
Thanks for the comment Mark. I’m not following you on the “real numbers” part – how is using a control group not real numbers?
To use the example from the post, the control group – which is a random sample of the targeted population – spends $100 per person. The test group – which is the remainder of the target population, which receives the campaign – spends $115.50 over the same time period. What about the $15.50 in incremental spend for test versus control is not a real number for you?
I’m truly interested in your perception of why this result is “not real”. What part of this approach makes you question the result? Or are you saying some version of what Ron said – that anything other than “response” is too difficult to explain (to the CFO)?
Thanks yet again Jim. Yes does make perfect sense.
Greatly indebted to you for your time!
“Study up”? I suppose so. :-)
I see it as more expanding my horizons. Value Add and the prior reasons I supplied back when.
Ha. Learning Marketing to better Market myself and the skills I can offer such that I can help better Market my clients services to their Marketable clients. Peter Piper Picked…. ;-)
Seriously, for a change, what I am personally finding interesting is the ability to use this gained knowledge in my highly technical core field. THAT was a very unexpected bonus. Little things, and not every 2nd minute, but often enough to be very valuable.
Cheers!
– Steve
Sir,
i am working with an indian telecon firm and like every one else have gone through kotler. Currently looking at refreshing my skills with some good book of your (Specially with live exampl understanding) need your advise in selecting the book on basin product management( product ,pricing, customer and channel engagement etc)
I assume by “gone through kotler” you mean his book Marketing Management or Principles of Marketing? If you’ve only read one of these read the other, or any of the other books Kotler has written. One that might be relevant in your industry is Kotler on Marketing : How to Create, Win, and Dominate Markets.
Thanks Jim, very interesting and insightful. Avinash had a great post in July called “I got no e-commerce, how do I measure success”. A version of this post targeted at the same audience (read me!), would be tremendously helpful.
Well, Avinsash talks about Recency as a key metric and that’s exactly what I’m doing here with a lot more tactical detail. The longer it has been since last action – whether it be visit, purchase, post, etc. – the less likely the action is to happen again, relative to other customers / visitors who have acted more Recently.
So it is important to:
1. Monitor the average Recency of segments to detect if they are weakening
2. Communicate to the segments differently, properly reflecting their level of engagement
So, for whatever action you deem important on your site, people highly engaged in it should get one message, people who are starting to drift away should get another message, and people who have detached should get a 3rd message, if you want to optimize the marketing effort.
Speaking to these 3 groups with the same message will diminish the overall response / profitability of the marketing effort, because these 3 groups have different levels of engagement with the site.
Thanks Jim for the update. Sorry if my question was not specific enough. In the end my focus is on the other axis of your graph. Without e-commerce, measuring response effectively becomes more challenging. When I say no e-commerce, I mean no online sales and no lead collection. That means we need to measure “engagement” using less straight forward metrics like PDF downloads, video play, PV / Visit etc (or an index of all of the above). The segmentation method stays intact as far as I can see. Does that make sense to you?
Sure, and I think I answered your question, or (quite possibly!) we have different definitions of engagement?
If by engagement you mean “interaction with site during a single visit” then what I’m saying here probably will not have much relevance for you. These behavioral metrics are meant to look at the relationship of a visitor to your site over time, and be predictive of the visitor continuing to be engaged with your site.
Said another way, what’s important at this more strategic level is not that the visitor came to your site and had an engaging visit, but that they came to the site at all, and what their likelihood is to come back – it’s the engagement of the relationship as opposed to the visit.
So for example, you have two campaigns generating visits. You can look at the visits and find Campaign 1 generates a more engaging visit than Campaign 2, based on the level of interaction with various activities. So you judge Campaign 1 better than Campaign 2.
However, just a week later you find the average days since last visit of visitors from Campaign 1 is 5 days, where the average days since last visit of Visitors from Campaign 2 is 2 days. These metrics are predicting that the visitors from Campaign 1, though engaged in the initial visit, are less likely to come back and visit again. 30 days later, you find in fact the prediction was true – only 10% of visitors from Campaign 1 ever came back, whereas 40% of visitors from Campaign 2 keep coming back.
The importance of the distinction between engagement with the visit and engagement with the relationship will depend, of course, on your business model. But in most cases, a site is more optimized when more visitors come back, and the point of the prediction above is you can change what you are doing before you waste of lot of time or effort.
You may find, for example, that days since last visit for new visitors who download a PDF is much higher than days since last visit for new visitors who play a video. That tells me I want to push the video to new visitors over the PDF, and changes should be made to the site.
Sir,
I have recently starting reading your news letter, Though have not read your book, but i am planning to buy one and go through it .
I based out in india and am taking care of product marketing telecom – Prepaid product.
In this market, there is high level of competition resulting in external churn.and also since high commissions are involved results in to internal churn.
We Map churn to validity expiry of the customer, and categorize it in to new customer validity expiry(Customer entering grace first time) and existing customer validity expiry (Customer entering grace 2nd time onwards).
The existing customer validity recharge is at 56% and new customer recharge is at 50%. This level we have reached from a level of 38-40% in last 1 year.
we have a grace period of 7 days after which we can not contact the customers. We are sending SMS and doing telecalling to our customer bases 3 days before validity expiry and in 7 days grace period also.
how can we take the validity recharge further, in both new and existing base
please advise
Sarab
Unfortunately, this business model is “built for churn”, there is really not much you can do outside of pricing / bundling, for example, better prices if they continue to renew.
If you are willing to take a hit to sales in favor of boosting profits, I would look carefully at the sources of new customers. Some sources will have higher churn rates than others, and some will be particularly bad. Cut out those sources and your churn rate should fall, increasing profits.
Hi Jim,
You are dead on as usual. It is about Customer States. Regardless of the segment, regardless of the brand messaging schedule, the consumer must in a state where they will be receptive to the message you are sending. No point in sending you a discount for new tires if you just bought some last week.
To this end, we have developed a Customer State Marketing engine. It allows marketers to include the concept of customer state when they develop their marketing programs. I would be greatly interested in getting your comments, as an expert in the field, on our product and our approach.
Thanks for the great posts.