|Turning Customer Data |
into Profits with a Spreadsheet
The Guide to Maximizing Customer Marketing ROI
Note the time between events gets shorter and shorter, then around the fourth event, begins to lengthen. This is a classic customer LifeCycle, and the last half of the pattern is typical of what you see when a customer is about to defect and stop or substantially slow down doing business with you. The time between events gets longer and longer until no next event takes place at all.
Will all customer LifeCycles look like this? No; this is the average behavior of a customer in your business as it stands now. Some customers may cycle more rapidly, others more slowly through these events.
Customer LifeCycles are a reality: there is going to be a LifeCycle and you will not be able to stop it. You probably don't know about LifeCycles. because you have not measured them;perhaps you don't know how. Besides, you don't even hear the pundits talking about them.
This is most amusing given all the jaw flapping and tongue wagging about LifeTime Value; if you don't understand the customer LifeCycle, how would you ever know when the "LifeTime" was over to measure value? The plain fact is people have it backwards; LifeTime Value is the last thing you want to try to wrestle with when just starting out with customer relationship and value management. You start with the LifeCycle, and only after fully playing out that card, do you move on to the idea of LifeTime Value. You do not have to mess around with calculating absolute customer LifeTime Value to be successful using data-driven marketing. If you don't believe this, you must have missed the tutorial which explains this concept in detail.
Customers are not just customers one day and then not the next day; there is a process to customer defection, and the smart data-driven marketer creates High ROI Customer Marketing programs by taking advantage of understanding the complete customer defection process.
There are two ways you can increase the value of customers:
1. Extend the customer LifeCycle, leaving
2. Increase the value of the customer within the existing LifeCycle. The customer still defects pretty much on schedule, but you have done everything you can to increase their value before the defection happens.
The first approach usually requires some pretty sophisticated tools and can be expensive; loyalty programs are a classic example of extending the LifeCycle. Not for the faint of heart financially and organizationally, loyalty programs also do not work well for every type of business. But they do work and can be extremely profitable if they are designed and executed correctly. If you are interested in how this type of loyalty program is constructed, here are some good examples.
The second approach to increasing customer value above is easier to execute, and for many companies, is the right way to go. It involves what I would call a customer retention program as opposed to a loyalty program, and this is how you go about setting it up.
The first place I would look to address the above customer LifeCycle is the fourth event. Why? This event looks to be the one that sets up the beginning of the defection, since it is this fourth event which starts out the pattern of longer and longer times between customer activity events.
For the average customer, this fourth event happens at 180 days after the first event. We can assume some customers reach the 4th event before 180 days, and some reach it after 180 days. Any customer who is 180 days old and has not yet made a 4th purchase, a 4th visit to the web site - whatever the event is you are tracking - is acting outside the behavior of the average customer and is a prime candidate for an earlier than normal defection. This is where you focus your efforts. You set up this fourth event as the "trip wire" - if the customer doesn't trip the wire by engaging in the 4th event by day 180, you take action and affect this behavior.
If you can save just a small percentage of defecting customers, the ROI can be very high, because these customers represent "found profits" which would not have existed without your efforts. And yes, you can measure and track these found profits - I am going to show you how to do this below.
1. Don't spend until you have to
You don't have to spend on customers who make the fourth purchase or visit within 180 days, because they are acting like "average" customers. Why spend on them if everything there is OK and they are behaving normally? You want to concentrate your spending where it will have maximum impact - on the customers who "roll over" the 180 day barrier without engaging in "average" behavior. These customers are the most likely candidates for a complete defection, and by focusing your resources laser-like on these people, you can spend more per customer and really have some impact on their behavior.
Put another way, let's say you have a customer retention budget of $20,000 and you have 20,000 customers. You currently spend $1 per customer each year sending all your customers the same lame retention stuff - statement stuffers that say you care and so forth. But if you could tell which 5,000 customers were the most likely to defect, and only spent on them at the point of maximum impact - when the defection was taking place - you could spend $4 per customer trying to stop or slow the defection with the same budget, have a much higher success rate, and actually realize the "found profits" I spoke of earlier. Make sense?
How To Execute a Latency Promotion
We'll use a retail example because the numbers are easiest to
understand and convey. But the same thought process is valid for
Utility / Telco,
/ Services, or Durable Goods / Long
Sales Cycle businesses.
2. Create the offer. In a retailing business, this could be as simple as a discount of some kind. You could sub-divide the 180 day old / no 4th purchase customers into "best" and "other," creating a VIP service offer to best customers and a discount offer to other customers.
3. Prepare the list. Select all your 180 day / no 4th
purchase customers, and then randomly
select 10% of them to not contact. This is called
your control group. People will tell you to only use 2% or 3% as
control, and statistically they could be right about
this. But the first
a. It's a "no argument" control group size. If your effort works and you can prove it, there won't be chattering from the sidelines about the possibility of a "defective" control group.
b. Why spend more than you have to the first time? By taking a large control, you reduce the number of people you are spending on to execute your promotion.
If you created the two groups "best" and "other," you need to take a 10% random sample of each.
The other 90% of a group is called the test group; they are the ones who will receive the promotion by direct mail, e-mail, other means.
The creation of proper control groups is absolutely essential to measuring the "found profits" referred to above. If this step has you puzzled, you can read more about creating control groups and random samples here.
4. Now you have two lists of people, control and test. Set up your tracking capability, which at minimum is the ability to run a report every 30 days that reveals the sales of each group starting from the beginning of the promotion, which is when you execute the e-mail, snail mail, or other communication of your offer to the test group. The metric you are interested in here is revenue per customer, so you would take the total sales of each group from the time the promotion is delivered and divide by the number of customers in the group, for both control and test groups.
5. Deliver your promotion to the test group.
6. Monitor the revenue activity of test and control groups. Run a sales report weekly or every 30 days, and look for divergence in the revenue per customer. The customers in the test group should be registering a higher sales per customer level (you hope). Keep running the report until the increase in revenue between test and control remains stable or begins to fall. When this happens, the LifeCycle of the promotion is over (promotions have LifeCycles too!). Let's say this takes 90 days, so 90 days after the event, you have a revenue per customer number since the promotion started, for both control and test groups.
7. Calculate ROI. I'll use some plug numbers as an example. The idea here is to compare the revenue behavior of the test group with the control group, and determine how much additional revenue occurred because of your promotion. Since the control group experienced no promotion, any difference in revenue between test and control can logically be attributed to the promotion. We then take out costs, and see if we added value to the customer LifeCycle - in more mercenary terms, did we make money or not?
Here's the key to the above. The people in control generated $30 in Gross Margin per customer over 90 days; the people in test generated $33 per customer. So $3 in additional Gross Margin per customer was created because of your promotion, since the two groups are the same in all other ways (if control was truly a random sample).
This $3 nets down to $2.50 because the cost of doing the promotion was $.50 per customer. Note: nowhere in here are we talking about response rates. Response doesn't matter; what matters is actual buying behavior. When you use control groups, you pick up buying behavior you never could have measured by just looking at response rates.
Now, the Per Customer Cost of Event is usually where you get into some arguments. If the event included a discount, the per customer cost of this discount must be included in the calculation of the true promotional cost:
Also, in the strictest sense, there is probably additional overhead attributable to the additional revenue: the cost to take the additional call and ship the box, the cost of additional salespeople needed to cover the promotion, and so on. These costs would not exist if you had not executed your promotion, so they should be included in the calculation to the extent you can calculate these additional overhead costs.
This $1.60 is profit after all expenses have been paid back. You have added $1.60 in value to the LifeCycle (and LifeTime Value) of each customer in the promotion.
To get to ROI, we need to look at what the promotion cost, and compare this to the value we generated; this is the definition of ROI. How much did we invest, and how much did we get back? We know what we got back $1.60 per customer Net of all costs, so we need to calculate total costs:
You spent $1.40 and you generated $1.60 after all costs. It's a 90-Day ROI because the additional revenue generated was measured over 90 days.
A 114% return is not something the CFO is going to be against, trust me. In fact, you could make the argument that since ROI in financial circles is usually measured on an annual basis, and this is a 90-day ROI, the real ROI here is 4x the 90-day ROI, or 456% on an "annualized basis." Not bad for a first-timer.
These are the found profits you have generated from your effort. By comparing the test group with the control group, you have proven these profits would not exist without your 180 day trip wire promotion. A smaller percentage of customers in the test group defected when compared with the control group; at least some portion of test made a purchase, and some kept right on buying for at least 90 days. These are found profits that would not have existed without your effort.
You have proven the 180 day / no 4th purchase trip wire promotion added value to the customer LifeCycle, a total of $1.60 per customer x 5000 customers = $8000 to be specific, and you did this without costing the company a single dime, since you paid back all your costs with profit from the promotion, and still had $8000 left over to put in the bank.
I can hear you now. C'mon Jim, looks good on paper, but 485% annualized ROI? An $8000 profit on a promotion that with every cost imaginable thrown in costs $7000? How is that remotely possible?
Folks, it's not just possible, this kind of return is normal in LifeCycle-based promotions. Remember the two rules of High ROI Customer Marketing:
1. Don't spend until you have to
By focusing your resources squarely on the problem, each dollar you spend works much harder. By waiting for the trip wire you narrowed the population you were promoting to, weeding out people you would normally waste money on. And by acting when the wire was tripped, you spent at the point of maximum impact. Efficient and effective.
Here is why this type of promotion makes so much money. It's anti-defection. You literally kept customers from leaving the company, and the control group proves this. The people you did not promote to in the control continued to slip away, while some portion of folks in the test group were stopped and their behavior reversed. This is where the huge returns come from - it's the relative spending disparity between the groups that creates the "found profits," which would have slipped away had you not done the promotion. It's a "tipping point" kind of idea - if you can be in the right place at the right time with the right catalyst, it doesn't take much change to create a big impact.
This promotion was not designed to extend the customer LifeCycle,
but to add value to the LifeCycle. Still, are you curious?
Did you actually extend the LifeCycle, and how would you measure this
effect? We'll answer these questions next month.
Copyright 2001, The Drilling Down Project by Jim Novo. All
rights reserved. You are free to use material from this
newsletter in whole or in part as long as you include complete