Andrew M. Heller Professor at the Wharton School, Senior Fellow Leonard Davis Institute for Health Economics Co-Director, Mack Institute of Innovation Management The Wharton School

[MUSIC]

All right, we're back.

And now we're going to talk about customer retention.

And I hope you know where I'm going with this.

There's really two things that I'm going to see through all of this

discussions.

I'm going to begin by asking the same exact question that I asked about costumer

acquisition, which is, what metric do we use to gauge and

guide our retention activities?

And so once again, there's a very common metric out there.

In fact, it goes by several different names,

some companies call it a retention rate, some call it churn rate or attrition rate.

Lets just explain what it is real quickly,

although you probably get the idea if your not familiar with it already.

We look at all the customers we had, say at the beginning of the period.

We ask ourselves how many of them stayed with us?

So what percent of that original chunk of customers stayed with us?

That's the retention rate.

Of course, we can look at what percent of customers left.

And that goes by different names, as we said, a churn rate, or an attrition rate.

So either way, whether you look at the glass is half full, or

half empty, it's that kind of metric that firms use to say how

good of a job are we doing at keeping our existing customers.

So again, very common metric and here's my view on it.

Unlike CPA, cost per acquisition, which I view as a very bad metric,

I see the retention rate or the churn rate as a pretty good metric, but

you have to be careful about how you look at it, how you use it, and

how you make decisions based on it.

So I'm going to start with a real world example.

One of the things that I like to do is I like to look at the information that

public companies put out there about their retention rates or their attrition rates.

What do they say about how many of their customers stick around and

what implications does it have?

How do they manage around the attrition rate?

Instead of viewing it as just a nice to know number,

let's understand how we can actually use that information to go all the way to even

coming up with a financial value of the entire customer base.

So here's the example that I'd like to work with.

The example comes from Vodafone,

a big cell phone operator I'm sure many of you are familiar with.

And they actually put statements out there to their investors to say,

here's our attrition rate and here's how it varies over time.

And so you can see the graphic right here and from a very kind of quick look,

it appears that their attrition rate on an annualized basis, as we look at it from

one quarter to another, is around 20%, a little bit lower, but around 20%.

So let's think for a second, about what that means?

Okay, so if we are losing around 20% of our customers at any given time,

how long do we expect those customers to stay with us?

Pretty simple math question.

Let me make it even simpler.

Suppose the attrition rate were 50%,

suppose we're losing half our customers every period, God forbid.

Then how long is a typical customer staying with us?

So if our attrition rate is 50%,

then a typical lifetime would be around two years, right?

So if the attrition rate's around 20%, the typical lifetime is around five years.

It's a little bit less than 20%, it'll be a little bit greater than five years.

So if we want to come up with kind of our quick and dirty valuation, not just for

a given customer but for

the entire customer base, if our attrition rate is a little under 20%,

it means the typical customer's with us for a little over five years.

We can multiply that by the amount of revenue per customer, multiply that

by the size of the customer base, and boom, that's our customer equity,

that's the value of the firm, at least again, as a first pass approximation.

And I don't want to understate that.

There's a lot of companies out there that are doing exactly that kind of calculation

to figure out what their customers are worth.

So here's my question.

What's wrong with this picture?

That sounds like a very nice calculation, right?

It's great to see their attrition rate, but what's wrong with it?

What's missing?

What is it that we really want to see?

I contend that it's not enough to see their attrition rate for the customer base

as a whole, but what is it that we celebrate in the customer eccentric world?

You got it, we celebrate heterogeneity, right?

We don't wanted to see a single number, we don't want to say what is the attrition

rate look like for in average customer, because there is no average customer.

We want to know how that attrition rate,

how that attrition propensity varies across the customers.

So here's what I want you to think about,

and this is a very very important question.

How does the attrition propensity vary across the customers?

Just imagine, if we can reach into the mind of each and every customer and

pull out just how churn prone or

not churn prone they are, how likely they are to leave at any given time.

And we look at those numbers across the customer base.

What will that distribution look like?

Will we tend to have a lot of very churn prone customers?

Will we have a lot of customers who tend to stick around for awhile?

Will it be kind of a nice bell-shaped normal distribution?

Who knows?

Well, I know and you know and I think it's a very important question.

So let me show you what it looks like for Vodafone, but before I do,

I want to emphasis to you that the figure you are about to see is actually very,

very typical.

This is the basic shape that we see for almost all businesses,

not just in telecommunications, not just for a phone company that primarily

operates in Europe, but for pretty much any company that operates

on some kind of contractual subscription basis, and here's what it looks like.

Here's the celebration of heterogeneity for Vodafone.

So don't ask me where the numbers come from, again,

these are numbers that Vodafone calculated on their own.

And to their credit, shared with their investors, and

they broke their customer base into three groups.

And they found that these three groups vary in terms of their churn or

attrition rates.

They found that there's one real small group that has a very high Attrition Rate

because of those people are very likely to leave the next possible opportunity.

And then there's this middle sized, middle attrition group, and

then there this largest group to the left that has a fairly low attrition rate.

So first thing I want to ask you is, is this good news or bad news for Vodafone or

for most companies?

And the answer is yes, it's a pretty good news and

suggest that most of the costumers are tend to stick around for a while and

don't have a propensity to leave right away.

Now, the next question is, why is that?

I mean, again, it's great to know.

But what is it about those customers?

What words would be used to describe the customers

with the fairly low attrition rates?

What most companies, what most managers like to use would be a word like, loyal,

those customers are really loyal.

They love us.

They'll run through the gates of hell to stay with us.

Maybe, that's probably true for some of the customers, but

what other words could we use to describe them.

How about words like lazy, inertial, indifferent.

Maybe they just don't care very much.

They're not very involved with this particular product of the service.

They're working with the service provider and

it doesn't really matter enough to them to decide whether to stay or whether to go.

They just don't care very much.

Now, I don't know what the breakdown is here.

We have this big bar to the left, some of them are loyal, some of them are lazy,

I don't know and for the purpose of this exercise, I don't necessarily care.

But I do want to emphasize that when we see all of those costumers who tend

to have a low attrition rate, it's not necessarily a reflection of

great marketing or stong branding on our part, although it might be.

It might be that a lot of customers just don't care very much.

And so let's keep that in mind.

So here is my question, now that we're celebrating heterogeneity,

what difference does it make?

How are we going to use the information that you see on this chart in order to

make a more informed assessment of what the customer base is worth?

So when you make a little bit easier for you.

In addition to the graph, I'm just going to look carefully at the graph and

pick off the various numbers, you see this table over here that shows you

both the size of each of the three groups, what percent of the customer base is

associated with each of these three groups.

And the attrition rate again, as we see a high, medium, low risk of churning.

So let's do the math right now, in fact, take a moment and think about

how you would take this information that's in front of you and combine it together

in order to come up with an overall value of the customer base.

What would you do?

Think about that.

Okay, let's talk about the calculation.

Here's what I suspect most of you did.

You probably took a weighted average.

You said look we understand the breakdown of our customer base,

70%, 20%, 10% into the low, medium, and high risk groups.

And then they have there associated attrition rates.

And see, you took a simple way that average.

You multiply .7 by .06 and so on, and so on, and so on.

You did the calculation and

it says right here the overall average attrition rate would be 17.7%.

And what is that average attrition rate tell us about

the overall length of the customer's life and

therefore the overall financial value of the customer base?

Well, once again one over 17.7%.

Do that calculation, you get 5.6 years.

Well, that sounds kind of familiar, right?

Isn't that just about the same number that we got

when we didn't celebrate heterogeneity?

So what difference does it make?

Why is taking heterogeneity to account can help us all here?

I'll tell you why, because you did the wrong calculation.

You did calculate in overall average, but

like we've said before there is no average customer.

Telling me that the average attrition rate is 17.7% doesn't do me any good

because there is no customer out there, at least according to the Vodafone analysis,

who has that kind of attrition rate.

So we've just calculated an expected lifetime,

an expected financial value for a customer who does not exist.

And so the question is, what's wrong with this calculation?

How do we do it the right way?

How do we truly celebrate heterogeneity?

because what we did here is we eliminated heterogeneity.

We just squashed it all together and said heterogeneity is gone and

that's why we ended up getting the same result

that we would have gotten if we didn't look at heterogeneity in the first place.

So what's wrong with this calculation and what's the right way to do it?

Think about that for a moment and then I'll give you a super big hint

from one of the world's leading thinkers about customer loyalty and so on.

So let's turn to the words of Frederick Rykeld,

who a number of years ago wrote a book called The Loyalty Effect.

And he laid out some many, many good ideas about what loyalty is,

how we measure it, how we capture it.

He's a consultant for Baines, so not an academic.

So he's really seen loyalty in action and has helped a lot of companies create and

monetize it.

And here is the quote from Reichheld and you can see it here.

And I hope the logic makes a lot of sense.

The average makes no sense at all,

we need to do the calculations separately group by group.

That's where the celebration of heterogeneity is going to come in.

Let's understand the separate value for the high, medium and low, and

then combine them together, instead of combining them together first.

I hope that makes sense.

Perhaps your inclination is to say, seems kind of similar.

Let's take an average and then calculate the lifetime.

Or let's calculate a lifetime and then take an average, seems kind of similar but

you know what, it's not.

So let's revisit this example, but do the lifetime calculation first, okay?

So for our low risk customers, their attrition rate is 0.06.

So what does that mean?

Just for those customers, how long are they going to stick around on average?

So what's 1 over 0.06?

Well, as you can see here, it's about 16.7 years.

And if we repeat that calculation for

the other two groups, you can see what their expected lifetime would be.

And here you see the really dramatic differences.

This is the celebration of heterogeneity.

We see an order of magnitude difference between the best and the worst customers.

We don't want to ignore that, we don't want to eliminate that,

we don't want to average over that, we want to celebrate that.

Once we see the expected lifetimes for

each of these different customer groups, then let's take the weighted average.

Then we'll multiply by 70, 20, and

10 percent, and when we do that what do we get?

Our overall expected lifetime for this customer base, 12.4 years.

That's a big difference, I'm sure you'll agree.

So instead of doing the calculation the wrong way,

where we weren't celebrating heterogeneity, once we acknowledge and

explicitly take into account heterogeneity,

we have more than doubled the value of our customer base just like that,

not by doing anything, just by doing the correct calculation.

This is the celebration of heterogeneity.

Now you might be wondering, all right so

this is this Vodafone example, how does this work in general?

Answer, always works the same way.

Whenever there's any head originate at all,

any spread among the customers in terms of their attrition rates,

there will always be money left on the table if we ignore heterogeneity.

We will always understate the value of the customer base by ignoring heterogeneity.

The only question is how much?

So in this case, it's a more that two fold increase.

And as you might imagine, the magnitude of that increase depends entirely on

the magnitude and nature of heterogeneity.

So, if the customers are more spread out, then ignoring heterogeneity

is going to cause us an even steeper understatement of the customer value.

So again, it's not a question of whether it will occur,

it's not a question of whether will be an over statement or

an under statement, it always works this way, it's just a question of how much?

Now, if that's not a celebration of heterogeneity, I don't know what is.

But this shows you, that by explicitly accounting for a heterogeneity,

by really understanding the differences among our good customers and our not so

good customers, we have just doubled the value in our company.

And you might wonder, so what are the implications of that for

managing our customer base?

This is really, really important, especially for

all of you who voted for customer retention

as the activity we should be putting our incremental dollars on.

So let's summarize our thinking about customer retention.

And there's two really important points I want to make over here.

First and foremost, there is no average customer, and

you can't do calculations based on an average customer.

Yes, it's easy.

Yes, it's convenient.

But it's wrong.

You will always understate the value of your customer base and

that difference can be huge if your ignore the heterogeneity.

And here's the second point that's much more subtle and

I'll spare you the painful math to get there.

But if we ever want to calculate an elasticity,

if we ever want to find out what's the incremental gain that we get for

say, a 1% reduction in the attrition rate.

It turns out we can do that calculation two ways.

If we do it by ignoring heterogeneity, as many firms do, or

if we do it by taking heterogeneity into account as we just did.

We get a very different conclusion.

So in some of my research, we'd actually done that calculation separately and

we show that the retention elasticity,

the gain that we get by lowering the attrition rate by 1% is much,

much less when you account for heterogeneity.

Which means that efforts to boost retention or

decrease attrition or churn are much more modest

than you think they are once you explicitly account for heterogeneity.

Now, I want to be real clear about this.

I am not saying that you should stop spending on retention.

No, I'm not saying that at all.

Again, retention is one of our three major pillars of customer centricity.

It is very, very important.

You want to figure out who the good customers are and

do whatever it takes to keep them around for a long time, you must do that.

But at the margin, companies seem to want to spend more, and more, and

more on retention.

That seems to be the constraint for them.

I contend that some companies might actually be overspending on retention.

And taking some of those retention dollars, which are often being allocated

to customers who aren't really that great, to customers who'll have a fairly high

attrition rate and are always going to have a high attrition rate,

even if we kind of bribe them or incent them to stay around for

another year or two, they're going to leave at the first opportunity after that.

We're better of taking some of those dollars and spending them where?

On customer acquisition.

Let's spend some of those dollars finding new customers who might be really good.

So right away you're starting to see some of the implications,

some of the trade offs between acquisition and

retention that arise when we have an explicit focus on heterogeneity.

So that's the retention story.

Let's take a short break and come back and talk about customer development.