4:43

I may then see a period of inactivity where the customer has not purchased

a ticket for six months.

In certain kinds of businesses it's quite difficult to determine whether or

not a customer is active or inactive.

In fact,

there are whole series of models that I'll talk about at the end that we could employ

to determine the probability that a particular customer is still with us.

Now, of course, if we have a business that's more of a subscription service,

then it's much more obvious whether the customer is active versus inactive.

If I discontinue my cell phone service with AT&T, they'll certainly know that.

If I cancel my subscription to The Economist,

The Economist will also know that but there are many situations

in which the status of the customer is not always clear.

In addition to calculating customer lifetime value, we also need a method or

a model or a way of thinking about the chance that a customer's, in fact, active.

If we combine those two pieces of information, the value of the customer

with the status, we yield some fairly interesting and intuitive implications.

Customers who are very high value and are currently active, we want to do everything

we can to foster and encourage loyalty from among those customers.

Customers who are relatively low value but actually fairly active, still

part of our system, transacting with us, clearly we want to increase their average

transaction size or the number of products and services that they buy from us.

Customers who are high value who are either inactive or

maybe customers belonging to competitors, we may want to focus on those individuals

with very specific and targeted promotions to switch them into our business.

Of course, in the final quarter, customers who are relatively low value and

who appear to have declined in the usage, or be somewhat inactive,

we may really just minimize the effort that we focus on those particular groups.

One thing I can't emphasize strongly enough is that customer lifetime value

is not only a calculation but it's a very important conceptual way of thinking about

what customers do for a business and why, in fact, we should cultivate customers.

When we think about customers as assets rather than just someone with whom we have

a transaction, it really changes the way we do the entire business.

I'll give a shout out to a friend of mine,

Sir Neal Gupta at the Harvard Business School,

wrote a very interesting book about ten years ago called Customers as Assets.

Really emphasizing the fact that it's not just the calculation but

it's the mindset that we have that becomes very critical if we want to

have a successful entrepreneurial venture.

What are some of the practical things we might do with a customer

lifetime value calculation?

7:20

Most fundamentally, we're able to answer the question, is it worth the effort and

at what price should we attempt to acquire a particular customer?

We can also ask and

answer the question, what is it costing us to retain a particular customer?

We can also, if we have the CLV data, we see the distribution,

from very low value customers to high value customers, or

even customers for whom the CLV might start to look negative.

We might want to get rid of those customers.

Finally, if we took the value of all of the individual customers put together,

this would give us an estimate of the value of the enterprise, and

a way of potentially valuing companies.

And mergers and acquisitions and so on.

So, these are just some of the many practical decisions

that you as an entrepreneur might want to take as a result of doing the calculation.

So with that in mind, let's now start to step through the most basic calculation,

and then I'll give you some other references for

those of you who like to do some homework on some more sophisticated approaches but

again the most important thing here is the concept.

And so, in order to embrace the customer lifetime value philosophy and

start to do the calculations, you're going to need some data.

And the bare minimum data has the following four components.

So first of all, you will need to know what kind of a margin

you're getting from a particular individual customer.

So this is just really the revenue that you're getting and

minus any variable costs you have of servicing that customer.

You also need to have some estimate of the chance that the customer will be retained.

What's the probability that given the option,

the customer renews a contract with you, or the customer remains a customer

in a certain time period if it's a non-contractual business?

I'll talk a little bit about where these numbers can come from also.

The third thing that you'll need to know is of course, the discount rate, or

the interest rate that you might like to apply.

Obviously, money that I get from a customer in the future

is not as value as the money that I'm getting today, and you may need to apply,

or you will need to apply, an appropriate discount rate.

We'll take the easy way out in our calculations and we'll just use 10%.

And then finally, you need to think about what's the right period over which

customer lifetime value should be calculated.

So, a recent consulting project I was doing for

a friend of mine who graduated here at the Warden School

was focused on a business where I calculated the CLV every quarter.

So these were sellers in a home shopping network and a holding trunk shows on

their home, and these sellers would have a trunk show every season to sell apparel.

So, I basically hit four observations per year, the time period was the quarter.

For your business, the time period might be the week, or it might be the year, but

it's a very important decision to think about choosing the right

time period over which we're going to do the calculation.

So, these are the four things that we need to know.

Some of them are more difficult to come up with than others.

Probably, the research would say the most challenging thing to figure out is

what is the right retention rate?

Now, one way we can get at the retention rate is the following.

Imagine we begin our business.

Let's make it easy, January 1st,

we have 100 customers that start with our business.

We then wait a particular of time let's say, in this example, one year and

we see from those initial 100 customers how many customers are still

remaining with the business.

If 80 customers are still in the business,

one simple estimate of the retention rate is 0.8 or 80%.

So looking at a core order of customers that entered our business at the same

time, waiting a certain amount of time to then see how many of those are left.

That's one way that we can estimate the retention rate.

So with this is mind, let's now go forward and do some calculations.

11:05

So in this diagram, we're going to visualize the process of revenue coming in

from a particular customer and I encourage you perhaps even to draw a diagram of this

sort for your own customers in your own business.

So let's assume a fairly simply example,

that the business that we're running is something like a mobile phone service and

that every year, the customer is spending $250 with us in terms of the net margin.

So the revenue may be a little bit higher minus some costs that we have of

variable costs maintaining that customer, we're getting $250 at margin.

Now, in addition, that cost is $400 in the beginning to acquire the customer.

Perhaps we had to give them a free cellular phone or something like that.

So, with this information in hand, over five periods, so

that's the lifetime that we're assuming that the customer is going to live for.

We can start to execute on the calculation of customer lifetime value.

So, let's go through this and do it together.

And let's do it in a very, very simple sense, so

we'll assume that there's no customer attrition.

Wow, wouldn't that be a great business.

The customer never leaves us,

stays with us with probability one every period for five years.

Secondly, let's assume that we're not going to do any discounting in

the calculation, that the money that we receive from the customer in year five

is just as valuable as the money that we receive in year one.

Of course our finance colleagues would not like us to make such an assumption, so

we'll modify that shortly.

But le's go ahead and do the calculation.

So, in this case, the customer lifetime value is just the five increments of $250

at margin, $1250 minus the acquisition cost

of $400 which leaves a net customer lifetime value of $850.

So the reason I'm showing you this example is as we start to layer in other things

like attrition and discount rate, what you'll notice is

that the customer's lifetime value the number starts to decline.

And it starts to decline quite dramatically so for

those of you who are students of business history which I hope many of you are,

if you think about things like the Internet boom and bust ,there are a lot of

companies during the 1.0, maybe still even today, that were grossly over valued or

grossly optimistic about what the value of the enterprise was because they made some

rather heroic assumptions about the values of their customers.

Either the margin that they were getting every period or the chance that

the customers would actually stay with the business and be retained, so.

That's one important point that I want you to look for, even in this very,

very simple example.

Just to see how dramatically the value is going to change

as we start to relax those assumptions.

So, let's go ahead and do that.

And so, now let's turn again to our familiar diagram where we have the flow

of margin coming in from the customer every period, but

this time around we're going to add one additional assumption,

which is that customers, unfortunately, might decide to leave us.

There's going to be some probability of attrition that's going to

cause the customer lifetime value number that we ultimately calculate to go down.

Now, this is something that actually is quite subtle and quite important.

And for those of you who are students of business history, which I hope many of you

are, if you think back to the boom and bust of things like Web 1.0,

part of the reason that happened is that people made gross overstatements or

grossly misstated assumptions about what the value of different businesses were.

Based on the underlying customer transactions and the underlying customer

asset in particular, they either assume that the margin that was going to

come in from the customer was much higher than what it turned out to

be because the customers couldn't be monetized in the way people thought.

Or the entrepreneurs though that the chance of retaining

customers was much higher than it actually was,

because they discounted the fact that a competitor could steal a customer, or

a customer might just leave of natural causes, doesn't really find the value in

the service that we, as the entrepreneurs would like to believe that they have.

And what you'll notice here, this is a very important element of the customer

lifetime value calculation is that even a fairly small degradation in the retention

rate can have a dramatic effect on the number that ultimately get's calculated.

So, lets go through and do this, and

then the example, we're going to assume a retention rate of 80%.

Now that's actually a pretty good number.

So, I think we'd be fairly pleased as your colleagues and

instructors in this course if there was a 80%chance from week to week that you

kept engaged with the class and applying the materials.

What we'll see however, is even with an 80% retention rate,

there's going to be quite a dramatic reduction in the value that we got

from the model previously, the value previously, was a total of $850.

Where we would assume customers would always stay with us.

For five years, and we also assume no time value of money,

we'll get to that one in a second.

So, what we're also going to do here, is you notice in the calculation is that at

the end of every period, the customer can either leave with probability 0.2,

or stay with probability 0.8, and

this process gets repeated until the end of the 5th period.

Now, those of you who're sort of looking at the slide,

there are those of you have studied a little bit of statistics in mathematics,

might say, hang on a minute, there's quite a simplifying assumption here.

You've assumed that the probability that the customer is there by the end of

year 2 is 0.8, at the end of year 3 is just 0.8 squared,

at the end of year 4 is 0.8 cubed.

So implicitly I've assumed

that the retention rates are independent from one year to the next.

Which seems like a pretty unrealistic, shall we say, assumption.

But, let's go back to where we started our session today,

with the notion that all models are wrong.

Thank you.

But some are useful.

So clearly it's a bit of an unrealistic assumption but it's made not only just for

mathematical convenience but

also because it may not in fact be such a bad assumption after all.

So imagine that my colleague Stephanie is a customer of AT&T.

The longer she stays with AT&T we might argue that her retention rates going up,

she likes the service, she's getting used to it,

she really doesn't want to go anywhere else.

At the same time, there could be other factors

causing her retention rate to be potentially going down.

Competitors like Verizon chasing after her.

She's just getting a little bit tired of the service, and so on.

So, as my colleague and friend, Sunil Gupta,

at Harvard Business School might argue,

there are forces pushing retention rates up, there are forces pushing them down.

So assuming that they're roughly constant, is not such a bad thing to do.

Now of course, getting the right number in the first place by doing as I said before,

looking at a cohort that entered at the same time period and

asking how many remained after a certain point through time

to calculate the retention rate that's probably the most critical thing of all.

So let's go through now and do the numbers.

I'm now just computing the expected contribution which is

the margin modified by the retention rate.

So I've done that there in the slide.

We add all those numbers up.

We get a net value now of $840, not $1250, and of course we have to

subtract out the initial acquisition cost, which I just assumed to be $400.

So now we have a customer lifetime value of $440.

Wow, that's a big drop from $850.

So you can imagine why this retention rate is just so critical.

In fact, the academic research suggests of all the four elements in

the mathematical formula, the one that produces the most leverage or

impact over the final number that you calculate is in fact the retention rate so

always be wary of somebody who's making an assumption about retention.

That's just too heroic or too unrealistic.

Because if they're doing that they're going to be grossly over estimating

the value of individual customers so as entrepenuers we always want to err

on the side of caution and off course to sensitivity.

Let's see what the result looks like if we assume 90%, or 80%, or 70%, and so on.

So we now have completed the second calculation, let's continue on, and

do a third one.

And so now on the screen in front of us we have the familiar flow diagram of,

margin coming in every period from the customer for five periods.

We also have on top of that the retention factor,

which we assumed in the beginning that they are with probability one.

And then in every period they have a chance of 80% of staying with us

20% of leaving, and on top of that I've computed the expected contribution,

$250 dollars, $200 dollars, $160 and so on down.

Now in this case we are just going to have one final component, which is something

that our finance colleagues, or your CFO at your start up, or your venture would

be really concerned about, which is of course the time value of money.

So we're retaining the assumption that customers will be with us,

with probability 80%.

So they'll churn or we'll lose them with probability 0.2.

In addition let's just assume that the value of money that comes in

is not as valuable in the future as it is in the present.

That's a pretty good assumption.

And let's have a discount rate of 10% just to make the math easy and

now let's go through and do that calculation so

the discount factor that's applied to the first piece of money that comes in at

the end of year one is just one divided one plus the discount factor which is 10%.

At the end of two years that's just again,

one, divided by one

plus the discount rate 10%, and

then we square the whole thing, and then we cube and so on.

What we see is we get now a value of $664.

We subtract out the $400 to acquire the customer.

Wow, almost nothing left.

We're down to $264.

So just think about this for a moment.

We've done a very, very simple and very stripped down example.

I really hope that you're going through and

thinking about your own customers in applying the same kind of logic and

we started with a customer lifetime value of $850.

We're now down to a number that's roughly about a third of that even with having

what seems to be a pretty good retention rate.

And also applying a fairly modest discount rate of

about 10% to the time value of money.

So, we can see that when one starts to really build in some more realistic

assumptions about whether or not customers will be retained and

the value of revenue that we might be getting in the future.

When you do those two things, you end up with a dramatically lower number.

That's really the bottom line here.

So, with this in mind, I'm now going to give you a couple of quick and

dirty simple ways to do this calculation without going through and

doing the adding up from every single period.

And also talk about some extensions that I would really love for some of you

to do for your homework, if you really need to dig a little bit deeper, and

go into a more sophisticated approach.

So now let me just give you one other very simple way,

just in a very rough sense, to calculate customer lifetime value.

In the examples that we just went through, remember that the customer

was sticking around for five periods, five years in our example.

We only added up the data for a period of five.

Let's imagine now however that the customer keeps on going, period six,

period seven, period eight, but

of course in every period, the chance that they stick around is declining,

declining, so it would be 0.8 to the five, 0.8 to the six and so on.

So, if we do then, we make that assumption that the customer is, in some sense,

almost going to be around forever, but with the declining chance every period.

So, in this case the customer lifetime value is simply the return, or the margin

divided by the churn rate, the churn rate is 0.2 or 1 minus the retention rate.

That gives us 1,250.

Subtract off the $400 acquisition cost.

If we wanted to then modify the formula again, assuming that the customer

is not there for five periods but six, seven, eight, keeps on going.

It's just going to be the margin $250 divided by the churn rate,

0.2 + the discount rate of 0.1.

So these are some really simple heuristics that you can use to apply to

to different customers to see which ones are more valuable, which ones are less so.

And of course, in all of these cases, whether you do the summing up.

Over a certain number of periods, or whether you use the direct formula

assuming that the customer is always going to be around.

Always be critical as entrepreneurs about the assumption around retention.

And always try to do some experimentation or some sensitivity.

24:12

And now as we conclude this lecture,

there is three things I would like you to think about.

And in fact to do as a homework exercise as we always do

at the end of our sessions together or the end of our time together.

So, let's try and take the customer lifetime value concept and

put it into action.

So first and foremost what I'd like you to do is to look at your own business as

an entrepreneur, and try and make those four decisions.

First of all, try to figure out what you think

the margin is that you're getting from every customer, number one.

Number two related to that, think about what is the right time period

over which you should be doing the calculation.

Number three, and probably most challenging, try to

develop an estimate of the probability that the customer is being retained.

And then number four, talk to your finance people, your CFO, and

think about what's a reasonable way to discount

the revenue stream that you're getting from the customers.

So try and put those four components together, and then start to apply those

calculations to the individual customers that you have at your particular business.

Number two, if you'd like to do it for a business that's already out there,

that I think some of you, at least in the United States might be familiar with,

a business that's challenging Gillette.

It's called Dollar Shave Club.

I'd like you to think what the customer lifetime value might be for

a customer who entered Dollar Shave Club.

Now I can assure you that a company like Dollar Shave Club

is most certainly thinking about customer lifetime value as a calculation.

So I'd like you to go through the exercise of trying to apply it to this

particular business.

As you do that, also think about which of the three razors in front of you

might be the one that's most likely to be purchased by most customers.

I'll give you a clue.

There's something called the compromise effect that suggests that customers might

gravitate towards buying the $6 option as opposed to the 1 or the 9.

And then finally, for some of you out there who have a strong background in

mathematics, or have a penchant to do things that are a little bit more

sophisticated, you've got the basic framework in mind, that's good.

All models are wrong, but some are useful.

You've got the basic elements and the philosophy, which is critical.

I would really encourage you to look up my friend and

colleague here at the Wharton School, Peter Fader.

F-A-D-E-R, Pete Fader.

He's probably one of the world's leading experts in the study

of customer lifetime value with more sophisticated mathematical modeling, and

Pete has kindly made most of the software available.