How To Use, and Misuse, Customer Lifetime Value (CLV)

Or, you could do it that way but why would you do that?

*** Updates to this page are likely to be especially significant while the page is being developed. Please do tell me when you find errors. I believe in being forthright but the good part of that is that I can cope it if you disagree or simply have spotted obvious errors that I have made. ***

Customer Lifetime Value (CLV) is a very useful metric. That said, it has to be one of the most mis-used of all the marketing metrics. Whenever anyone says CLV I cringe in fear that what they do will be truly appalling. I’d like to say this was just me being a professor and blowing up minor issues to grade down papers but much of the advice I see simply doesn’t make sense. This is true of work written by academics as well as practizing marketers. See my work on CLV in teaching materials with Charan Bagga — even Harvard Business School has some unhelpful, confusing, and mutually contradictory material on CLV. The extremely popular Starbucks case is widely used but, frankly, pretty unhelpful at CLV.

When it comes to searching online you will see broadly speaking two types of advice. Some bad advice, and some truly terrible advice. It is depressing to see if you care about the quality of marketing metrics. (It is quite good to read if you want to feel superior though so it isn’t all bad).

Want Some BAD CLV advice?

Just Google ‘CLV’. This came up for me: https://www.crazyegg.com/blog/customer-lifetime-value/
As of September 21, 2020 this said: “Customer lifetime value (or life-time value (LTV), is the average amount of money your customers will spend on your business over the entire life of your relationship”. We have a number of sloppy points in the text definition alone. Is ‘money’ profit? revenue? The writer’s use of language is, at best, confusing.

In my blog I have aleady shared my thoughts on Qualtrics’ advice.

Simple Marketing Metrics Good. Simplifying To Meaninglessness Not So Good

BTW many sources make the same mistakes/seem to share elements of the same bad advice. Consultants attempting to be thought leaders seem to converge on the same bad thoughts. I have many objectives to the way CLV is taught in instructional materials: http://neilbendle.com/teaching-clv-badly/. Other people have also messed CLV up: http://neilbendle.com/more-on-clv/. To be clear this section could go on for much longer but at some point I need to become constructive and say what I think should be done rather than just criticize others.

What Should We Look For When Using CLV

Firstly, the basic (good) idea of CLV is that it brings in long-term profitability. It is theoretically superior to Customer Profit as it covers mutliple periods. CLV gives a metric that allows us to better understand the full value of the customer to the company long term — not just over a short period of time.

CLV can be historic or forward looking. By this I mean CLV can be used to assess the value of a customer at the end of the relationship, e.g., ‘the customer was worth $150’, or can be a prediction of what they will be worth, e.g., ‘the customer will be worth $150 going forward’.

Forward looking CLV is generally harder to estimate — it is after all a prediction. Yet, forward looking CLV is more interesting as it is directly useful for decisions. Generally speaking historic CLV is probably most useful to gain a estimate about forward looking CLV, e.g., customer A looks like customer B who was worth $150 so customer A is like worth around $150.

Such a forward looking CLV involves lots of estimates/assumptions, so although it is theoretically a strong metric be careful when looking at CLV estimates. If you see CLV estimates push for more detail about how the metric was created. How much you plan to charge a customer in five years time is a hard question to answer.

Digression: A Common Mistake Using CLV

Don’t simply chose high CLV customers to invest in. This ignores fact that funds are scarce. This will bias your customer acquisition towards high value customers at exclusion of more numerous, but individually less valuable, customers. I will have more to say on this in future work.

What is CLV For?

As I have a clear idea what CLV is for, I am able to say that some advice is really bad. When the advice you give doesn’t do what it is supposed to do then it isn’t that useful.

My starting position is that the value of a customer could be useful for:

Revenue or Profit

Many use revenue and not profit as the ‘money’ being made by the firm. Think about this for a moment. If you try and work out what a customer is worth and you use revenue you will go bankrupt very quickly.

If you are willing to spend (something approaching) the amount of revenue coming in to acquire a customer you will never cover the costs of them being a customer. Revenue can be a useful metric but using revenue in CLV is shockingly bad. (Of course, there are counter examples of firms with practicaly zero marginal cost but let’s ignore them). Marketers often think finance people don’t treat them with sufficient respect. If you think revenue gives what a customer is worth to your firm you might want to consider that maybe your finance colleagues are right about you.

If You Don’t Discount

A second egregious problem is lack of discounting. Here is Hubspot’s advice as of September 21, 2020:

“To calculate customer lifetime value you need to calculate average purchase value, and then multiply that number by the average purchase frequency rate to determine customer value. Then, once you calculate average customer lifespan, you can multiply that by customer value to determine customer lifetime value” Hubspot.

To be clear this advice is comically bad in any organizations where customers have lifespans longer than a fruitfly. There is no real excuse for not discounting cashflows that occur over many years. Would you be fine if someone borrowed $10 now and promised to give you $10 back in 20 years even assuming you completely trusted them to do as they say. Of course, not, you’d be sure that the returned $20 won’t buy as much in twenty years time as it buys now.

Not discounting means that the resulting CLV:

I   Can’t be used to decide how much to invest in current customer

F  Can’t be used to decide whether to fire current customer

A  Can’t be used to decide how to spend on acquiring new customer

V  Can’t be used to value customer base

A  Can’t be used to assess assets for financial accounts

What exactly is a ‘non-discounted CLV’ for then? The only thing I can think of is that you need some numbers in your report to look professional and you can’t be bothered to put meaningful numbers in.

Subtracting Acquisition Costs Before Reporting CLV

This final popular way of doing CLV is a less obvious error but is a bit more in the category of ‘you could do that, but why would you?’ People subtract acquisition costs from CLV before reporting the number to give what I will call ‘net CLV’. (Proponents just would call it CLV).

“NET CLV” = CLV minus acqusition costs

Doing this involves an additional step and makes the metric a lot less useful. Net CLV limits uses of CLV while at the same time encouraging users to make mistakes especially in respect of sunk cost bias. (I have sunk cost bias in my cartoon book, Behavioral Economics for Kids which is available for a nominal amount — occassionally free — on Amazon).

What is Wrong With ‘Net CLV’?

Subtracting acquisition costs, which are irrelevant costs, going forward means ‘Net CLV’ is not useful for any decisions involving current customers – i.e. the IF & VA of IF AVA.

‘Net CLV’

I   Can’t be used to decide how much to invest in current customer

F  Can’t be used to decide whether to fire current customer

V  Can’t be used to value customer base

A  Can’t be used to assess assets for financial accounts’Net CLV’ can be used for acquisition using the rule acquire if CLV>0. (This rule is actually incorrect when funds are scarce – i.e. pretty much all the time — but I agree it is not as bad as some of the other errors I’ve highlighted on this page). What I say in reponse is that subtracting acquisition costs adds no value so why do it?

‘Net CLV’ can theoretically be used for acquisition but CLV can do everything that ‘Net CLV’ does. Simply comparing CLV to acqusition cost replicates what ‘Net CLV’ does. The CLV I recommend is easier to create, does not encourage sunk cost bias, and can be used for so many more purposes than ‘Net CLV.’ So what exactly is the point of ‘Net CLV’.

Please Don’t Subract Acquisition Costs from CLV Before Reporting it

Please do not subtract acqusition costs from CLV before reporting it. It is harder to do, encourages sunk cost bias, and makes the metric not useful for many purposes.

Given that it is very simple to compare CLV to acquisition costs, there seems no benefit to subtracting acquisition costs before reporting CLV.

Neil Bendle and Charan Bagga 2016

How to Calculate CLV

You can calculate historic CLV from data.

The basic formula for forward looking CLV is:

CLV($)=Margin($)∗(Retention Rate (%))/(1+Discount Rate (%)-Retention Rate(%) )

Retention is % of customers at risk of stopping being customers who stay. (Churn measures customers at risk who leave, so 1- Retention Rate). The formula above is the infinite life formula. This projects returns to infinity (which makes the math easier and makes less of a difffernce than one might sometimes expect.

An initial margin is sometimes claimed, i.e. formula + margin. In such cases CLV can be written as:

CLV($)=Margin($)∗(1+Discount Rate (%))/(1+Discount Rate (%)-Retention Rate(%) )

You Can Change the Strong Assumpions

There are some strong assumptions when using the formulas above. An obivous strong assumption is that the retention rate will be stable over time. Yet, it is quite reasonable to argue that retention rate could be low in initial periods and get better as the customer pool is left with only the most committed customers.

If you don’t believe that these assumptions will hold the basic idea behind CLV remains the same. In such cases just create any CLV estimates using a spreadsheet or similar technolgy that allows you to change values per period. This doesn’t change what CLV is, only how you estimate it. If I say your assumptions are wrong behind you multiplication this doesn’t mean that the idea of multiplication no longer works. The theory behind CLV is sound but of course sometimes (often) people get the numbers wrong.

Communications Matter

In some financial valuation models CLV can sometimes be applied to future prospects to help estimate the value of the firm. This is great but there is absoultely no need to change the CLV formula to do this. Estimate cashflows from future customers easily using CLV. Estimate acquisition costs for the same future customers. Together these inputs allow you to calcualte the value of customer growth.

When acqusition costs are netted off before reporting CLV the resultant number cannot be used for decisions about current customers. This gets you to the bizarre situation where your estimates of CLV do not apply to customers, only to current non-customers, i.e. potential future customers. Such lingustic contortions seem odd to me and seem likely to hamper usage of CLV.

A solution to this I have seen is to call the value of current customers; residual lifetime value, to distinguish their values from CLV. This seems like a solution in search of a problem to me. In addition residual lifetime value seems like a term invented by finance people to apply to machinery etc… presumably because it is. It is vaguely demeaning of current customers as it seems to imply the current customers are a bit ‘used up’. I see one objective when advocating for more use of CLV being to improve the way we look at the importance of customers to the firm. Current customers being seen as shop-soiled seems to work against this objective. Why do this?

Digression the Common Language Dictionary

Use the Marketing Common Language Dictionary to look up the meaning of marketing terms.

Common Language Marketing Dictionary

Here Is A Handy Flow Chart

The Sloan Management Review published a handy flowchart I produced with Charan Bagga.

For Some More On CLV

http://neilbendle.com/category/marketing-metrics/customer-lifetime-value/

Summary: CLV Is Useful But Be Careful

Consultants, marketers and academics have all put out pretty awful CLV advice. When using CLV you should be aware of this. If you want the metric to be useful please consider this advice.