LTV:CAC Ratio Is a Sham (Use Payback Period Instead)

You may have heard that a “good” LTV:CAC ratio for a SaaS startup is 3:1 or more.

Well, I’m here to tell you something contrary.

The LTV:CAC Ratio is an utterly useless metric.

Unlock this article and many more with a membership

Become a member
Already a member? Log in

You may have heard that a “good” LTV:CAC ratio for a SaaS startup is 3:1 or more.

Well, I’m here to tell you something contrary.

The LTV:CAC Ratio is an utterly useless metric.

In fact, it’s destructive.

We should wipe it from our vocabulary completely.

The main issue is with the definition of lifetime value.

Traditionally, lifetime value was defined as the average total revenue collected from a customer over time.

As acquisition costs rose and marketers could no longer justify spending money to acquire a customer based on their first purchase, they invented a new metric to show profitability over time: The LTV:CAC Ratio.

Maybe a customer wasn’t profitable to acquire based on their first purchase, but over their lifetime and through multiple purchases, their value outweighed the cost.

It’s fine for ecommerce and other non-subscription business models. But it never should have made its way into the world of SaaS.

In SaaS, we calculate lifetime value differently to try to account for the nature of subscription revenue:

Average monthly revenue per customer / average monthly customer churn

This seems acceptable until you realize a few things:

  1. It assumes every customer eventually churns (not true)
  2. It assumes customers churn at around the same time on average
  3. It doesn’t account for multiple plans with huge variances in prices
  4. It doesn’t account for a free plan or free trial

Let’s dig into the nuances of each and then I’ll tell you why Payback Period is a perfect alternative to LTV:CAC.

LTV:CAC assumes every customer eventually churns

The assumption that every customer eventually churns is literally not true.

Colin shows Customer.io's cohort retention over several years. The bottom orange layer shows steady revenue all the way back from their 2013 cohort.

In fact, they have exceptional expansion revenue with most cohorts GROWING in revenue over time.

A more accurate representation of LTV:CAC would be if you calculated the cumulative LTV of each annual cohort and compared that against the CAC for each year.

But what you’d still find is that LTV:CAC keeps getting better and better for each cohort since customers stick around for years and years.

And I struggle to find how that’s useful information.

It’s simply not a true or useful concept to make decisions on.

LTV:CAC assumes customers churn at around the same time on average

Beware of averages.

The danger in relying on averages is that the range of historic outcomes may be very wide. Too wide to be able to accurately represent a sample. ****

It's also quite likely the outcome will be nowhere near the historic average.

Since the traditional calculation of LTV is based on ARPU / User Churn, we need to take a closer look at churn and why it’s used.

ConvertKit’s User Churn report shows that with a monthly churn rate of 3.4%, customers will churn out after 2 years, 5 months, & 1 week on average.

So what the original equation of LTV = ARPU / User Churn is saying is ARPU x 27 (27 months = 2 years, 5 months, 1 week-ish).

But if you conduct a cohort analysis, you’ll see a much different story.

25% of customers will be gone within 3 months and 50% gone within 12 months.

After looking at thousands of tables like this in my time as Head of Growth at Baremetrics, I can tell you that more churn occurs in the first 3 months of the customer lifecycle than in any other period.

It makes sense too. The first 3 months will weed out the majority of bad fit customers, price-sensitive customers, customers who never truly got onboarded, what are effectively paid trials, and customers who ended up switching to a competitor.

So the reality is that only a fraction of your original customers will still be around by the time your supposed “average time to churn” comes around.

You’ll be making decisions based on a small subset of sticky customers, which, in and of itself isn’t a bad thing, but it’s contrary to how the data is portrayed.

Going back to the warning about averages... the problem is that the range of data gets wider and wider as time goes on, thus skewing the average further and further away from a true representation of what you’re looking for.

On a customer-by-customer LTV basis, we had an absolutely enormous range from just a few hundred dollars collected from new customers to tens of thousands of dollars from long-time customers.

LTV:CAC doesn’t account for multiple plans with huge variances in prices

This is the real kicker.

The reason why LTV:CAC works just fine for one-time sale business models is that price points are consistently close to each other. Even if you have hundreds of SKUs, it’s unlikely that they’re going to vary in price too drastically.

An ecommerce brand might sell A at $49, B at $69, and C at $79, for an average order value of ~$66.

You don’t have to be a statistician to understand that $66 is pretty representative of the true price points.

SaaS businesses regularly have multiple plans with huge variances in prices.

A SaaS business might sell A at $9/mo, B at $99/mo, and C at $999/mo, for an ARPU of $369.

You also don’t have to be a statistician to understand that $369 is not representative of the true price points.

Why does this matter?

Let’s say your CAC is $1,000 and you use $369 as your ARPU, divide by user churn of 3.4%, and get a LTV of $10,853.

10,853:1,000 = 10.8:1 = your LTV:CAC is amazing!

Think again. It’s only amazing for your highest-paying customers.

Spend $1,000 to acquire a $9/mo customer = horrible 👎

Spend $1,000 to acquire a $99/mo customer = decent 🤞

Spend $1,000 to acquire a $999/mo customer = amazing 🔥

LTV:CAC doesn’t account for a free plan, free trial, or a lengthy sales cycle

Again, with one-time sale business models, the revenue is collected immediately at the time of the sale.

Not so in SaaS.

A free trial can delay collecting revenue by 7-30 days.

A free plan can delay collecting revenue by 30-180 days.

A lengthy sales cycle can delay collecting revenue by 180-365 days.

So even if your CAC is reasonable for the amount that you charge customers on various plans, you still have to factor in the time it takes to start collecting that revenue.

Delayed revenue presents a huge problem for LTV:CAC because you have to be able to float the acquisition costs before recouping with the revenue new customers generate.

Taking the earlier example from above... a $999/mo customer might take 6 months to close from the time you spent $1,000 to get them in your pipeline.

If you’re acquiring 10 of those customers a month, you’ll be spending $60,000 over 6 months before those customers even start to generate revenue for you and recoup the cost.

Do you have $60k to float customer acquisition costs? LTV:CAC won’t help you figure that out.

Why Payback Period is better than LTV:CAC

The LTV:CAC Ratio is supposed to be a measurement of how long it takes a new customer to become profitable after recouping acquisition costs.

However, hopefully by now you’re convinced that there are too many flaws with LTV for the LTV:CAC Ratio to be even remotely helpful in figuring that out.

There’s a much simpler, much more reliable way of measuring how long it takes a new customer to become profitable after recouping acquisition costs.

It’s called the Payback Period.

It has similar roots to LTV in that it involves ARPU, except it skips all the roundabout calculations and gets straight to the heart of the issue.

Payback Period = CAC / ARPU

Isn’t that so much more straightforward?

To use the same numbers as before, if CAC is $1,000 and ARPU is $369, we get a Payback Period of ~2.7 months.

Looking at the Payback Period per plan paints a much clearer picture of how long it takes customers on each plan to become profitable.

$1,000 / $9/mo = 111 months (yikes!)

$1,000 / $99/mo = 10 months (manageable!)

$1,000 / $999/mo = 1 months (printing money!)

Ideally, you want your Payback Period to be between 3-12 months. Anything shorter and you don’t even have to blink twice about CAC. Anything longer and you’ll need a lot of cash and a lot of patience.

I love how ProfitWell helps visualize it on a graph in their article on Payback Period.

This makes it incredibly easy to model in Excel or Summit.

We can even layer in assumptions about a free plan, free trial, or lengthy sales cycle.

Let’s assume that, on average, we figure out that a free trial delays revenue by 1 month, a free plan delays revenue by 3 months, and a lengthy sales cycle delays revenue by 6 months.

$1,000 / $9/mo = 111 + 3 = 114 months (more yikes!)

$1,000 / $99/mo = 10 + 1 = 11 months (still manageable!)

$1,000 / $999/mo = 1 + 6 = 7 months (pretty good!)

If you want to model this even further, you can account for churn by “discounting” ARPU.

Churn eats into the true payback period since a portion of the customers you acquire will end up canceling and you can’t collect the revenue needed to recoup the cost of acquiring them.

Let’s call it Discounted Payback Period:

CAC / (ARPU x annual retention rate)

To use the same numbers as before, if CAC is $1,000, ARPU is $369, and annual retention is 85%, we’d calculate:

1,000 / (369 x .85) = 1,000 / ~314 = ~3.2 months

We’d tack on an additional .5 months compared to the ~2.7 months before.

This is far more helpful than the 10.8:1 LTV:CAC example we originally started with.

Now that you know how long it will take for a new customer to become profitable, you also know (1) if it’s even profitable in the first place and (2) how much cash you need to have in order to float the acquisition costs.

In both LTV:CAC and Payback Period, CAC stays the same. While it’d be ideal to be able to segment CAC based on customers on different pricing plans, that requires near-perfect attribution, which we all know is far from possible.

And while ARPU is used in both metric calculations, Payback Period more accurately accounts for expansion revenue since the final calculation isn’t heavily manipulated by User Churn.

To recap...

  • Aim for a Payback Period of 3-12 months
  • Account for delayed revenue with free plans, free trials, and lengthy sales cycles
  • Account for churn with Discounted Payback Period

For more tactical SaaS marketing advice and a community of founders & marketers to learn from, consider becoming a Swipe Files member.

You may have heard that a “good” LTV:CAC ratio for a SaaS startup is 3:1 or more.

Well, I’m here to tell you something contrary.

The LTV:CAC Ratio is an utterly useless metric.

In fact, it’s destructive.

We should wipe it from our vocabulary completely.

The main issue is with the definition of lifetime value.

Traditionally, lifetime value was defined as the average total revenue collected from a customer over time.

As acquisition costs rose and marketers could no longer justify spending money to acquire a customer based on their first purchase, they invented a new metric to show profitability over time: The LTV:CAC Ratio.

Maybe a customer wasn’t profitable to acquire based on their first purchase, but over their lifetime and through multiple purchases, their value outweighed the cost.

It’s fine for ecommerce and other non-subscription business models. But it never should have made its way into the world of SaaS.

In SaaS, we calculate lifetime value differently to try to account for the nature of subscription revenue:

Average monthly revenue per customer / average monthly customer churn

This seems acceptable until you realize a few things:

  1. It assumes every customer eventually churns (not true)
  2. It assumes customers churn at around the same time on average
  3. It doesn’t account for multiple plans with huge variances in prices
  4. It doesn’t account for a free plan or free trial

Let’s dig into the nuances of each and then I’ll tell you why Payback Period is a perfect alternative to LTV:CAC.

LTV:CAC assumes every customer eventually churns

The assumption that every customer eventually churns is literally not true.

Colin shows Customer.io's cohort retention over several years. The bottom orange layer shows steady revenue all the way back from their 2013 cohort.

In fact, they have exceptional expansion revenue with most cohorts GROWING in revenue over time.

A more accurate representation of LTV:CAC would be if you calculated the cumulative LTV of each annual cohort and compared that against the CAC for each year.

But what you’d still find is that LTV:CAC keeps getting better and better for each cohort since customers stick around for years and years.

And I struggle to find how that’s useful information.

It’s simply not a true or useful concept to make decisions on.

LTV:CAC assumes customers churn at around the same time on average

Beware of averages.

The danger in relying on averages is that the range of historic outcomes may be very wide. Too wide to be able to accurately represent a sample. ****

It's also quite likely the outcome will be nowhere near the historic average.

Since the traditional calculation of LTV is based on ARPU / User Churn, we need to take a closer look at churn and why it’s used.

ConvertKit’s User Churn report shows that with a monthly churn rate of 3.4%, customers will churn out after 2 years, 5 months, & 1 week on average.

So what the original equation of LTV = ARPU / User Churn is saying is ARPU x 27 (27 months = 2 years, 5 months, 1 week-ish).

But if you conduct a cohort analysis, you’ll see a much different story.

25% of customers will be gone within 3 months and 50% gone within 12 months.

After looking at thousands of tables like this in my time as Head of Growth at Baremetrics, I can tell you that more churn occurs in the first 3 months of the customer lifecycle than in any other period.

It makes sense too. The first 3 months will weed out the majority of bad fit customers, price-sensitive customers, customers who never truly got onboarded, what are effectively paid trials, and customers who ended up switching to a competitor.

So the reality is that only a fraction of your original customers will still be around by the time your supposed “average time to churn” comes around.

You’ll be making decisions based on a small subset of sticky customers, which, in and of itself isn’t a bad thing, but it’s contrary to how the data is portrayed.

Going back to the warning about averages... the problem is that the range of data gets wider and wider as time goes on, thus skewing the average further and further away from a true representation of what you’re looking for.

On a customer-by-customer LTV basis, we had an absolutely enormous range from just a few hundred dollars collected from new customers to tens of thousands of dollars from long-time customers.

LTV:CAC doesn’t account for multiple plans with huge variances in prices

This is the real kicker.

The reason why LTV:CAC works just fine for one-time sale business models is that price points are consistently close to each other. Even if you have hundreds of SKUs, it’s unlikely that they’re going to vary in price too drastically.

An ecommerce brand might sell A at $49, B at $69, and C at $79, for an average order value of ~$66.

You don’t have to be a statistician to understand that $66 is pretty representative of the true price points.

SaaS businesses regularly have multiple plans with huge variances in prices.

A SaaS business might sell A at $9/mo, B at $99/mo, and C at $999/mo, for an ARPU of $369.

You also don’t have to be a statistician to understand that $369 is not representative of the true price points.

Why does this matter?

Let’s say your CAC is $1,000 and you use $369 as your ARPU, divide by user churn of 3.4%, and get a LTV of $10,853.

10,853:1,000 = 10.8:1 = your LTV:CAC is amazing!

Think again. It’s only amazing for your highest-paying customers.

Spend $1,000 to acquire a $9/mo customer = horrible 👎

Spend $1,000 to acquire a $99/mo customer = decent 🤞

Spend $1,000 to acquire a $999/mo customer = amazing 🔥

LTV:CAC doesn’t account for a free plan, free trial, or a lengthy sales cycle

Again, with one-time sale business models, the revenue is collected immediately at the time of the sale.

Not so in SaaS.

A free trial can delay collecting revenue by 7-30 days.

A free plan can delay collecting revenue by 30-180 days.

A lengthy sales cycle can delay collecting revenue by 180-365 days.

So even if your CAC is reasonable for the amount that you charge customers on various plans, you still have to factor in the time it takes to start collecting that revenue.

Delayed revenue presents a huge problem for LTV:CAC because you have to be able to float the acquisition costs before recouping with the revenue new customers generate.

Taking the earlier example from above... a $999/mo customer might take 6 months to close from the time you spent $1,000 to get them in your pipeline.

If you’re acquiring 10 of those customers a month, you’ll be spending $60,000 over 6 months before those customers even start to generate revenue for you and recoup the cost.

Do you have $60k to float customer acquisition costs? LTV:CAC won’t help you figure that out.

Why Payback Period is better than LTV:CAC

The LTV:CAC Ratio is supposed to be a measurement of how long it takes a new customer to become profitable after recouping acquisition costs.

However, hopefully by now you’re convinced that there are too many flaws with LTV for the LTV:CAC Ratio to be even remotely helpful in figuring that out.

There’s a much simpler, much more reliable way of measuring how long it takes a new customer to become profitable after recouping acquisition costs.

It’s called the Payback Period.

It has similar roots to LTV in that it involves ARPU, except it skips all the roundabout calculations and gets straight to the heart of the issue.

Payback Period = CAC / ARPU

Isn’t that so much more straightforward?

To use the same numbers as before, if CAC is $1,000 and ARPU is $369, we get a Payback Period of ~2.7 months.

Looking at the Payback Period per plan paints a much clearer picture of how long it takes customers on each plan to become profitable.

$1,000 / $9/mo = 111 months (yikes!)

$1,000 / $99/mo = 10 months (manageable!)

$1,000 / $999/mo = 1 months (printing money!)

Ideally, you want your Payback Period to be between 3-12 months. Anything shorter and you don’t even have to blink twice about CAC. Anything longer and you’ll need a lot of cash and a lot of patience.

I love how ProfitWell helps visualize it on a graph in their article on Payback Period.

This makes it incredibly easy to model in Excel or Summit.

We can even layer in assumptions about a free plan, free trial, or lengthy sales cycle.

Let’s assume that, on average, we figure out that a free trial delays revenue by 1 month, a free plan delays revenue by 3 months, and a lengthy sales cycle delays revenue by 6 months.

$1,000 / $9/mo = 111 + 3 = 114 months (more yikes!)

$1,000 / $99/mo = 10 + 1 = 11 months (still manageable!)

$1,000 / $999/mo = 1 + 6 = 7 months (pretty good!)

If you want to model this even further, you can account for churn by “discounting” ARPU.

Churn eats into the true payback period since a portion of the customers you acquire will end up canceling and you can’t collect the revenue needed to recoup the cost of acquiring them.

Let’s call it Discounted Payback Period:

CAC / (ARPU x annual retention rate)

To use the same numbers as before, if CAC is $1,000, ARPU is $369, and annual retention is 85%, we’d calculate:

1,000 / (369 x .85) = 1,000 / ~314 = ~3.2 months

We’d tack on an additional .5 months compared to the ~2.7 months before.

This is far more helpful than the 10.8:1 LTV:CAC example we originally started with.

Now that you know how long it will take for a new customer to become profitable, you also know (1) if it’s even profitable in the first place and (2) how much cash you need to have in order to float the acquisition costs.

In both LTV:CAC and Payback Period, CAC stays the same. While it’d be ideal to be able to segment CAC based on customers on different pricing plans, that requires near-perfect attribution, which we all know is far from possible.

And while ARPU is used in both metric calculations, Payback Period more accurately accounts for expansion revenue since the final calculation isn’t heavily manipulated by User Churn.

To recap...

  • Aim for a Payback Period of 3-12 months
  • Account for delayed revenue with free plans, free trials, and lengthy sales cycles
  • Account for churn with Discounted Payback Period

For more tactical SaaS marketing advice and a community of founders & marketers to learn from, consider becoming a Swipe Files member.