A Deep Dive on Average Revenue Per User (ARPU)
And where it goes wrong
When I became an FP&A leader for the first time, I built our entire operating model from scratch. I had no context and no templates. And that blank workbook staring back at me was DAUNTING!
That’s why I’m excited to introduce our sponsor, Mercury.
Their VP of Finance, Dan Kang is sharing his personal template for building a financial forecast model.
Whether you’re preparing to fundraise, updating investors, or charting a new course for the company, you’ll likely reach a point where you’ll need to build a forecast model. It can help you align on clear business goals and understand why those goals matter.
The average age of an American who wears diapers is 22 years old.
This strange statistic highlights the inherent flaw of averages: they often gloss over crucial context. The same issue occurs in the world of SaaS metrics, particularly with Average Revenue Per User (ARPU), a number that can seem useful at first glance but often provides a misleading view of a company’s financial health.
The metric "ARPU" originated in the telecom industry in the early 2000’s. Companies needed a way to measure the average revenue generated by each individual subscriber, particularly during the early days of mobile phone subscriptions when companies were heavily reliant on user fees. It became a key indicator of the value each customer brought to the business, allowing them to compare performance across different subscriber groups and market segments. The first time I heard the metric I was a valuation analyst at a PE shop, responsible for monitoring three telecom companies - one in Brazil, one in Ukraine, and one in India. Change in ARPU was the front page metric we reported on each quarter. Full stop.
As our lives increasingly moved online, and we turned everything under the sun into a subscription (including your local car wash), the concept of ARPU migrated to other industries like online gaming, streaming services, and Software-as-a-Service (SaaS) companies, allowing them to measure the average revenue generated from each active user on their platforms.
Every SaaS company I know tracks ARPU almost religiously.
But I find this trend misguided.
In the words of my friend and data scientist Olga Berezovsky:
"ARPU is tricky, and in many cases, it's quite frankly useless for monitoring growth.”
Let's dive into five critical considerations for ARPU before you average yourself out of a job.
There’s No Benchmark for "Good"
Unlike other SaaS metrics, ARPU is not a rate; it's an absolute number.
This means that there’s no standardized benchmark for what a "good" ARPU should look like. The number fluctuates based on industry, pricing models, market segment, and location.
A $100 ARPU might be excellent in one sector, but disappointing in another. The lack of a universal benchmark makes comparisons almost meaningless unless you're looking at companies with identical business models, customer profiles, and market dynamics.
ARPU Can Go Up Even When Business Health Is Declining
ARPU can easily become a vanity metric. Instead of offering a meaningful signal of business health, it can actually increase when your company is in trouble. Here's an example from a friend’s startup: from January to February, their ARPU grew threefold—from $230 to $600 per customer. This sounds great, but did they upsell an existing feature or improve their pricing strategy? No. They simply landed one new large client that skewed the average significantly. This kind of growth might look good on paper, but it hides the fact that the underlying business hasn't fundamentally changed, and the overall customer base remains largely the same.
This example illustrates the flaw of ARPU—it reflects averages that mask important details. The truth is, in many businesses, revenue per customer follows a power law distribution, where a few high-value customers contribute the majority of revenue, while the rest contribute significantly less. Like movie box office revenues or city populations, most of the value is clustered among a minority, making the average essentially meaningless.
ARPU Alone Doesn’t Indicate Profitability
Another limitation of ARPU is that, as a standalone metric, it fails to tell you whether you're maximizing value from each customer or whether that value is cost-effective. ARPU only captures average revenue, without considering the cost associated with acquiring or retaining those users. As such, it doesn't provide the broader context needed to understand profitability. What if you are spending more to service the customer each year than you are getting from them?
It also doesn’t tell you if you’re maximizing revenue. Imagine you have 1,000 customers paying $15 per month. It’s highly unlikely that each customer values your product equally. In reality, some customers might be willing to pay much more, but ARPU doesn't account for this. If you don't have a pricing model that effectively segments these different levels of value, you could be leaving significant revenue on the table. Introducing pricing tiers that align with value delivered to different customer segments could help you better capture the true value being generated, but ARPU won’t guide you toward that decision.
High ARPU Customers Still Churn
A high ARPU can sometimes give a false sense of security about customer loyalty. You might think, “If a customer is paying us more, they must be happy.” But that’s not necessarily true. Often, high-ARPU customers churn for reasons unrelated to product value—they might be locked into a contract and decide not to renew, or they may become dissatisfied with your service. ARPU says nothing about satisfaction, stickiness, or retention risk.
In fact, ARPU could rise to an unsustainable point, due to over usage fees or aggressive price increases, and the customer starts to resent you. As a result, they churn out the top the first chance they get.
It’s important to remember Seth Godin's insight: "Averages almost always hide insights instead of exposing them."
ARPU Is Not LTV
Lastly, ARPU is sometimes confused with Lifetime Value (LTV), which can lead to misguided business decisions. Unlike LTV, which has a relationship with churn and can be paired with CAC to provide a picture of unit economics, ARPU is a one-dimensional metric. It can increase even if your overall customer count is declining or if churn is rising—meaning your average revenue per user looks better while your total revenue and growth trajectory might be suffering.
The Bottom Line: ARPU Is Best Used in Context
ARPU can be a helpful starting point for understanding customer revenue, but it should never be used in isolation. To really grasp your business’s health, look at ARPU alongside other SaaS metrics—customer counts, retention rates, CAC, and LTV—to get a holistic view. A high ARPU might indicate an opportunity to double down on value-based pricing, but it could also mask serious issues with churn or dependency on a few high-value customers.
Instead of relying on a metric that only scratches the surface, dig deeper. Averages are misleading by nature. Avoid the pitfalls of false positives and make better decisions for your startup's growth and sustainability.
Smart Stuff I Read at 2AM (Sources)
Run the Numbers
Apple | Spotify | YouTube
I was recently joined by Curt Sigfstead, the CFO of Clio, a leading vertical software provider for legal firms. This episode does a deep dive into stock-based compensation (SBC) and employee equity. Whew!!!!
The evolution of equity at tech companies
How leaders should budget for stock based comp
Clio’s recent fundraising process
Beating and raising to maintain investor confidence
Building trust with your colleagues
Quote I’ve Been Pondering
“When you understand that nobody wants to read your shit, your mind becomes powerfully concentrated. You begin to understand that writing/reading is, above all, a transaction. The reader donates his time and attention, which are supremely valuable commodities. In return, you the writer, must give him something worthy of his gift to you.”
-Steven Pressfield, Nobody Wants to Read Your Shit
Hey CJ, just found your cool blog! Very well explained, and an associated issue with these average metrics is that it doesn't really say much about customer segmentation and outlier values as you allude to.
Have you read Sam Savage's work on the Flaw of Averages? I highly recommend it. One of the most famous examples is the case of the statistician who drowns while fording a river that he calculates is, on average, three feet deep. Plans based on assumptions about average conditions usually go wrong!
ARPU is best used when determining the unit economics of your model.
ARPU can also include non-revenue data. For example, the ARPU can include a value for increased traffic leading to future revenue of user flow in a network. There is LTV there.