Average revenue per user, or per account, over a period: recurring revenue divided by the number of users or accounts. It is the numerator in lifetime value and most unit-economics work, and it is meaningful only when the population being averaged behaves consistently.

What is the difference between ARPU and ARPA?

ARPU divides by users, ARPA by accounts. In B2B the account is what pays and the distinction matters: an account with forty seats and one with two can produce the same ARPA and wildly different ARPU. Say which you mean, because people assume the other one.

Why is average ARPU misleading?

Because revenue distributions in B2B are not symmetrical. A handful of large accounts pull the mean far above what a typical customer pays, so the average describes a customer who does not exist. The median, reported beside it, usually tells a different and more useful story.

Does raising ARPU always help?

No. ARPU can rise because small accounts churned, which looks like improvement and is contraction. It can also rise through a price increase that raises churn enough to lower lifetime value. Read it beside account count and retention or it will mislead in both directions.

How does product work raise ARPU?

Mostly through expansion inside existing accounts: more seats, more usage, a higher tier reached because the product became load-bearing. That is a design and activation problem more than a pricing one, and it compounds where a price rise does not.

ARPU

Average revenue per user or account over a period — the numerator in most unit-economics calculations, and the one most often averaged badly.

Growth & Metrics

ARPU is average revenue per user over a period: recurring revenue divided by the number of users. Its sibling ARPA divides by accounts instead, and in B2B that distinction matters enough that the two should never be used interchangeably.

It is the numerator in lifetime value and in most unit-economics work, which means every error in it propagates.

The averaging problem

Revenue distributions in B2B are long-tailed. A handful of enterprise accounts sit far above everything else, and the mean sits between two groups rather than describing either.

The practical consequence is that a company's reported ARPU is frequently a number no customer pays. Reporting the median alongside it usually reveals the gap immediately, and where the two differ by a large factor the average has stopped being usable for planning.

Segmenting fixes it properly. ARPU by plan, by company size, by acquisition channel. Those numbers describe real populations and can be acted on; the blended figure mostly cannot.

Reading it in the right direction

ARPU moves for reasons that are not improvements, and the direction alone tells you nothing.

It rises when small accounts churn. The average improves because the bottom of the distribution left. That is contraction reported as progress, and it is the most common false positive on this metric.

It rises after a price increase that costs you customers. Higher revenue per remaining customer, fewer customers, and lifetime value possibly down. The metric improved and the business did not.

It falls when a cheaper tier succeeds. More customers, lower average, potentially a much larger business.

None of these are readable from ARPU alone. It has to be paired with account count and retention, or it will mislead in both directions with equal confidence.

What moves it, and what design touches

Expansion inside accounts. Seats added, usage growing, a higher tier reached because the product became load-bearing. This is the version that compounds, and it is largely an activation and adoption problem rather than a pricing one.

Mix. Winning more of the customers who pay more. An acquisition decision.

Price. The fastest lever and the one with a churn cost attached.

Design touches the first directly. A product that spreads from one user to a team, and from one workflow to three, raises ARPU through use rather than through a price list. A product that stays on one desk does not, whatever the pricing page says.

Where it belongs and where it does not

ARPU is an input, not a goal, and treating it as a target produces predictable damage.

It belongs in lifetime value, in pricing analysis segmented by plan, and in any comparison between acquisition channels, where a channel bringing cheaper customers is worth knowing about before you scale it.

It does not belong on a dashboard as a number to raise. Every direct lever on it — price rises, removing a cheap tier, qualifying harder — improves the average by shrinking the population, and all three look identical to growth on the chart.

The safeguard is the same one that applies to any metric: name the counter-metric when you choose it. For ARPU the counter is account count, because the fastest way to raise the average is to lose the customers at the bottom.

In practice

A company watches ARPU rise steadily for three quarters and treats it as evidence the product is moving upmarket.

Splitting the number ends that reading. Account count is flat, and the rise comes entirely from the bottom of the distribution leaving: the self-serve tier lost a third of its accounts over the period. The remaining base is more expensive per account and smaller, and total revenue is roughly where it started.

The metric had been reporting attrition as improvement for nine months, because nobody had looked at it beside the account count.

Where teams get it wrong

ARPU and ARPA used interchangeably. In B2B they answer different questions.
Mean with no median. A figure describing a customer who does not exist.
Reading the direction alone. It rises when small customers leave.
Blended across segments. Two populations averaged into neither.
Treating price as the lever. Expansion compounds; a price rise carries churn with it.

Learn more

Product redesign service →