ARPU / ARPA
ARPU and ARPA don't measure how much you sell. They measure how much value you extract, on average, from each active commercial relationship.
“A flat ARPA for eight consecutive quarters is not stability. It's an upsell engine that was never turned on.”
Otto GTM ObservatoryWhy to track ARPA over time, not in a single period
ARPA in a single month says little. Its trend across homogeneous cohorts says a lot: if the ARPA of the cohort acquired 12 months ago is higher today than at the time of acquisition, the company has a working expansion revenue engine.
Segmenting by product tier avoids the mix effect: if a new tier of enterprise customers enters the base, average ARPA rises even though no existing customer bought more.
Anti-patterns
Common mistake: looking at company-wide aggregate ARPA without segmenting by tier, confusing a richer sales mix with real per-customer value growth.
Second mistake: using ARPU instead of ARPA in multi-user B2B contexts, where a single account generates revenue from dozens of seats: the per-user figure artificially dilutes the real deal value.
Practical Application
A cohort of 40 customers acquired in January 2025 with an initial ARPA of €8,200/year reaches, after 12 months of customer success and expansion activity, an ARPA of €10,750/year on the same cohort: +31.1% value extracted without acquiring a single new customer.