Japan's NRR looks healthy on your dashboard. Ask for the logo churn.

Your regional dashboard shows Japan at 104% net revenue retention. Every other region sits between 98% and 106%. Japan looks unremarkable, so nobody asks about it in the quarterly review.

Then a renewal is missed and a fifth of Japan’s ARR disappears in one month.

Nothing on the dashboard was wrong. The problem is that one retention number, applied to a young and concentrated country operation, cannot tell you what it tells you about your home market.

The two numbers point in opposite directions

They do so even in the aggregate. The Fullstar Customer Success Survey (2025 edition), run by CloudCircus with 200 customer success and support staff at Japanese B2B information and communications companies between 15 and 18 August 2025, published two headline figures side by side: a mean monthly customer churn rate of 3.01% and a mean net revenue retention of 102.1%. It is a vendor survey, and worth reading as one.

Read the first and Japanese SaaS is leaking customers every month. Read the second and Japanese SaaS is growing inside its existing base. Same 200 companies, same period, two instruments measuring different things.

The gap is expansion revenue. NRR counts what existing customers added; logo churn counts who left. Neither is the other’s sanity check.

For reference, Benchmarkit’s 2025 SaaS Performance Metrics (published May 2025, covering FY2024) reports a median NRR of 101% against a median gross revenue retention of 88%. Note that the effective sample differs by metric: 228 for NRR, 225 for GRR, not the 583 total responses on the cover. Strip out expansion and the typical company loses 12% of its existing revenue every year, then buys it back through upsell.

Why the gap is wider in Japan than at home

Three structural reasons, and none of them are visible in a regional column on a global dashboard.

Your Japan revenue is concentrated. Most foreign SaaS operations in Japan land a small number of large accounts first, because that is who will sign a contract with a company that has no local track record. When twelve customers make up the country’s ARR, losing one is 8% logo churn and can be 25% revenue churn. In your home market, with hundreds of accounts, those two numbers stay close together. In Japan they do not, and the regional view gives them the same two columns.

Non-renewal arrives at the anniversary, not month by month. Leaving a vendor in Japan goes through the same internal approval document that buying did. Research by IDEATECH with Hiroyasu Kitagawa, published 16 April 2026, surveying 307 people who took part in a B2B purchase of at least ¥3 million alongside at least one colleague, found 60.9% required two approval stages and more than 80% required two or more, with consideration running three to eight months in 63.6% of cases. Budgets move on the same annual cycle. The decision not to renew is therefore made months before the date on your contract, and lands as a single event.

A monthly churn chart for Japan will be flat, flat, flat, cliff. It is close to the wrong instrument.

To be clear about what is not known: there is no published Japanese primary dataset on why B2B contracts are cancelled. The approval research above covers buying, not leaving. Anyone quoting you a Japanese churn-reason breakdown is quoting a vendor’s customer list, not a market.

Your expansion may be one account. Seat growth inside a single Japanese group company clears one approval and can carry the whole region above 100% NRR. That is not a repeatable motion, but it looks identical to one on the dashboard.

What to put in the regional report

Four lines, replacing the single retention number.

1. Logo churn: accounts lost over accounts at period start
2. Gross revenue churn: revenue lost, including downgrades on contracts that renewed
3. NRR: the same base after expansion
4. Concentration: the share of Japan ARR held by the top three accounts

The fourth line is what makes the first three readable. At 60% concentration, a 100% NRR tells you about three customers. At 10%, it tells you about a market. Headquarters has no way to know which one it is looking at unless the line is there.

Two fields, and one rule

In the CRM, add the customer’s own renewal decision date, which is when their internal approval cycle needs an answer, not your contract end date, and the notice deadline in the contract. In Japan these are usually one to three months apart, and the second one is the date you actually lose.

Then make churn reason mandatory, from the same picklist your other regions use. If Japan gets its own free-text field, the pattern will never reach a board deck.

A country operation of a dozen accounts does not have a retention rate. It has a list of twelve renewal decisions, and each one has a name and a date.

Never read a single retention number for Japan. Logo churn, gross revenue churn and NRR must appear together with the concentration of the account base, or the number describes one customer rather than a market.

Related reading

The same mistake shows up when headquarters benchmarks Japan against a market average. Why Japan’s “3% SaaS churn” benchmark misleads your HQ covers reading the median rather than the mean when the distribution is skewed.

Other notes on the same problem are collected under Revenue data.