Databricks Genie

August 2026 customer retention

Retention fell to 84%, down 4 percentage points from July. August lost 176 customers. New customers had the sharpest cohort decline; small businesses accounted for most losses.

Retention means month-start paid customers who remain paid at month-end, divided by the month-start population. In-month signups are excluded. Data: July–August 2026, analytics.customer_month_status.

Where retention fell

“New customers” joined in the prior calendar month: July for August’s cohort. “Established” customers joined earlier.

Where customers were lost

Small businesses represented 68.2% of August losses: 120 of 176. Enterprise had the largest segment decline, at 9 percentage points. Cohort and segment tables describe the same customers; their losses should not be added together.

Three recommended actions

1. Test guided setup for new customers. Among August’s new-customer cohort, 85 of 100 customers who completed the key workflow by August 1 were retained, versus 58 of 120 who had not: 85% versus 48.3%. Run a randomized test of guided setup and compare first-full-month retention. This association does not prove that setup caused better retention.
2. Review the 120 small-business losses before choosing a retention offer. Group cancellation reasons and usage patterns to distinguish onboarding problems, product gaps, and price concerns. This is the largest pool of losses; the current data does not identify their causes.
3. Audit the 32 enterprise losses. Losses rose from 14 in July to 32 in August despite an unchanged starting population of 200. Review renewal outcomes and account notes before deciding which intervention fits.

Customer growth masks the decline: August ended with 1,120 paid customers, comprising 924 retained customers plus 196 new customers still active at month-end, up from 1,100 in July.

I can turn the guided-setup recommendation into a test plan with eligibility criteria, a control group, and a first-full-month retention measure.