Data Quality & Reconciliation
Numbers you can't trust are worse than no numbers.
Put this metric to work: explore FlowMRR's Stripe MRR dashboard to track revenue movements, retention and forecasts from your Stripe account, or see how FlowMRR compares as a Baremetrics alternative for Stripe.
In FlowMRR, data quality covers two kinds of check. Source checks look at your Stripe records: duplicate customer emails, customers without an email, one product split across several IDs, unused products, and the currencies in use, including any with no published exchange rate. Identity checks run after every recompute: the month's movements must bridge opening to closing MRR, headline MRR must equal the subscriptions behind it, and each billing-interval split must add up to its total. Reconciliation against an independent source, a recomputation from Stripe or your bank statements, completes them.
Worked example
Dashboard MRR is $48,200; independently normalised Stripe subscription MRR is $48,847 on the same date, currency, discount, and status basis. The $647 gap (about 1.3%) needs reconciliation. Raw invoice totals are not comparable to MRR because annual bills, taxes, and one-off charges differ by design.
Other definitions you will see
Many teams mean reconciliation only: tying dashboard MRR to an independent recomputation from Stripe, or to revenue in the general ledger, and reporting the gap as a percentage. That measures drift after the fact, while source checks catch the records that cause it.
Illustrative comparisons
Acceptable driftThese unsourced ranges are illustrative scenarios, not measured industry benchmarks, percentiles or recommended targets. Their qualitative labels are editorial interpretations. Compare your own cohorts using the same segment, period, currency and measurement basis.
Why it matters
Investor due diligence will tie out your MRR to Stripe to the dollar. Forecast accuracy depends on input accuracy. Customer-facing pricing changes ride on the assumption that you know what customers actually pay. Data quality is invisible until it's a crisis — bake it in early.
Common mistakes
- Calculating MRR from CRM data instead of payment data — they always diverge.
- Not reconciling MRR each month against a recomputation from the Stripe subscription list on the same basis.
- Ignoring proration adjustments and credit notes — small per-event, large in aggregate.
- Treating refunds as churn (they're a revenue adjustment, not a customer event).
Calculating it from Stripe data
Most data-quality problems in Stripe analytics come from the source records, not the arithmetic. The common ones: one buyer under several customer records (a re-signup, a second workspace, a customer created by both Checkout and your app); customers with no email, which makes those duplicates impossible to group; near-identical products and prices created over the years, which split one plan into several lines; and subscriptions in several currencies summed without conversion. FlowMRR's Data Health page lists these: customers sharing an email, customers without one, similar and unused products, the currencies in use and any currency with no published exchange rate, whose revenue is left out of converted totals, with links to fix each record in Stripe.
Then come the rules applied to invoices. Proration lines from mid-cycle plan changes, setup fees, credit notes and voided invoices all appear on invoices but are not recurring revenue, so a renewal that also carries a proration must not produce a one-month MRR spike. Usage-based and tiered prices have no fixed list amount, so their MRR has to come from what was actually invoiced. Discounts need a stated rule: deducting only forever coupons, as FlowMRR's headline does, gives a different figure from deducting every active coupon, and both are defensible once declared.
Reconcile like with like. To tie dashboard MRR to Stripe, recompute it from the subscription list on the same date, with the same statuses (active and past_due in; trialing, incomplete, incomplete_expired and paused out; unpaid out from its last paid period), the same discount rule and the same exchange rates. To tie cash, compare gross receipts with paid invoices by payment date. Gaps that remain are usually timing: a webhook that arrived after a snapshot, or a sync that had not finished.
Internal consistency is a second check that needs no outside source: a month's movements (new, expansion, reactivation, contraction, churn) must bridge opening MRR to closing MRR exactly, headline MRR must equal the sum of the subscriptions behind it, and a split by billing interval must add up to the total it splits. FlowMRR checks these identities after every recompute and alerts its team when one fails; its investor report goes further, comparing figures printed in several places, such as the product total against headline MRR, and is withheld rather than sent when a money gap exceeds 1% of its base.
Reading the result
Look at the direction of a gap, not only its size. Dashboard MRR above a fresh recomputation usually means something is still counted that should not be: a cancellation whose webhook was missed, or a past_due subscription that has since become unpaid. Below it usually means missing records or a currency without a rate. A gap that grows every month points to a rule difference; a gap that appears once points to an event.
Fix the source rather than the report. Grouping duplicate customers, adding missing emails and archiving unused prices in Stripe improves every tool that reads the account, including the next investor's diligence. Re-run the comparison after each fix, and keep a short written note of your rules (statuses, discounts, exchange rates): most disagreements between two correct systems are disagreements about rules.
Frequently asked questions
Why does my MRR differ between Stripe and my analytics tool?
Usually because the two apply different rules, not because one is wrong: which statuses count (trialing, past_due, unpaid), which discounts are deducted, how annual and usage-based prices are normalised, which exchange rates are used and how duplicate customers are grouped. Align the date and the rules before looking for missing data.
How do I find duplicate customers in Stripe?
Export your customers and group them by email, then by company name or domain for B2B accounts. Several customer IDs sharing one email usually mean a re-signup or a record created twice by different integrations. Customer records cannot simply be merged together with their billing history, so the practical fix is to keep one record for future billing and have your analytics group the others with it.
How often should I reconcile SaaS metrics with Stripe?
At least once per closed month, before figures go to a board or into a fundraising process, and after any migration, pricing change or new billing integration. A monthly check catches drift while it is still small and its cause is recent enough to trace.
What do investors check in SaaS revenue due diligence?
They typically rebuild MRR or ARR from a raw Stripe export and compare it with what you reported, examine how churn, discounts and past-due accounts are treated, and test customer concentration and cohort retention. Written definitions and a reconciliation you have already done make the process faster and avoid late surprises.