Guides

What a 99.9% match rate hides.

A match rate is a fraction, and the denominator decides the answer. Count payments or turnover, leave out the unmatched, ignore providers without statements, or call an ID 'found' instead of reconciled, and the same day can look near perfect or far from it. Ask what the rate is a share of before you trust it.

A match rate is a fraction

A match rate is the share of payments, or of money, that a reconciliation confirmed. Every share has a numerator and a denominator. Figures such as 99.9% appear in reconciliation marketing. A figure like that is a claim about one fraction, built one way. It says nothing on its own about which payments were counted, which were left out, or what "matched" meant. The same merchant day can produce a near-perfect rate or a modest one, depending on those choices. None of the choices has to be dishonest. Each one is a reasonable simplification for some purpose. The trouble starts when a single headline figure is read as "everything is fine". This guide uses one anonymised merchant day, measured by Reconora, to show how far the answer moves. Every figure below comes from that one day and is not an industry benchmark.

Count or turnover

A match rate by count treats every payment as equal. A match rate by turnover weights each payment by its amount. The two can differ a great deal. Thousands of small payments can match while a handful of large ones do not. Take an illustrative synthetic day: 9,990 of 10,000 payments match by count, but the unmatched 10 carry a large part of the money. By count the day looks almost perfect. By turnover it does not. The reverse also happens: a tail of small, unmatched local-method payments can drag down a count-based rate while the money is mostly confirmed. Neither view is wrong. A finance team cares about money; a support team cares about players. Reconora reports outcomes as shares of turnover in the reconciliation scope, and keeps counts alongside. On one anonymised merchant day, 75.9% of turnover in scope was matched and confirmed by provider statements.

Leaving out the unmatched

A match rate improves sharply when hard cases leave the denominator. Some reports compute "matched out of all payments that could be compared". Payments with no counterpart then disappear from the fraction. On one anonymised merchant day, 10.1% of turnover in scope had nothing to reconcile against. A further 12.6% was awaiting a statement that did not yet cover its hour, often at the day edge. Remove both groups from the denominator and the same day's match rate moves close to the top of the scale. Not one additional payment has been checked. The only change is which payments are allowed to count. A useful report keeps these groups visible as outcomes of their own. Reconora returns four outcomes: matched, differs in money, differs in facts, and nothing to reconcile against. Awaiting is shown separately and is not a discrepancy.

Ignoring providers without statements

A reconciliation can only confirm what a statement covers. When a provider sends no statement, every payment through that provider is unverifiable. Some reports drop such providers from scope, so the rate is computed only over providers that did send statements. On one anonymised merchant day, 26 providers sent no statement at all. Others sent incomplete statements: 830 lines on the same day belonged to statements that held only part of the provider’s payments for the day. A rate computed without these providers can be accurate and still describe a smaller world than the one the merchant operates in. Statement coverage is the figure that tells you how large that smaller world is. Ask for it next to the match rate.

"Found" is not "reconciled"

A spreadsheet lookup answers one question: was the log ID found in the statement. A reconciliation answers another: did the amount and the outcome match. On one anonymised merchant day, a lookup found 75.5% of reconciliation lines. True reconciliation on amount and outcome gave 74.4%. The headline figures are close. The difference is in what sits underneath: 690 lines were found by ID but differed in amount or outcome. A "found" count books every one of them as reconciled. Some are reversed outcomes, where the player was charged but not credited. A lookup from the log also cannot see payments that exist only in the statement; on the same day there were 317 of them. The spreadsheet comparison walks through the same day line by line.

One day, four ways

A single merchant day, measured once, gives these figures. Each is correct for its own definition.

What the figure measuresOne anonymised merchant day
Log IDs found in a statement, by lookup75.5% of reconciliation lines
Lines truly reconciled on amount and outcome74.4% of reconciliation lines
Turnover matched and confirmed by statements75.9% of turnover in scope
Turnover with nothing to reconcile against10.1% of turnover in scope
Providers with no statement at all26
Lines found but wrong690

None of these is a headline to sell with. Together they describe the day. A report that shows only the first line, or only a rate computed after removing the hard cases, is not lying. It is answering a narrower question than the one a finance team usually asks. The methodology page explains how each figure was measured and why one day is not a distribution.

What a useful report shows next to the rate

A useful reconciliation report shows the match rate together with what surrounds it. That means the denominator, stated in words. It means the share of turnover in each outcome, so matched, differing and unverifiable money add up to the whole scope. It means a list of providers and currencies without statements, and the lines from incomplete statements. It means the payments still awaiting a final status, kept apart from real discrepancies. Finally, it means a way to open any single payment and see the log line, the callbacks and the statement row behind it. With those in place, the headline rate becomes one figure among several, and nobody has to take it on trust.

Questions to ask about any match rate

A short list is enough to turn a headline into a useful number.

  1. Is it a share of payments by count, or of turnover?
  2. Which payments are in the denominator, and which were left out?
  3. How are payments with no statement, or with an incomplete one, counted?
  4. Does "matched" mean the ID was found, or that amount and outcome agreed?
  5. How many payments sit in "nothing to reconcile against", and in "awaiting"?
  6. Can a single matched payment be traced back to its source rows?

The longer version, with good answers and red flags, is in 5 questions to ask any reconciliation vendor.

Try it on one day of your data

One day of your data, reconciled, and the report stays with you whether or not you continue.

By Reconora Research · Updated