Ames Systems

DATA RECONCILIATION · SCRIPTED IDENTITY

Two teams. Two numbers.
One reconciled truth.

Sales quotes one customer count, marketing another, and the dashboard a third. Every meeting starts with that argument. We end it the only honest way: input = clean + merged + exceptions, exactly ±0 — scripted, not "trust me."

Sound familiar?

Where the argument comes from.

Three dashboards

Nobody trusts the report.

Each team pulls "customers" differently: deduped or not, which lifecycle stage counts, whose export is fresher. Everyone is a little right, so the argument never ends.

  • We make counting rules explicit once — then enforce them by script
  • The identity check ships with the result: anyone can re-run it
Two systems after a merger

The acquisition doubled everything.

Two CRMs, two customer masters, two definitions of "active". Until someone reconciles them with a documented rule, every combined number is fiction.

  • Cross-source matching runs as supervised scoped work — rules you approve first
  • Unmatched records become explicit exceptions, never silently dropped
The manual spreadsheet

Someone reconciles it by hand. Monthly.

A named person spends a day before each board meeting making numbers agree. When they're on vacation, the numbers disagree again.

  • We turn that ritual into a repeatable procedure your team can re-run alone
  • Documented refresh steps — no tribal knowledge
How it works

An identity check, not an opinion.

# reconciliation identity (demo dataset, synthetic)
input(1821) = clean(1183) + merged(10) + exc(628)
TRUE

# independent recount (separate reviewer, raw input):
recount → 1183 / 10 / 628 → MATCH
verdict: PASS (hash-bound)

The count isn't asserted — it's proven. A separate review pass recounts from the raw input without seeing our working, and the verdict is hash-bound to the artifacts.

What we need from you

The exports in question plus what each team currently counts as "a customer". That definition dispute is the project.

What comes back

Reconciled dataset · counting-rules doc · exception register · scripted identity check ±0 · refresh procedure your team can re-run without us.

Honest limits

Single-source reconciliation is a standard cleanup pass ($300–800 per ≤10k batch). Cross-source joins are supervised scoped work — quoted individually after we see both schemas. We don't do strategic analytics consulting.

$0Free sample · ≤1,000 rows · inside 24h

Send both exports. Watch the argument end.

Two files under 1,000 rows each — free reconciliation inside 24 hours, with the identity check included. If it's useful after that: fixed price before start, pay on acceptance.

Reply within 24h · prefer email? [email protected]

Related

Where reconciliation starts.

Cleanup

Duplicates everywhere?

The cleanup pass removes them with a full change log first.

Migration

Moving systems?

Pre-migration packs keep old mess out of the new platform.

Objections

"Excel does this."

Until it silently doesn't. See where DIY stops being enough.