CRM & Notion Data Audit — Diff Your Records Against a Dataset

Nobody trusts a CRM they haven't checked. This Actor checks it.

Give it a dataset you treat as the source of truth and a connected app to compare against, and it reports three things:

• Missing — source rows that never made it into the app.
• Stale — records in the app with no matching row in the source any more.
• Drifted — records on both sides where a field no longer agrees.

It never writes. Point it at a live production CRM without a second thought.

Why run it

Sync jobs fail quietly. Someone edits a record by hand. An import half-finishes. Six months later nobody can say whether the CRM matches reality.

This turns that into a list you can act on — and pairs naturally with the write-side Actors: audit first, see the damage, then fix it with a sync or a field fill.

Signal, not noise

The comparison is deliberately forgiving about things that don't matter:

• Blank on both sides counts as agreement — an empty cell and a missing field are not a discrepancy.
• Scalars compare trimmed and case-folded, so " Acme " and acme don't generate a finding.
• 0 and false are treated as real values, not blanks.
• Identical records are omitted from the report by default.

How it works

1. Connect your app under Integrations in Apify Console.
2. Run in list-tools mode to learn the connector's tool names.
3. Run in audit mode.

{
  "destination": "<your connector>",
  "mode": "audit",
  "datasetId": "<source-of-truth dataset>",
  "readTool": "query_database",
  "readToolArgs": { "database_id": "abc123" },
  "cursorArgName": "start_cursor",
  "cursorPath": "next_cursor",
  "matchKey": "email",
  "compareFields": ["phone", "owner", "status"]
}

Set up pagination if your app pages. An audit against a partial read will wrongly report records as missing. cursorArgName + cursorPath make the read complete.

Input

Field	Required	Description
destination	yes	MCP connector to audit
mode		audit (default) or list-tools
datasetId	for audit	Dataset treated as source of truth
readTool / readToolArgs	for audit	Tool that lists the app's records
cursorArgName / cursorPath / maxPages		Pagination for a complete read
matchKey	for audit	Field joining both sides
destMatchKey		Only if the app names that field differently
compareFields		Fields to check for drift. Empty = presence only
fieldMap		Only if a compared field is named differently in the app
reportMissingAtDestination		Default on
reportMissingInSource		Default on
reportChanged		Default on
includeIdentical		Default off

Output

One row per finding:

{ "status": "changed", "key": "a@example.com",
  "differences": [{ "field": "phone", "source": "480-555-0100", "destination": "480-555-9999" }] }

{ "status": "missing_at_destination", "key": "b@example.com", "differences": [] }

Full counts — including how many records were identical — are saved to the SUMMARY key-value record, so you can track drift over time without paying per row.

Pricing

$5 per 1,000 findings reported. Records that agree are free, and an audit that finds nothing wrong costs nothing.

FAQ

Does it modify my CRM? No. This Actor only reads.

What if my app pages its results? Set cursorArgName and cursorPath, or the audit will report a false "missing" for everything past page one.

What if a record has no join key? It's counted as keyless in the summary and skipped, rather than reported as missing.

Can an AI agent call this? Yes — it's exposed through the Apify MCP server, and being read-only it's safe for an agent to call unattended.