Skip to content
BENIBIX Learn

Data Reconciliation

Also known as Data Recon, Source-to-Target Reconciliation, Row Count Check

Checking that copied data still matches its source. You compare counts and totals.

Think of it like

It is like counting your change after you shop. The receipt and your purse must match.

Example

The shop database shows 1,204 orders last night, but the warehouse has 1,192. The check fails and the load is held back. Emma finds 12 orders stuck on a bad date.

Why it matters

A pipeline can finish green and still drop rows. Only a check against the source shows it.

Definition

Data reconciliation checks that two copies of the same data agree. You compare the source with the place the data was copied to. The simplest test compares row counts. A stronger one also compares totals, like the sum of all orders. If the numbers differ, data was lost, doubled or changed on the way. Teams run it after each load, before anyone reads the new data.

Example code

sql
SELECT COUNT(*), SUM(total) FROM orders
WHERE order_date = '2026-11-15';
-- source:    1204 | 52300
-- warehouse: 1192 | 51780
The same query runs on both sides. The counts and totals differ, so 12 orders went missing.
The right half is reconciliation. Two copies of last night's orders disagree by 12, so the load stops.

Common mistake

Comparing only row counts. The counts can match while the amounts are wrong.

Key takeaways

  1. 1
    Two sides. The source against the copy.
  2. 2
    Counts and sums. Rows first, then totals.
  3. 3
    Before anyone reads. Hold the data until it matches.