DataMosaic#Why the usual lists don't help retail IT
Search for the best data reconciliation tools and most of the pages that come back are written by finance software vendors. They compare close platforms: bank matching, balance sheet sign-off, audit trails. That is a real job, but it belongs to the finance team.
The person who looks after the systems of a chain with 10 to 150 stores is asked something else. The tills say one number and the ERP says another. The web shop's stock doesn't match the warehouse. Last night's price file reached 117 stores out of 120. That is one system against another, and the lists mostly skip it.
So this comparison sorts the tools by the question each one answers, and then by the retail IT job that asks it.
#Five kinds of tool, five different questions
| Kind of tool | Examples | The question it answers | Who usually owns it |
|---|---|---|---|
| Finance reconciliation platform | Solvexia, ReconArt, BlackLine | Do the transactions match the bank and the ledger, and is the close signed off? | Finance |
| Table compare tool | Redgate SQL Data Compare, dbForge Data Compare | Which rows differ between these two SQL Server databases, and what script makes them the same? | DBA, developer |
| Pipeline diff and data tests | Datafold, GX Core | Did this load or this code change alter the data, and does the data pass its tests? | Data engineering |
| Spreadsheet | Excel with Power Query | Which lines in export A are not in export B? | Whoever exported them |
| System-to-system check | DataMosaic | Do these two live systems agree, store by store, and if not, why? | Retail IT |
#Finance reconciliation platforms
Solvexia says it helps "Finance teams reduce the time and effort needed to combine, reconcile, map and analyze data", with complex matching for high volumes of transactions. Its own site describes a retail customer reconciling payment gateways, the point of sale system and the bank.
ReconArt calls itself a reconciliation and financial close platform, and lists "POS reconciliation, marketplace settlements, and multi-currency transaction matching" for retail and e-commerce.
BlackLine sells account reconciliation and transaction matching software as part of its financial close platform.
What they answer: whether the money agrees. Tender against bank deposit, card settlement against sales, subledger against general ledger, with approvals and a record of who signed what.
Where they stop for retail IT: they reconcile transactions for the close. They are not built to tell you that a store's day never reached the ERP's interface, or that one store missed a price batch. None of the three publishes a price on its site; each asks you to book a demo.
#Table compare tools
Redgate SQL Data Compare describes itself as a tool "for comparing and deploying SQL Server database contents". It compares live databases, backups and script folders, on SQL Server from 2008 to 2025, Azure SQL and SQL Server on Amazon RDS, and can run deployments from the command line. Redgate lists it at $327 per user for a one-year subscription (November 2026).
dbForge Data Compare for SQL Server, from Devart, finds "differences in the data of two databases", including tables, views, backups and the results of a custom query. Comparisons can be saved and scheduled from the command line with Windows Task Scheduler. Devart lists it from $239.95 a year (November 2026).
What they answer: which rows differ between two copies of the same SQL Server database, and the script that would make one match the other. That is exactly right for test against production, or before and after a migration.
Where they stop for retail IT: both sides must be SQL Server, with the same tables. Tills on Oracle against an ERP on SQL Server, or a Shopify store against anything, is outside what they compare. And a row that differs is the end of the answer, not the reason.
#Pipeline diffs and data tests
Datafold says its Data Diff compares "datasets within or across databases with value-level precision", and its documentation gives PostgreSQL against Snowflake, or two SQL Server instances, as examples. Rows are matched on a primary key you name, which need not be declared in the database.
GX Core, from Great Expectations, is an open source Python library under the Apache 2.0 licence. It tests one dataset against Expectations, which its README calls "unit tests for your data": not null, in range, unique.
What they answer: did this load or this code change alter the data, and does the data still pass its tests. They live in the data team's pipeline, next to the warehouse.
Where they stop for retail IT: they assume a warehouse and an engineer to run them. GX Core checks one dataset at a time, so two datasets that are each valid can still disagree with each other and pass. For more on that difference, see data reconciliation: techniques and a worked retail example.
#Excel and Power Query
The tool most chains really use is two exports and a spreadsheet. Power Query makes it repeatable:
- Load both exports into Power Query, one query each.
- On the Home tab, open the arrow next to Merge Queries and choose Merge Queries as New.
- Pick the key columns on both sides, store and business date.
- Under Join Kind, choose Left anti join for rows only in the first file, Right anti join for rows only in the second, or Full outer to keep everything and compare the values.
Microsoft documents this for Excel for Microsoft 365, Excel 2024 and Excel 2021.
What it answers: which lines differ between two files, today, for whoever has time. Step by step: how to compare two Excel files.
Where it stops: someone has to export both sides every morning, and nothing learned yesterday is kept. When that person is on holiday, nobody checks.
#DataMosaic: why two systems disagree
DataMosaic asks a different question from all four: not which rows differ but why these two systems disagree, every night, for every store.
- A check reads two systems, lines them up store by store on a key such as store and business date, and compares the measures you choose: receipts, sales, tax. It reads POS, ERP, e-commerce, WMS and CRM systems, databases including SQL Server, Oracle and PostgreSQL, REST APIs, and CSV and Excel exports. The two sides need not be the same kind of system.
- Rules set the thresholds, so 0.04 of rounding passes and 480 does not.
- A check runs on a schedule, overnight, so the answer is waiting at nine.
- On a row that differs, an investigation follows it back: was the store mapped, did the ERP refuse lines, do they share a cause, until there is a reason.
- A workflow is a fix you build, such as sending a price batch again, running the check again and emailing the store manager only if it still differs. A check only reads. Mosaic acts on a system only through a workflow an administrator builds, step by step, and no AI decides what changes.
It runs on your own server, Windows, macOS or Linux, or on Mosaic Cloud. Every feature is free while it is in beta, and sources and stores are never counted on any plan (November 2026).
Where it stops: Mosaic is not a close platform. It doesn't replace the finance team's sign-off, and it doesn't write deployment scripts between two databases.
#Which retail IT job needs which
| The job | The tool for it |
|---|---|
| Month-end close: tenders against bank deposits, sign-off and audit trail | A finance reconciliation platform |
| Refresh a test database from production, or check one after a restore | A table compare tool |
| Prove a warehouse load or a dbt change didn't alter the numbers | A pipeline diff and data tests |
| A one-off question about two files | Excel with Power Query |
| Tills against ERP sales, every store, every night | DataMosaic |
| Web shop stock against the warehouse or the ERP | DataMosaic |
| Did every store get last night's price file? | DataMosaic |
| After a POS or ERP migration, the first month of nightly comparisons | A table compare tool on cutover day, DataMosaic after |
Many chains need two of these. Finance keeps its close platform; the DBA keeps a table compare tool. The gap is usually the morning question between two live systems, which today is answered by an export, a VLOOKUP and a phone call to the store.
#One Monday, four tools
An invented chain with 120 stores. Monday's sales, at the tills and in the ERP, for four of them:
| Store | Tills | ERP | Difference | What a table compare tool shows | What Power Query shows | What Mosaic shows |
|---|---|---|---|---|---|---|
| ST-014 | 12,400.00 | 11,920.00 | 480.00 | Can't compare: tills on Oracle, ERP on SQL Server | A row that differs by 480.00 | Differs by 480.00. Why: lines refused for no tax code |
| ST-022 | 6,410.75 | none | 6,410.75 | Can't compare | A row only in the tills file | Missing in the ERP. Why: the store's day never arrived |
| ST-031 | 8,120.00 | 8,119.96 | 0.04 | Can't compare | A row that differs by 0.04 | Within the rule: rounding |
| ST-037 | 9,875.00 | 9,875.00 | 0.00 | Can't compare | Matches | Matches |
Invented figures. The table compare tool can't start, because the two sides are on different engines. Power Query finds the same three rows Mosaic does, but only once somebody has exported both sides, and it treats four cents of rounding like 480. Only an investigation turns a difference into a reason someone can act on. A finance platform would see this Monday later, when the deposits fail to match the sales. The six usual reasons are in the POS vs ERP post.
#How to choose
Ask four things before any demo:
- Which two systems? Two copies of one SQL Server database point to a table compare tool. Two different systems, on different engines or behind an API, need a tool that reads both.
- Who is asked? If it is the controller at month end, it is a finance job. If it is the IT manager before nine, it is a systems job.
- How often? Once, after a migration, a one-off comparison will do. Every night, for every store, it needs a schedule and a history.
- What happens after the difference? If the answer is still a phone call to the store, the tool has stopped one step short.
#Questions people ask
#What is the best data reconciliation tool?
It depends on the question. For the financial close, a finance reconciliation platform; for two copies of a SQL Server database, a table compare tool; for two live systems such as tills and ERP across many stores, a tool that compares them every night and explains the differences.
#Can Excel be used for data reconciliation?
Yes. Power Query's merge, with a left anti, right anti or full outer join, finds the rows that are missing or differ between two exports. It needs someone to export both sides each time, and it keeps nothing from one day to the next.
#What is the difference between data reconciliation and account reconciliation?
Account reconciliation proves a ledger balance against evidence such as a bank statement, for the close. Data reconciliation checks that two systems holding the same facts agree, such as sales at the tills and in the ERP.
#Is there a free data reconciliation tool?
GX Core is open source, though it tests one dataset rather than comparing two. Power Query comes with Excel. DataMosaic is free while in beta, as of November 2026.
