Free local evaluation · v1.3.1

Run your first
CRM export check.

Compare two CSV exports on your own computer. Inspect missing records, duplicate IDs and selected field differences, then open the evidence behind each finding.

Python 3.10+ · macOS or Windows launcher · No account or payment required

The evaluation has no expiry. CRM files are processed locally; this website has no upload facility.

Watch the checker in action

Same row count. Different data.

Two short fictional demonstrations: follow a changed value and understand the coverage.

Watch the video demos →

Choose the help you need

Try the workflow.
Bring a reviewer when needed.

The download lets you evaluate the comparison yourself. A paid pilot adds agreed scope, human review and a handover.

Free

Local evaluation

Run the sample, then assess whether suitable CSV exports and retained identifiers fit the checker.

  • Self-guided setup and interpretation.
  • Local reports and a repeatable comparison.
  • Source code and operating instructions in the package.
Download v1.3.1 →
£395 indicative pilot fee

Reviewed pilot

Agree the rules with Serhat, investigate the exceptions and receive evidence for your selected scope.

  • Two exports of one object or agreed subset.
  • Up to 5,000 records per export and five agreed fields.
  • Human review, handover and one scoped rerun.
Discuss a reviewed pilot →

Written quote confirms scope, availability and the final total, including any applicable tax.

Set up on your computer

Start with the fictional sample.

Check the sample before working with customer exports. It gives you known findings to recognise and a safe way to learn the report.

Compare with the website’s sample evidence →
  1. Download and extract

    Save the ZIP and extract the whole folder. Open the included README. Keep the package files together.

  2. Check Python and launch

    With Python 3.10 or later installed, open a terminal in the extracted folder and run python3 launch.py on macOS or Linux, or py -3 launch.py on Windows. Included macOS and Windows convenience launchers offer another route. Follow the README if your system asks for permission.

  3. Run the fictional example

    Use the included sample and inspect its report. Read the matching rules, compared-record count and exceptions together.

  4. Evaluate your agreed subset

    Use only exports you are authorised to process. Confirm retained unique IDs, selected fields and expected transformations first. Keep both inputs and generated reports in an appropriate local location.

Read the coverage
before the conclusion.

The useful question is what was compared under which rules. A zero-findings report cannot establish that an entire migration succeeded.

01

Resolve ambiguous IDs

Duplicate retained identifiers prevent a reliable one-to-one comparison. Investigate them and rerun; do not treat skipped records as passes.

02

Explain expected changes

Agree transformations explicitly. Exact text and decimal equality answer different questions; a changed value may be legitimate only under the intended rule.

03

Keep wider checks visible

Supplied exports cannot establish live permissions, attachment access, workflow behaviour or whether the extraction included every required record.

Use the reconciliation acceptance guide →

Release information

Know what you downloaded.

Version 1.3.1 · local evaluation package prepared 16 October 2026. Check the file details and tested environment before running it.

New in 1.3.1: changing a record ID refreshes comparison-field choices and asks you to review conflicting rules again.

Package size
42,978 bytes (42.0 KiB)
Requires
Python 3.10 or later; Python is not bundled.
Verified environment
macOS 27.2 ARM · Python 3.12.14
Checks
26 automated checks passed: 23 release checks and three ID-selection regressions. Browser regression also verified.

SHA-256
27fc1e44f0ddc3191cc7c97c89b394042531d011b99c889ae6666a8188f77ea7

Windows launcher included; native Windows and Linux execution have not been tested. Python 3.10 and 3.11 have not been tested separately. The metadata records the full verification coverage.

Self-guided evaluation has a defined role.It does not include human review or certify a migration. Each export supports up to 5,000 rows, 10 MiB, 128 columns and 250,000 cells, with 1–5 selected comparison fields. All limits apply together; the README records the remaining cell, findings and report-size limits. Use an agreed process for customer work and keep report files under the same access controls as the source exports.

Before you download

Do I need to share an email address or pay?

The evaluation download is ungated and has no expiry. You can contact Serhat if you want to discuss the tool, an agency collaboration or a reviewed pilot.

Does it connect to my CRM?

It works from CSV files. It does not log in to a CRM, move records or repair data. Retained source identifiers and supported comparison rules are essential to a useful check.

Can I use it on a work computer?

You need Python 3.10 or later and permission to run the package. Follow your organisation’s software and data-handling rules. The package includes launchers and source; it is not a signed native app or a browser-store extension.

Where do my files go?

The evaluation processes input files on your computer and writes local reports. The website does not accept CSV uploads. You control any later sharing of inputs or reports; reports may contain values from your exports.

Can an agency use this before a handover?

You can evaluate the workflow on authorised exports. Your agency retains responsibility for the scope, rules, interpretation, client permissions and acceptance. Discuss a scoped second review or overflow check if human review would help.