Does git history matter? Why commit history is worth more than a code snapshot

Git history matters because commits, pull requests and review comments record how code changed and why, which a snapshot cannot show. For a data license, full history is the more useful asset. A fractional CTO should preserve it before any migration, then introduce qualifying companies to SourceX.

Does git history matter for a data license?

Yes. A code snapshot shows what the software is today; git history shows how it got there. Commits, pull requests, review comments and merge decisions record each change, who proposed it, who questioned it and why it was accepted or reversed. For AI developers training and evaluating coding agents, that sequence is closer to real engineering work than any single version of a repository.

A fractional chief technology officer (CTO) often sees the moment this history is at risk: a platform move, a monorepo consolidation, a squash-and-import, or an acquisition clean-up. Each can flatten years of commits into a single "initial import."

This page explains what history adds, what to preserve before a migration, and how to raise it with a company before introducing it to SourceX.

What does history hold that a snapshot cannot?

ArtifactWhat it showsLost when squashed or snapshotted
Commit sequenceIncremental steps, reverts and fixes in orderBecomes one commit
Commit messagesThe author's stated intentReplaced by a generic message
Pull requestsProposal, discussion and merge decisionStay in the old hosting platform
Review commentsReviewer objections and the author's responsesStay in the old platform
Issue linksWhich bug or request a change addressedReference numbers become dead links
Branch and tag historyRelease lines and hotfixesOften not migrated
CI resultsWhich changes passed or broke the buildRetained only if the old CI is archived

The key pairing is change plus reason plus outcome. A snapshot cannot supply any of the three. The same logic makes test suites linked to the changes they verify valuable, since a test is an objective check on a change.

When is history most at risk?

Raise the point when a technical event is already scheduled.

EventTypical riskWhat to ask the engineering lead
Hosting platform movePull requests and comments do not travel with the repositoryIs the old platform's review data being exported or just git objects?
Repository consolidation into a monorepoHistories are rewritten or truncatedWill the original histories be archived as is?
Squash on importYears collapse into one commitIs a full-history mirror kept somewhere?
Acquisition integrationAcquired code is re-created under the buyer's structureWhere does the acquired company's original repository live?
Vendor or agency handoverCode arrives as a zipDoes the handover include history?
Team departureLocal clones and personal forks vanishIs the main repository mirrored and protected?

If the history is already gone, say so; the company may still qualify through other systems. The question is what exists today and who can export it.

What to preserve before a migration

A short sequence works for most engineering teams:

  1. Take a full mirror clone of each repository, including all branches and tags, before any rewrite.
  2. Export pull request and review data from the hosting platform using its documented export or API, and store the output next to the mirror.
  3. Archive the issue tracker and its links to commits, since references are what connect the change to its reason.
  4. Archive CI logs or at least the pass-fail results for released versions.
  5. Record who owns each archive and where it lives, so an inventory can list it later.
  6. Keep the old platform account alive until the export is verified, not just requested.

None of this requires giving a copy to anyone outside the company. A company can list these archives as systems and years of history with the data inventory builder, which helps list systems without sharing code.

Common mistakes

MistakeWhy it hurtsFix
Treating a code freeze snapshot as the archiveNo reasons or outcomes surviveMirror full history and review data
Squashing to "clean up" before movingAuthorship and sequence are erasedMove history intact; squash only in new work
Exporting git objects but not pull requestsDiscussion and approvals vanishExport review data separately
Letting the old platform subscription lapseExport window closesExtend until verified
Ignoring secrets committed in pastCredentials sit in history and create riskRotate any exposed credentials and scrub where needed with counsel and security input
Assuming the company owns all the codeContractor or client code may belong to othersCheck assignment terms and customer contracts

What rights questions apply to code history?

Code history contains more than the company's own code. Check:

  • Contributor agreements: did employees and contractors assign their work to the company?
  • Client-owned work: agency and consulting repositories often belong to the client.
  • Open-source components: license terms attach to the code that was included.
  • Secrets and personal data embedded in commits or comments, which need removal or redaction.
  • Customer data accidentally committed in test fixtures.

A company that mainly writes code for clients without consent is a red flag. A product company with its own repositories, a long history and clear assignments is a strong candidate. The PII redaction matrix helps frame what needs removing from embedded data.

What to say to a CTO or founder

How the introduction works

You introduce; the company and SourceX run the rest. You never see or hold code.

  1. Register and send the founder or CTO your referral link, or use the referral form.
  2. The company applies and SourceX checks headcount, repository age, system breadth and rights.
  3. The company inventories repositories, review tools, trackers and CI, with the years each covers.
  4. One all-in price and the license terms are agreed before any buyer review.
  5. Once the company is deal-ready, buyers typically respond within about two weeks.
  6. After signature, secrets and personal data are redacted under the agreed rules and the data is delivered.
  7. The company is paid; your reward follows after SourceX receives its fee.

How rewards work for a fractional CTO

Partners earn 25% of the eligible platform fees SourceX actually collects from the referred company's licensing deals, capped at $100,000 per referred company. The reward is paid only after the buyer pays and SourceX receives its fee; an introduction, meeting or signed agreement alone does not trigger payment, and no reward is guaranteed. The reward is a share of SourceX's fee and does not reduce the company's price. Check your client agreements for any disclosure requirement before registering. More on the role is at referral opportunities for fractional CTOs.

When it is not worth raising

Skip it for a company under the size baseline, a team that writes only client-owned code, or one that has already licensed the same code history for AI training. Related record types like PSA ticket histories fit better when the client is a managed service provider rather than a software product firm.

Next step

Ask one client whether the full commit and review history will survive their next platform change. If it will and they might consider a license, register as a partner, or have the sponsor apply at sourcex.si/apply.

  1. Step 1Share your linkSend your personal link to a company you know.
  2. Step 2Company appliesThe company applies itself at /apply.
  3. Step 3Buyer selects and paysThe buyer selects and pays for the data and SourceX receives its fee.
  4. Step 4You get your rewardYour share of SourceX fees becomes payable.

Common questions

Is a squashed repository worthless for licensing?

No, but it is less useful. A squashed repository still shows the code at points in time, and the company may hold other records that qualify. Full history with pull requests and reviews is considerably richer, so a mirror clone taken before squashing is the safer choice.

Do pull requests live inside the git repository?

Not usually. Git stores commits and branches; pull request discussions and approvals are typically held by the hosting platform. Export them separately using the platform's documented tools, and keep that export alongside the mirror clone.

What about credentials committed years ago?

They are a security issue regardless of any license. Rotate exposed credentials and decide with security and counsel how to handle them. Redaction requirements are agreed with the company before any work begins, and nothing is delivered without an executed agreement.

Who owns code written by contractors?

It depends on the assignment terms in each contractor agreement. A company should confirm that contributors assigned their work before it relies on a licensing right. Code written for clients under agency contracts may belong to those clients and should be excluded unless they consent.

Does the CTO see the code during an introduction?

No. A fractional CTO passes along basic fit facts and a referral link, nothing else. The company lists repositories, tools and years of history in its inventory rather than contents, and works with SourceX on scope, price and delivery.

Free resources

By SourceX Partnerships Team · Published 2026-10-09 · Updated 2026-10-09

Know a US company with valuable proprietary data?

Become a referral partner from anywhere we support, get your link and introduce an owner or authorized decision-maker.

Refer a company →

I own a business

Explore licensing your company's data to AI developers worldwide. Start a short assessment; no uploads needed.

Start an assessment