SaaS Evidence Notebook

Editorial and review guides for teams explaining B2B software and its security.

SaaS Evidence Notebook

Editorial and review guides for teams explaining B2B software and its security.

Put Static Analysis Findings Into a SaaS Customer Evidence Packet

Start with a statement you can trace back to a release record. If a buyer asks how you review source code findings, show what was analyzed, when someone reviewed the results, and what happened to the items that mattered. Give the buyer a short summary with the limits visible on the same page. Keep the underlying records together so your team can reproduce the summary when the buyer asks a follow-up.

Write the claim the packet needs to support

Write the buyer’s question above the draft. “Did you review findings for the version we will receive?” calls for release evidence. “How do you assign and track findings?” calls for evidence of your review process. If the buyer asks both, answer them with separate statements and attach the supporting record to each.

NIST’s Secure Software Development Framework recommends recording and triaging issues found in code review or analysis. For a customer packet, ask the engineering reviewer for three things before opening the export:

  • The code in scope. Record the product, repositories or components, and revisions that correspond to the release. For a multi-repository service, ask for the revision of each included component.
  • The source record. Save the completed run reference, scan date, ruleset or configuration identifier if available, and any filters applied when exporting. Write down any skipped or failed components.
  • The review cutoff. Record when the team confirmed the dispositions and who can explain them. The scan date and the date of the last review may differ, so give both a label.

If an excluded repository implements part of the feature the buyer asked about, name the missing component in the summary. Narrow the claim to the code your team can identify and substantiate.

Reconcile the export before you publish a count

Imagine a SaaS team shipping release 4.8 of its document approval service. An October 6 export for the release candidate contains 12 rows from two included repositories. The team finds that three rows repeat findings already listed in the export. The reviewer confirms each match against the underlying code location and tracked issue and records the mapping. The reconciliation yields nine distinct findings to account for. A separate archive connector is outside the scan scope and outside those counts.

By the October 10 review cutoff, the engineer and reviewer have assigned all nine findings a disposition:

  • Four fixed and verified. The review record names the release build containing the changes and links to the check that verified each fix on that build.
  • Two fixes merged, verification pending. The changes have been merged, and the reviewer still needs results for the shipping build.
  • One confirmed and open. An engineering owner has a due date and a ticket that records the remaining work.
  • Two dismissed as false positives. Each has a dated reviewer decision and a recorded reason tied to the code reviewed.

The editor writes a customer summary that the reviewer can check directly: “We reviewed nine distinct findings from the October 6 scan of the two repositories in scope for release 4.8. As of October 10, four fixes were verified on the release build, two were awaiting verification, one finding remained open, and two were dismissed after review. The archive connector was outside scope.”

The team shares that scoped statement and assigns an owner to update it after release verification. The customer can see why the number in the packet differs from the export total. Internally, every row still leads to a finding record, including duplicates and dismissals.

Keep a review record behind every status

Give each distinct finding a durable internal ID. Record the source row IDs, deciding reviewer, disposition date, and evidence reference on the same entry. If the review status changes, append the new decision so you can rebuild an earlier packet using its original cutoff.

When using a source code security scanner for remediation review, check what its history actually records. Topscan’s SAST documentation describes a dated history of open findings and remediation progress. To support a claim for a shipping release, match the finding record to the revision that contains the fix and the evidence that the reviewer used to verify it.

Review the register for gaps before sharing a summary. A closed ticket with no linked release needs a follow-up before you count it as a verified fix. If two rows look similar but the reviewer has yet to confirm a match, leave the reconciliation open and hold the count. For a finding with a documented risk acceptance, retain the decision owner and review date and use a status that reflects that decision.

Prepare the customer copy and give it a review trigger

Put the version, dates, coverage, and disposition totals together in the customer copy. Attach a document ID and the contact who can arrange a deeper review. Where disclosure is limited, use stable references to the private evidence and give the buyer enough information to understand the impact and next step.

Before sending the packet:

  • Check that totals add up and that each label names the unit being counted.
  • Review the outgoing file for source excerpts, internal paths, credentials, customer identifiers, and sensitive details. Share deeper evidence through the agreed channel.
  • Set a review trigger for the next relevant release, a changed scan configuration, or a material status change. Preserve the old packet with its own date and scope.

When pull request scanning is configured, GitHub code scanning can annotate changed lines with new alerts. If you include one of those annotations as a process example, label it as evidence about that pull request. Use a separate scoped record to support a claim about a branch or release.

Frequently asked questions

What should we report when an in-scope run returns zero rows? Say that the identified run reported zero findings for the recorded revision and scope. First check that it completed, that the intended components were included, and that the export filters did not hide results. If a job failed, report that gap explicitly.

Should false positives be included in the customer count? Keep them in a reconciliation of all distinct reviewed items and label their disposition. If a buyer asks for confirmed issues only, give that subtotal and state the criterion. Retain the decisions privately so reviewers can reproduce either number.

Can we compare this packet with last quarter’s? Compare the same counting unit, scope, export filters, and review rules. If the ruleset or repository coverage changed, explain the change or make a like-for-like comparison for the common scope before describing a trend.

What if the buyer asks about a branch or repository outside the packet? Record the gap and ask the engineering owner for evidence covering the requested revision. Give the buyer a date for the follow-up and retain the original report as evidence for its documented scope.

Put Static Analysis Findings Into a SaaS Customer Evidence Packet
Scroll to top