Back to home

StackPeep Data Sources and Detection Methodology

How StackPeep uses public website signals, subprocessor disclosures and curated records, and how to interpret scan dates, evidence labels and limitations.

Last updated: 2026-10-23

Where does StackPeep's data come from?

StackPeep combines publicly observable website signals with recorded public disclosures and curated company entries. These evidence types answer different questions. A detected script suggests a website technology; a subprocessor disclosure describes a provider relationship. Neither establishes a complete picture of the company's internal software. The Sources and scan date panel on a lookup result shows evidence types and public links for the company record.

What do the evidence labels mean?

Website signal

A recognizable signature was found in public website content or response headers. Signals can include framework markup, script URLs, analytics tags, hosting headers or provider names. They can also occur in product documentation, integration examples or unused code. A match is a research lead, not proof that the company currently uses that tool throughout its business.

Disclosed provider

A tool or provider was recorded from public subprocessor or trust material. Check the original source for the vendor's purpose, product coverage, region and effective date. A company disclosing a provider does not mean every customer uses it or that the provider powers the company's public website.

Curated record

An entry was included in a prepared company record. It should be checked against the linked sources and current disclosures. It is not evidence that the live scanner detected the tool during your lookup.

New scans and reviewed company pages attach a source URL, observation date and matching evidence to each supported finding. Website signatures and rows in public subprocessor tables are evaluated separately. Older entries without item-level evidence are marked instead of borrowing the company's source list as proof. Internal confidence values are heuristics, not audited accuracy percentages.

What happens during a lookup?

  1. The input is resolved to a public company domain. A small set of example company names has explicit domain mappings; ambiguous names can resolve incorrectly, so use the domain if you know it.
  2. If a stored record is less than seven days old, it can be returned without a new scan.
  3. Otherwise, the scanner reads the public website's HTML and response headers. It may follow up to two same-site links that appear to be subprocessor disclosures.
  4. Recognized signals are matched against a supported tool catalog. The scanner does not log into company accounts or inspect private systems, and it does not execute the website's JavaScript in a browser.
  5. The result and its scan date are stored for later lookups. A site that blocks access or exposes no recognized tools may produce no result.

The prepared example records can include public disclosures collected separately from the live scanner. Their coverage can therefore differ from a fresh scan of the same domain.

How fresh are the results?

Reviewed company pages are fixed snapshots with their own review and observation dates; they are not refreshed by a page visit. Interactive company records may be cached for up to seven days. A lookup can return that cached record, and some source material may be older than its scan date. The date identifies the record's scan timestamp; it is not a claim that every vendor relationship was checked on that date. When recency matters, open the original sources and confirm their current contents.

What can StackPeep miss or misidentify?

  • Tools used only inside the company or behind authentication.
  • Technology loaded only after browser interaction or client-side JavaScript execution.
  • Unrecognized, renamed, removed or deliberately concealed technologies.
  • Provider names appearing in examples, text or scripts without current production use.
  • Differences between a company's marketing website, its product and its internal infrastructure.
  • Changes made after a record was collected.

A missing match does not prove absence. A displayed match does not prove company-wide use. Vendor links describe the product, while company source links support further investigation. Use direct company confirmation for security, procurement and architecture decisions.

Do sponsors influence the data?

No. Sponsorship purchases an advertising placement for a defined period. It does not alter company records, detected tools or evidence labels. The homepage logo pile is an exploration interface; logos alone are not evidence of a company's use or endorsement.

How do I report a correction?

Submit a support ticket, which requires sign-in. Include the company domain, the tool or provider, the disputed finding and a current public source. Do not include private contracts, credentials or confidential company information. A correction request is reviewed through support; submitting it does not automatically rewrite the public record.

Continue with About StackPeep or run a company lookup.

Run a free company lookup · Browse reviewed company examples