Bing Webmaster data

Bing Webmaster adds a second search-engine view to the evidence you already have. Its report compares traffic periods, highlights crawl changes, and finds query and page opportunities for a verified site. It does not replace a live crawl or prove that one URL is indexed.

Connect a verified site

Open Bing Webmaster Tools, go to Settings, then API Access, and generate an API key. Run the guided connection in a terminal:

seo providers bing connect

The prompt masks the key, validates it by listing the sites available to the account, and asks you to choose a verified site when the project hostname does not identify one on its own. The selected Bing site is saved on the local project profile.

The key stays on your machine. The system keychain is used when available. A private file with 0600 permissions is the fallback for headless machines and locked keychains.

Check or remove the connection later:

seo providers bing status --project example
seo providers bing sites
seo providers bing disconnect

Disconnecting removes the saved credential. It keeps the harmless site mapping on the project so a later reconnect can reuse it.

Microsoft’s Bing Webmaster API access guide explains where the API key is generated and how it applies to the verified sites owned by that account.

Run the report

Use the site attached during connection:

seo providers bing report --project example

Or pass a verified Bing site directly:

seo providers bing report --site https://example.com/ --json

The same core report is available to agents through the report catalog:

seo reports describe bing-webmaster-overview --json
seo reports run bing-webmaster-overview \
  --params '{"site":"https://example.com/"}' \
  --json

The report requests Bing’s daily search traffic and crawl statistics plus its weekly top query and page lists. It does not write these responses to the local cache. Every response has a hard byte limit. Daily sections keep at most 400 rows and weekly sections keep at most 8,000 validated source rows before reducing them to bounded insights. Structured output retains at most 14 recent daily rows per section, 10 items per insight list, and 40 query and page insight items across the report.

The useful parts come first:

• findings ranks traffic changes, crawl regressions, and query or page review opportunities with the evidence and a verification step.
• traffic.analysis compares the latest 28 days with the previous 28 days. It shows missing days rather than silently comparing incomplete periods.
• crawl.analysis compares the latest daily snapshot with the snapshot on or before the 28-day boundary.
• queries.analysis and pages.analysis aggregate four weekly top lists, compare movements only for entries observed in every week of both periods, and return a small review list.

The structured result also exposes returned rows, excluded invalid rows, caps, method names, observation time, and the credential source.

Read dataStatus, provenance, and section coverage before the findings. Bing’s query and page endpoints return weekly top lists, not complete exports, so those sections remain partial by design. An entry missing from one period is unknown and is never treated as zero. Invalid, capped, or failed sections also stay visible.

Review referring links

The connected site can also return a bounded sample of pages Bing says link to it:

seo links --project example --json

The command first requests target-page link counts, then requests referring URLs for a limited number of those targets. Provider pages, target pages, parallel requests, response bytes, retained rows, and returned rows all have hard limits. The result says when any of those limits made the evidence partial.

You can use a local export instead of Bing:

seo links --file ./links.csv --row-limit 10000 --json

CSV and JSONL files stream from disk. Regular JSON arrays are accepted only under a smaller file limit. Common source URL, target URL, anchor text, first-seen, last-seen, and nofollow column names are normalized. Invalid and duplicate rows stay visible in provenance.

This report does not download a web-scale backlink database. A missing row is not proof that no link exists, and a returned row is not an authority, quality, or ranking verdict. Open an important referring URL and confirm the link is still present before acting on it.

Keep the claims narrow

Bing’s search traffic includes data observed by Bing. It is not Google Search Console data and it does not describe the whole search market.

The query and page opportunity range is a prioritization heuristic. It helps choose what to review first, but it is not a Bing requirement or a forecast. Check query intent and the live page before changing content.

The crawl response can include an inIndex statistic. Treat that as a provider-reported site statistic for the observed date. It is not independent proof that a particular URL is indexed now. Run a live crawl when you need current status, redirect, robots, canonical, or page evidence.

Use an environment secret for agents and CI

Agents and unattended jobs do not prompt. Pass the key through the process environment:

SEO_BING_API_KEY=your-key \
  seo providers bing report --site https://example.com/ --json

Store the value in the platform’s secret manager. Do not put it in a script, repository, report parameter, or command-line flag. The structured provenance shows that the environment supplied the credential, but never includes the key.