Competitor intelligence
How to use it
- Add competitors to the registry. Type a domain directly, or promote one from discovery.
- Crawl them. Content signals come from a crawl of their public site.
- Read the scorecard for the summary, then the matrix for the detail.
- Work the keyword gap — it is usually the most directly actionable view.
- Set recrawl cadence so postures stay current without you remembering.
⏱ ~15 min setup · 💳 Starter+ (AI analysis Pro+) · 🎯 Know where you actually stand
The registry
The registry is the spine. Everything else reads it.
Competitors are added to it automatically whenever you start tracking one — adding a price-monitored site, starting a content crawl, or running a gap snapshot all register the competitor. You can also add one by hand from the registry panel.
On builds before August 2026 the registry was only populated by manual entry and a couple of specific flows, so a store that had been price-monitoring and crawling competitors for months could still show "No competitors tracked yet" — and, because the recrawl scheduler reads the registry, nothing was ever recrawled either. Opening the registry now backfills it from your existing tracking surfaces automatically.
Domains are normalised consistently. www.example.com and example.com are
the same competitor. Storing one form and looking the row up by another is how a
competitor you just crawled ends up reported as "no crawl yet" on the gap view,
and how a legacy www. row shows up as a duplicate column in the matrix.
The views
| View | What it answers |
|---|---|
| Scorecard | One blended read per competitor across price, content and keywords |
| Keyword gap | Terms they rank for or cover that you do not |
| Price monitor | Their prices against yours, per matched product |
| Content crawl | Their crawled pages, and the comparison against your site |
| Matrix | The ten-signal side-by-side |
| Backlink gap | Domains linking to them and not to you |
| Share of voice trend | Your visibility share over time against the set |
| Briefs | Content briefs generated from the gaps |
| Reprice | Price change proposals derived from the monitor |
| Executive report | A shareable summary of the whole picture |
The comparison matrix
Ten signals, computed from crawls of each site:
| Signal | Direction |
|---|---|
| Pages crawled | Higher is better |
| Title coverage | Higher is better |
| Meta description coverage | Higher is better |
| H1 coverage | Higher is better |
| Schema coverage | Higher is better |
| Product schema coverage | Higher is better |
| Open Graph image coverage | Higher is better |
| Average word count | Higher is better |
| Average internal links | Higher is better |
| Noindex rate | Lower is better |
The leader per signal is highlighted. Each competitor column shows when it was crawled and flags a stale one, because a comparison against a six-month-old snapshot presented without a date is worse than no comparison at all.
Crawling a competitor
A competitor crawl fetches their public site, extracts the same signals your own crawl does, and stores the comparison.
Internal links are scoped to the page's own host. A site that redirects apex to
www — or HTTP to HTTPS — meant every discovered link was classified as
external, the queue emptied after the first page, and the entire comparison
(coverage percentages, gaps, matrix, scorecard) was computed from a single
page and presented as fact.
Extraction now follows the final resolved URL, so a redirecting competitor crawls properly. If you have historic comparisons that look implausibly thin, recrawl.
Keyword gap
The keyword gap joins your Search Console data against a competitor's crawled content: terms where their content covers something you rank poorly for, or not at all.
Because it depends on both halves, it reports honestly when one is missing — "no SEO crawl of this domain yet" is a different message from "no gaps found".
Gaps trend over time, so you can see whether you are closing them.
Repricing proposals
The price monitor produces repricing proposals, which land in an approval queue rather than changing prices. Each proposal shows the real product title, the current price, the proposed price and the competitor evidence behind it.
Nothing changes a price without an explicit approval.
Live search results versus AI estimates
Several views need to know who ranks in Google's top 10 for a keyword: competitor discovery, share of voice, the SERP report, the SERP rollup, and the SERP panel in the keyword drawer. That data has two possible origins, and they aren't interchangeable:
| Origin | What it is | How it's labelled |
|---|---|---|
| Live Google results | The actual organic top 10 for the keyword in its market, fetched from DataForSEO. Positions, URLs and titles are measured | Live Google results from DataForSEO |
| AI estimate | A model's recollection of the results. Domains can be wrong and positions are guesses | Estimated by AI analysis, not a live search of Google |
Every number is calculated from live results only. Competitor discovery, share of voice, the SERP rollup and the Decision Intelligence market-research score skip estimated rows entirely. When no live results exist for a keyword, those views have no data for it. They don't fall back to an estimate.
The SERP report lists both kinds of analysis, with a Source column saying which each row is. Difficulty and opportunity scores are model estimates, so they appear only on estimated rows.
A SERP analysis tries live results first. A live SERP is reused for 24 hours, so repeated analyses of the same keyword don't pay for it twice. It uses an AI estimate only when live results aren't available (the provider isn't configured or is out of credit), and labels it wherever it's shown. In the keyword drawer, a live SERP links each result to its page and leaves out the word-count and on-page-signal columns, because a live SERP doesn't measure those.
Until September 2026 the competitor top-10 lists were produced by asking a model to recall the search results, not by a live search. Competitor discovery, share of voice and the Decision Intelligence market-research score were calculated from those estimates as if they were measured, and the image analysis tried to fetch pages the model had invented.
Every existing row is now marked as an AI estimate and excluded from those calculations. If competitor discovery or share of voice now look emptier than before, that's why. Re-analyse a keyword once live results are available to replace the estimate with measured data.
Competitor X-Ray
For a full profile of one competitor, including catalogue and price against yours, promotions, tech stack, search and AI-answer visibility, positioning and sentiment, run a Competitor X-Ray from Market Intelligence. It profiles the competitors in this registry and reads the price monitor, crawls and backlink research described on this page.
Alerts
Competitor movements raise alerts, and the Competitors navigation badge counts unread ones. You can mark them read in bulk from the executive report.
Plan tiers
| Capability | Free | Starter | Pro | Agency | Enterprise |
|---|---|---|---|---|---|
| Registry + manual add | — | ✓ | ✓ | ✓ | ✓ |
| Price monitoring | — | ✓ | ✓ | ✓ | ✓ |
| Competitor content crawl | — | ✓ | ✓ | ✓ | ✓ |
| Ten-signal matrix | — | ✓ | ✓ | ✓ | ✓ |
| Keyword gap | — | ✓ | ✓ | ✓ | ✓ |
| Scheduled recrawl cadence | — | ✓ | ✓ | ✓ | ✓ |
| Repricing proposals + approval queue | — | ✓ | ✓ | ✓ | ✓ |
| AI competitor analysis | — | — | ✓ | ✓ | ✓ |
| Content briefs from gaps | — | — | ✓ | ✓ | ✓ |
| Backlink gap | — | — | ✓ | ✓ | ✓ |
| Share-of-voice trend | — | — | ✓ | ✓ | ✓ |
| Executive report | — | — | ✓ | ✓ | ✓ |
AI-backed competitor analysis makes model calls per run and sits at Pro with the other AI SEO tools; basic price and content monitoring is Starter.
Best practices
- Pick three to five real competitors. A registry of twenty produces a matrix nobody reads and a recrawl bill you did not intend.
- Recrawl on a cadence, not on a hunch. Stale columns are the main way this view misleads.
- Compare like for like. A competitor with 40,000 pages will beat you on "pages crawled" and lose on average word count. The signals that matter are the coverage percentages.
- Start with the keyword gap. It converts directly into briefs and content; the matrix is diagnostic, not a to-do list.
- Treat repricing proposals as evidence, not instruction. The approval queue exists because margin is yours to decide.
See also
- Competitor discovery — finding who your competitors actually are
- Content gap — the content-side view of the same question
- Backlink intelligence — your own link profile
- Link-building outreach — turning backlink gaps into a pipeline
- Topical clusters — turning gaps into a content programme
- Competitor X-Ray — a cited 11-dimension profile of one competitor