AI-search competitor analysis has two separate jobs. One examines the content and sources competitors publish; the other measures which brands and URLs appear, are cited, or are recommended across a defined set of AI prompts. Content Radar can support the first job with competitor publishing evidence. It does not perform the second.
Google's Search Central documentation on AI features says the usual SEO fundamentals still apply, so traditional SEO competitor analysis remains the foundation. AI answers add a different question: which sources get used, cited, or recommended, and why. Answering it takes measurement of the answers themselves, which needs its own method and its own record.
Separate content evidence from AI-visibility measurement
Most confusion in this work comes from mixing two kinds of evidence. Publishing evidence describes what a competitor put on the web and when it was found. Visibility evidence describes what an AI system said in response to a specific prompt, on a specific date, in a specific product. Neither substitutes for the other, so each question below names where its answer has to come from.
| Question | Required evidence | System of record | Content Radar contribution |
|---|---|---|---|
| Which brands and URLs are cited for our priority prompts? | Captured answers for a fixed prompt set, with engine, date, and location | An AI visibility tool or a dated manual sample | None. Content Radar does not capture AI answers or citations. |
| How often does each competitor appear across those prompts? | Repeated captures of the same prompt set over time | An AI visibility tool or a dated manual log | None |
| What has a cited competitor published, and when did it appear? | Competitor URLs with source and first-detection context | Competitor feeds, sitemaps, and reviewed Articles | Surfaces new URLs from supported competitor publishing sources |
| Does the cited page answer the task better than ours? | A side-by-side reading of both pages | Editorial review notes | Keeps reviewed competitor URLs together for that comparison |
| Does the topic also carry traditional search demand? | Keyword, SERP, and ranking data, plus Search Console for your own site | An SEO platform and Search Console | None |
| Did our content change affect how often we are cited? | Re-measurement on the same prompts and conditions | The same visibility tool or manual method | None |
Define the prompt and competitor set externally
Start with a written prompt set: the questions a buyer asks while learning about the problem, comparing approaches, and choosing a vendor. Twenty to forty prompts grouped by stage are enough for a first pass. Record the conditions that change answers, including the assistant or search product, the date, the location, and whether the session was signed in.
The competitor set for this analysis is often wider than your sales competitor list. Publishers, review sites, and documentation hubs can be cited more often than direct rivals. Content Radar has no prompt or answer object, so the prompt set and the visibility competitor list belong in the measurement tool or a spreadsheet.
Collect mentions and citations with an external tool or controlled manual sample
Some SEO platforms, including Semrush and SE Ranking, now list AI visibility tracking alongside their search datasets. Specialist trackers can run prompts on a schedule and keep history. A controlled manual sample also works for a small program when every run follows the same conditions and every observation is logged.
- Prompt text and the buying stage it represents.
- Assistant or search product, date, location, and session state.
- Brands mentioned in the answer and the order they appear.
- URLs cited or linked, copied exactly.
- Whether your brand or pages appear at all.
- A second run of the same prompt, because answers vary between sessions.
Trace cited competitor pages back to content patterns
Once the citation log exists, group cited URLs by competitor, page type, and topic. This is the point where publishing evidence becomes useful: review competitor publishing that may help explain an externally observed AI-visibility pattern. If you monitor supported competitor publishing sources, you can see which related pages were detected around the cited URL and whether it belongs to a cluster or stands alone.
Keep the inference modest. A detection date shows when a URL was found through a feed or sitemap, not when the page was last edited or why an AI system chose it. Content Radar does not watch arbitrary pages for in-place changes, so edits to an existing cited page need a manual check or a page-change tool.
Compare answer coverage, evidence, freshness, and source credibility
Two competitors may target the same phrase but answer different parts of the reader's task. Map the decision journey: definition, evaluation criteria, process, examples, risks, alternatives, implementation, and next steps. Then note where each cited page gives a complete answer and where it relies on vague summaries. These are reviewer judgments made by reading pages; no monitoring or visibility tool scores them for you.
| Dimension | What to compare on the page | Content response |
|---|---|---|
| Source credibility | Does the page identify authors, methods, evidence, and firsthand experience? | Add transparent sourcing and real operational detail |
| Freshness | Are facts, examples, and product details current, based on what the page itself shows? | Create an update cadence for time-sensitive pages |
| Answer completeness | Does the page resolve the full task or only define the term? | Cover steps, tradeoffs, examples, and limits |
| Topic depth | Is the answer supported by related pages with distinct jobs? | Build connected pages and strong internal links |
| Clarity | Can key claims be understood without promotional language? | Use direct headings, concise definitions, and specific examples |
- Core answer: Can the page state the answer clearly in a short, accurate passage?
- Supporting evidence: Does it include examples, data, methods, or practical experience that justify the answer?
- Decision support: Does it explain tradeoffs, fit, limits, and alternatives?
- Connected depth: Can readers move to related pages that handle narrower follow-up questions?
Choose a content action
- Refresh an existing page when it covers the task but lacks evidence, examples, or current detail.
- Create an original page when the cited competitor material answers a task you have not addressed.
- Strengthen supporting pages and internal links when a single page is carrying a topic alone.
- Monitor without acting when the citation gap is small, unstable between runs, or outside your buyers' questions.
Avoid chasing generated answers page by page. Interfaces and citation patterns change often, while credible, current, and useful source material keeps its value across them.
Worked example: from a citation gap to an original page
The numbers below are hypothetical and only illustrate the method.
- Measurement, outside Content Radar: a 30-prompt sample about SOC 2 evidence collection is run in two AI assistants on two dates. A competitor is cited in 11 answers; your site is cited in 2.
- Tracing: 9 of those 11 citations point to three competitor pages, a control-mapping guide, an evidence checklist, and a glossary entry.
- Publishing evidence, from Content Radar: the checklist and glossary URLs were first detected in the competitor's resources sitemap during the same six weeks as four related guides. That suggests a deliberate cluster. It does not explain the citations.
- Page review: the competitor checklist names the specific artifact an auditor requests for each control. Your page describes the process in general terms.
- Recommendation: publish an original evidence matrix built from your own implementation experience, link it from the existing guide, and leave the competitor's checklist structure alone.
- Re-measurement: rerun the same 30 prompts in the same assistants four and eight weeks after the page is indexed.
Re-measure outside Content Radar
Use the same prompts, products, and conditions as the baseline, and compare like with like. Treat small movements cautiously: a handful of prompts can swing between runs without any content change. When the sample is small, report the direction and the variance, and avoid attributing a change to one page.
Where Content Radar fits
Content Radar monitors supported competitor publishing sources, such as RSS and Atom feeds and public sitemaps, and organizes the URLs it finds for review. It does not natively measure AI citations, AI answer visibility, LLM rankings, AI share of voice, AI search positions, prompt performance, brand mentions, or recommendations. For the ongoing publishing side of this work, see competitive content monitoring for SEO teams.
A practical competitor review prompt What does this competitor explain more clearly, prove more convincingly, or update more consistently than we do? The answer produces better work than asking how to copy a page that appeared in an AI result.
Add publishing evidence to AI-search analysis
Use Content Radar to follow what competitors publish through supported sources, and keep AI citation and visibility measurement in the tools built for it.