How do I run an SEO audit specifically for AI-powered search engines? Our client checklist
The audit we run for AI-powered search engines: bot access, indexing, site health, content gaps, citations and attribution, with the Promptwatch features behind each.
Clients sometimes ask whether an AI search audit is just an SEO audit with a new name. Partly. Google's guidance on AI features in Search points back to the usual foundations: crawlable, indexable pages with helpful content. So we start with a standard technical audit. Then we add the layer that is specific to AI engines. This is the order we use, with the Promptwatch features we rely on.
1. Bot access, checked against logs
We list the bots and test each against robots.txt, CDN rules and firewall settings: GPTBot, OAI-SearchBot and ChatGPT-User; ClaudeBot, Claude-User and Claude-SearchBot; PerplexityBot and Perplexity-User; Google-Agent and GoogleOther; the Meta fetchers; MistralAI-User. OpenAI documents OAI-SearchBot as its search crawler, so blocking it is a decision with consequences. Training bots are a separate choice, and the client owns it.
A configuration review is not proof. We have found sites where a rule looked fine but a CDN bot-protection layer was still challenging the crawlers. So we check logs. Promptwatch Agent Analytics gives real-time, IP-verified crawler logs with spoofed user agents removed, error tracking with failure explanations and a crawl-to-citation path with a citation rate per page. Crawler logs begin on the Professional plan, not Essential. Integrations cover Cloudflare, CloudFront, Fastly, Vercel, Netlify, Akamai, Google Cloud CDN, custom HTTP and a WordPress plugin.
2. Indexing and eligibility
200 responses, no stray noindex, main text present in the HTML, clean titles and headings, sitemaps current. We read the coverage reports in Google Search Console and Bing Webmaster Tools. Bing has been previewing AI-related reporting in Webmaster Tools, which we check for each client. Promptwatch's Search Insights overlays Search Console and Bing data on AI crawls and citations, so we can see a page that is indexed but never fetched by AI bots, or the reverse.
3. Site health
The weekly Site Health audit in Promptwatch runs 21 checks, produces a health score, shows what changed since last time and sends a weekly email. Site Content and Site Linking reports find thin or orphaned pages. We care about this for AI too: a broken internal link structure makes it harder for any crawler to reach deep pages.
4. Content against real prompts
We build a prompt list from the client's sales and support questions, tag it by topic and intent, and test whether each question has a page that answers it in the opening lines. Content gap analysis scores coverage per prompt. Topic Insights shows pages that underdeliver for their topic. Where pages exist but are weak, Content Refresh suggests diff-style edits.
5. Citations
We run the prompts across engines and record who is cited, from which domains and in what position. Promptwatch's citation analytics gives page and domain views, URL positions, Page Tracker for specific URLs, offsite mentions and Reddit, YouTube and LinkedIn reports. Citation Trends classifies each citation by content type and source type, with first cited, ramp-up days, peak rate and decay half-life per page. We pay attention to the source types competitors win from, because that tells us whether the fix is on-site or off-site.
6. Traffic and conversions
AI referrals in analytics, and conversions where possible. Visitor analytics in Promptwatch counts sessions and conversions from ChatGPT, Perplexity, Gemini, Claude, Copilot and others, via a script, GTM template or WordPress plugin. 360 Insights joins crawler logs, citation clicks and AI-referred visits per page, which is the closest we get to a single view.
7. The report and the plan
We rank fixes by revenue and effort. Blocked bots first. Missing answers on money pages second. Off-site gaps third. The Action Board in Promptwatch drafts a prioritized list, with a Discovery agent suggesting up to five evidence-backed actions per weekly run. We edit that list with the client.
What we do not claim
We do not promise that fixing an audit finding leads to a citation. Engines vary, answers change between runs and some topics are crowded. The audit removes what you control and gives you a baseline.
Where to start
Run the free checks first: the Agentic Readiness Check, the AI Bots directory and the no-login AI Visibility Report at Promptwatch. Then decide whether you need logs and daily prompt tracking, which means at least the Professional plan. Repeat bot access and site health monthly, and prompt checks weekly on your top pages.