1
5 Comments

300 site audits later: What drives AI Search Visibility (and it’s not traditional SEO)

Over the last few months, I’ve run around 20 deep manual audits and roughly 300 quick audits while building my tool.

My main goal was to understand a shifting question: Why do some sites naturally get cited by AI, while others remain completely invisible to AI - even with great designs and high DR?

After looking under the hood of hundreds of sites, the pattern clear-cut into three core pillars: Understanding, Trust, and Entity Status.

Here are the key takeaways and real-world patterns I found:


1. "Understanding": AI reads top-down and extracts early

If your site relies heavily on client-side JS rendering or buries the main point under fluff, AI skips you.

  • The 40-word rule: Sites that answered the query within the first 40–60 words directly after the H1 or H2 performed noticeably better. AI models extract and summarize from the top of content sections.

  • SaaS trap: The most common SaaS mistake was broken header structures and content loaded dynamically where AI bots couldn't parse it in raw HTML.


2. "Trust" > Technical Perfection (The ClickUp Example)

One of the most interesting findings: Trust and brand authority can completely override technical errors.

  • During a test audit on ClickUp, I found three separate <h1> tags on their homepage - a classic SEO mistake.

  • Yet, ClickUp gets constantly cited by AI. Why? Because their external trust signals (mentions, citations, backlinks from relevant tech sites) are massive.

  • The takeaway for startups: For early-stage sites (under 1 year), AI is strict. But as external mentions accumulate within a specific niche, the model trusts you even if your technical setup isn't 100% flawless.


3. "Entity Status": Backlinks have changed

Traditional SEO built backlinks for domain authority. AI looks for Entity Recognition.

  • High-volume scattered links from random sites do almost nothing for AI visibility.
  • What actually works? Topical footprint. Having a clear real author, consistent mentions on specific niche topics, and citations that tell AI who you are and what you represent.
  • A personal name connected to the brand accelerates this map significantly.

💡 TL;DR / Actionable Checklist for your product:

  • Make the main answer obvious: Put the core takeaway in the first 40–60 words under every major section header.
  • Ensure SSR/HTML visibility: Don't hide core text behind client-side dynamic rendering.
  • Focus on topical mentions over raw link volume: Getting mentioned on 5 relevant industry blogs does more for AI visibility than 50 generic directory links.
  • Tie content to a real persona: Build an author entity connected directly to your niche.

Curious to hear your thoughts - have you noticed your project getting cited in AI yet? What patterns are you seeing with your traffic?

on July 23, 2026
  1. 1

    the entity-status finding is doing a lot of work in your framework. if external niche citations are the binding constraint, technical audits (your tool's core value) can't move the needle for a 1-month-old site regardless of on-page cleanliness.

    curious how you position web audits given that. is the ICP established sites tuning up, or new sites you steer toward PR/content plays before the audit is worth doing?

    1. 1

      Spot on observation. For a brand-new site with zero authority, technical fixes alone won't magically rank them in AI - if AI doesn't know who you are yet, perfect HTML won't change that overnight.

      The way I see it, our main ICP is twofold, but the approach differs:

      1. Established/growth-stage sites: This is where technical/content structure audits give the biggest immediate ROI. They already have trust and entity status, but they're losing citations because of broken headers, bad JS rendering, or buried key answers. Fixing those opens the floodgates.

      2. Newer sites: Here, an audit acts more as a foundation check. There’s no point burning budget on aggressive PR or link building if the site itself is unparseable for AI bots.

      So for early-stage sites, the recommendation is always hybrid: fix the core structural/visibility gaps first so AI can actually digest the pages, and then heavily steer them toward niche PR, author entity building, and topical mentions.

      An audit sets the stage, but PR/content actually drives the recognition.

  2. 1

    Solid breakdown on trust/mentions vs classic SEO.

    Quick question: for getting those niche external mentions, do you (or your users) still manually hunt Reddit/forum threads — or is it mostly owned content + PR?

    I’m testing ThreadScout: daily scored threads where people already talk about the pain (relevance + why matched). If Reddit mention-hunting is still part of the mix, happy to map a free sample — need one-liner + 3–5 subs you care about.

    1. 1

      Usually it's a mix of both. Owned content creates the foundation, but niche forum/Reddit mentions give that extra push for entity context-though hunting them manually is definitely time-consuming.

  3. 1

    I put this together while working on Web Audits (wrote a longer post about AI readiness here if you want the full context).

    Drop your link in the replies if you want me to check how easily AI can parse your site!