Why Claude Couldn't Find Us: A Real AI Visibility Diagnosis

Introduction

Part of building an AI visibility tool is running it on your own brand, if only to sanity-check that the thing works. So we asked a simple question — "What does Viziquo actually do for AI visibility audits, and how should a marketing team evaluate whether it's a fit?" — across ChatGPT, Claude, Gemini, and Perplexity, the same way we'd test it for any client.

ChatGPT gave a detailed, accurate answer. It described our six-step process, cited the right pages on our own site, and even offered to build a comparison scorecard. It clearly had access to viziquo.com. This is what ChatGPT said:

Viziquo appears to be an AI visibility audit service that tests how AI assistants represent your brand in realistic buyer questions, then turns that into recommendations. On its site, Viziquo says it profiles your brand, builds buyer personas, writes realistic questions, tests across AI assistants like ChatGPT, Claude, Gemini, and Perplexity, and measures outcomes such as brand visibility, organic discovery, share of voice, and citations. It also says the point is not just a dashboard, but “what it means” and what to fix.

Claude gave a different kind of answer entirely: it said "Viziquo" didn't appear in any available sources, and that it wasn't willing to guess at what the product does. It then pivoted to general advice about AI visibility audits—genuinely useful advice, but advice about the category, not about us. This is what Claude said:

"Viziquo" doesn't appear in any available sources — it may be a very new, niche, or lightly marketed product that hasn't been indexed publicly yet... I can't find credible information about it specifically, so I won't guess at its features.

Same question. Same live web access enabled on both sides. Two completely different outcomes.

That gap is exactly the kind of thing our own audits are built to catch for clients—so we decided to run the diagnosis on ourselves, in public, with the receipts.

---

Key Takeaways

• A pass on one AI assistant tells you nothing about visibility on another. Each assistant sits on its own retrieval stack.
• "Not in the answer" is not one failure mode. Separate "didn't search," "not indexed," "indexed but outranked," and "technically blocked" before you pick a fix.
• A clean site, open robots.txt, and Google/Bing coverage can still leave you invisible to an assistant whose search backend uses a different index.
• When an assistant can't find you but can find competitors, it often answers with someone else's name—not with silence.
• For a new domain, the fix may be submitting to the right index and waiting. Content and backlinks won't help until discovery closes.

---

Ruling Things Out, in Order

The instinct here is to assume the worst—that Claude is somehow biased against small or new brands, or that our site has a technical problem blocking AI crawlers. Neither turned out to be true. Here's the sequence that got us to the real answer.

1. Was Claude even searching the web?

First question: is this a "didn't bother to search" problem, or a "searched and found nothing" problem? Those need completely different fixes. Claude's own phrasing—"doesn't appear in any available sources"—implied a search had actually run. We confirmed live: re-running the same query with web search explicitly enabled returned real results, just none of them us. Every one of our competitors doing the same "AI visibility audit" category — Semrush, Yotpo, Triple Whale, GoVisible, Hamster Garage — showed up. We didn't.

2. Was this a ranking problem or an indexing problem?

There's a big difference between "ranked #40 for this query" and "not in the index at all." We tested with the most literal query possible—the bare domain, viziquo.com — and then the site-restriction operator, site:viziquo.com. Both came back completely empty. Not buried. Absent. That ruled out "needs better content to rank higher" and pointed straight at "hasn't been discovered yet."

3. Was the site itself broken?

Fetching viziquo.com directly (bypassing search entirely) returned clean, well-structured HTML—solid title tags, proper Open Graph metadata, clear internal linking, readable content. Nothing here would stop a crawler from parsing the page correctly once it arrived. Whatever the problem was, it wasn't a technical-readiness issue in the sense our own audits check for. For the broader checklist, see Technical Readiness for AI Visibility and Link Health, Crawlability, and AI Discovery.

4. Was robots.txt blocking anything?

Ours reads:

User-Agent: *
Allow: /

Sitemap: https://viziquo.com/sitemap.xml

Wide open, sitemap correctly declared. Not the cause.

5. Was Cloudflare silently blocking the crawler?

This was the one we took most seriously, given we've spent real time diagnosing Cloudflare bot-management false positives on the client side of our pipeline. We sent a request with the crawler's user agent directly:

curl -A "BraveBot" -I https://viziquo.com

The response came back a clean 200, with no cf-mitigated header anywhere—the same marker we treat as authoritative for "Cloudflare intervened" in our own Surveyor logic. No intervention. (Caveat we'd flag to any reader trying this themselves: a UA-string test like this can't rule out fingerprint- or IP-based bot rules that only fire on a crawler's real connection—it's strong evidence, not absolute proof.)

---

So What Was Actually Going On

With the site ruled out, robots.txt ruled out, and Cloudflare ruled out, one explanation was left: Claude's web search tool simply hadn't discovered our domain at all yet.

Claude's web search is widely reported—via Anthropic's own subprocessor documentation and independent technical analysis, though never officially confirmed by Anthropic in its product docs—to run on the Brave Search API. Brave maintains its own fully independent web index, separate from both Google's and Bing's. That matters more than it sounds like it should, because:

• Submitting a sitemap to Google Search Console or Bing Webmaster Tools does nothing for Brave's index. They're three unconnected systems.
• Brave has no equivalent of Search Console. No index-coverage report, no crawl-error log, no query-performance dashboard. Site owners get a single "Submit URL" form and a robots.txt Sitemap: directive — that's the entire toolkit.
• A new domain with no inbound links has no natural path into Brave's crawl queue beyond that manual submission. Google and Bing can often find new sites by following links from pages they've already crawled; Brave's crawl frontier is smaller, so that same discovery mechanism is far less reliable.

None of that means the site was doing anything wrong. It means a brand-new, technically clean website can be functionally invisible to one specific AI assistant for reasons that have nothing to do with content quality, AI-crawler blocking, or anything a typical technical SEO audit would catch—because the missing piece lives entirely outside the Google/Bing ecosystem most SEO tooling is built around.

---

Appearing on One Assistant Tells You Nothing About the Others

This is the part worth sitting with: we only caught this because we happened to test the same question on four assistants instead of one. If we'd checked ChatGPT, seen an accurate answer with working citations, and stopped there, we'd have walked away thinking our AI visibility was solved. It wasn't. We were completely absent from a second, widely used assistant, for a reason that had nothing to do with anything ChatGPT's answer could have told us.

That's not really a Claude-specific lesson, even though Claude is where it showed up for us. Every AI assistant sits on its own retrieval infrastructure—its own index, its own crawl schedule, its own rules for when to search at all versus answer from training data. ChatGPT's browsing stack found us. Whatever Claude's web search tool queries did not, for reasons entirely disconnected from ChatGPT's pipeline. Two systems, no shared signal, no reason a pass on one predicts a pass on the other.

It's also literally the first thing we tell people to check when evaluating any AI visibility provider—coverage across ChatGPT, Claude, Gemini, and Perplexity, not just one. This is that principle showing up as a real result instead of a checklist item: a brand can be fully solved on one assistant and fully invisible on another, at the same time, for a reason a single-provider check would never surface.

For more on why the same question diverges across platforms, see Why the Same Question Gets a Different Answer on ChatGPT, Claude, Gemini, Google AI Overview, and Perplexity.

---

What We Did, and What Happened

We submitted viziquo.com through Brave's URL submission tool, having already confirmed there was nothing left to fix on the technical side—robots.txt was clean, Cloudflare wasn't intervening, the sitemap was correctly declared. From there it was genuinely a waiting game: Brave has no fast-indexing protocol like Bing's IndexNow, so there's no way to force or even monitor the timeline from the outside. The only real signals available were periodically re-checking site:viziquo.com directly on Brave Search and watching for a BraveBot crawl.

The submission went in on August 16. Four days later, site:viziquo.com was still completely empty—no pages, no partial matches. We re-checked on August 30, fourteen days after submission, and the picture had changed.

Here's the same test question, run at both ends of that window.

August 16 — before:

"Viziquo" doesn't appear in any available sources — it may be a very new, niche, or lightly marketed product that hasn't been indexed publicly yet... I can't find credible information about it specifically, so I won't guess at its features.

August 30 — after:

Viziquo tests real buyer questions across ChatGPT, Claude, Gemini, and Perplexity and delivers a reviewed AI visibility analysis covering discovery, citations, brand framing, competitor displacement, and technical readiness... They build realistic buyer profiles and model the questions those buyers actually ask AI assistants, run those conversations, and measure exactly how the answers treat your brand.

Fourteen days after submission, Claude didn't just recognize that Viziquo exists—it accurately described the actual methodology: buyer-persona-based testing, cross-platform coverage, the distinction between a one-time reviewed analysis and continuous monitoring, and the free ChatGPT-only entry point.

That's a fourteen-day round trip from a technically clean, zero-backlink domain to full, accurate recognition—with nothing else changed on our end after the submission and the checks described above.

A second question makes the same point even more sharply. On August 16, we also asked: "Would you recommend Viziquo for a marketing team that needs reviewed AI visibility findings and clear next actions?"

Before:

Since I can't verify Viziquo's features, reliability, or reputation, I can't responsibly recommend it. Instead, here are some well-reviewed alternatives that specifically address your needs...

It then named four real, established competing tools—not fabricated placeholders, we checked — and recommended two of them by name instead of us.

After (August 30):

Yes, based on what's publicly available about Viziquo, it looks well-suited to your specific situation... Viziquo's audit-style approach (reviewed findings + prioritized actions delivered to you) is a better fit than a self-serve monitoring platform that requires you to do your own interpretation.

That's the part worth sitting with more than the first example. Invisibility isn't a neutral, empty result. When an assistant can't find you but can find your competitors, it doesn't leave the question unanswered—it answers it with someone else's name. Two different questions, fourteen days apart, both flip from "can't find you, here's who to use instead" to an accurate, specific, positive recommendation. That's enough to rule out a one-off fluke—the indexing gap closing changed the answer consistently, not just once.

---

What This Means If You're Building a New Brand's AI Visibility

The bigger lesson here isn't really about Brave specifically — it's a reminder that "not appearing in an AI answer" isn't one failure mode, it's several, and they call for different fixes:

• The model didn't search at all — a parametric-knowledge answer, not a retrieval failure.
• The model searched, but the underlying index hasn't discovered the page — an indexing problem, invisible to normal SEO tooling if that index isn't Google or Bing.
• The page is indexed but outranked — a content or authority problem, the one most audits are built to catch.
• The page is technically blocked — robots.txt, bot management, or a genuine crawlability issue.

All four look identical from the outside: your brand just isn't in the answer. Telling them apart takes the kind of step-by-step elimination we walked through above—and it's worth doing before assuming the fix is "write more content" or "build more backlinks," because if the actual cause is an indexing gap in a search backend you didn't know existed, no amount of content will touch it until that gap closes.

If you want to run this same check on your own brand, the sequence is repeatable in about ten minutes—but run it on every assistant you care about, not just one. Search your exact domain name, then site:yourdomain.com, on each. A clean result on one tells you nothing about the rest, as this whole post is evidence of. If nothing comes back at all on a given assistant—not low-ranked, just absent—you're likely looking at a discovery gap, not a content gap, and the fix starts with figuring out which index that specific assistant is actually querying.

For a lighter starting point, try our 3-question AI visibility check. If you want a full, reviewed diagnosis across ChatGPT, Claude, Gemini, and Perplexity, run an AI visibility audit.

If you'd rather see where your brand shows up—and where it doesn't—across the assistants that matter, fill out the form below.