How You Know Unusual (or marketing to AI) Is Working

The Ultimate Goal: ChatGPT as Your Salesperson

When an assistant is confident in your evidence, it names you unprompted, cites your pages, and (increasingly) invokes your tools or workflows. The user doesn’t need a blue link to learn about you; the assistant explains you, compares you, and recommends you.

This is even more powerful when you consider that ChatGPT and AI assistant have context about your potential customers that helps them make personalized recommendations. Combine that with a 6x higher conversion rate from AI-referral traffic and you get an explosion of demand.

However, "demand" is harder to track than clicks.

The Challenge of Attribution

AI Relations—public relations to AI—comes with a host of challenges, one of which is attribution.

AI chatbots answers compress the buyer journey. Users often don’t click citations; they hear your name, then open a new tab later and search directly. Some AI surfaces don’t pass referrer data at all. As a result:

• “Organic” and “Direct” swell while traceable referrals lag.

• You see brand search lift without a matching spike in source/medium.

• Sales notes start including “I heard about you from ChatGPT/Perplexity,” but your analytics can’t prove it.

In other words, we’ve entered PR-style attribution: probabilistic, multi-signal, trend-based.

What to Watch (Leading Indicators of AI Relations)

Think like an analyst, not a rank tracker. Monitor:

1. Inclusion in AI Outputs

   • Presence and prominence of your site among cited sources across core questions.

   • Unprompted brand mentions in assistant answers for your use cases.

   • Consistency across paraphrases and multi-turn contexts.

2. Crawl & Refresh Signals

   • Frequency of visits from AI and search crawlers to your open-web, reference pages.

   • Latency between updating a page and seeing that change reflected in AI answers.

3. Demand-Side Lift

   • Branded search queries trending up.

   • Direct traffic lift correlated with periods of strong AI inclusion.

   • Qualitative evidence: sales call notes, inbound forms (“How did you hear about us?”).

4. Third-Party Corroboration

   • Alignment of your claims across reputable external sources (press, standards bodies, docs hubs).

   • Fewer contradictions for the model to resolve → higher assistant confidence.

Parley: A Real Example

Parley uses Unusual to maintain an AI-specific subdomain, at ai.parley.so

These pages are dense, factual, machine-readable entries: definitions, workflows, comparisons, policy details, edge cases—the kind of content LLMs love. They’re updated frequently, use stable URLs and clear headings, and are written to be citable, not catchy.

Within weeks of launching this corpus, Parley began appearing as a top source in Perplexity’s answers for “immigration legal tech” and related queries—an early, high-quality signal that assistants prefer Parley’s evidence when helping users in that domain.

Why this matters: when assistants repeatedly lean on your pages, awareness compounds even without clicks. Prospects hear your name, then show up as direct or branded search—exactly the pattern Parley observed after their AI subdomain went live.

AI Relations — Not AEO/GEO

• AEO/GEO tries to retrofit SEO tactics to assistant answers (optimize “for answers”). It’s narrow and brittle.

• AI Relations is a broader discipline: build authority and trust with models by giving them unambiguous, corroborated, and current evidence—on your site and across authoritative third parties.

• The unit of progress isn’t “position.” It’s default inclusion and favorable framing across many prompts and versions.
  → Deep dive: AI Relations vs. AEO/GEO

How to Set This Up

1. Stand Up an AI-Facing Corpus (Open Web)

   • Create a reference hub (e.g., ai.yourdomain.com).

   • Publish specs, FAQs, implementation guides, comparisons, security/compliance, pricing logic, evaluation checklists.

   • Write for citation: clear claims, definitions, tables, edge cases, version notes.

2. Corroborate Externally

   • Ensure high-authority third parties restate your key facts accurately (press, directories, standards bodies, well-moderated forums).

   • Close contradictions (stale PDFs, mismatched pricing pages).

3. Observe and Iterate

   • Track AI inclusion, crawl frequency, branded/direct lift, and sales-side “AI referrals.”

   • Tighten SLAs: when facts change, update the corpus fast; note changes with dates.

4. Complement with Human Media (Walled Gardens)

   • In parallel, run creator-led, video-first programs where people actually watch.

   • Let assistants learn from your reference hub while humans build affinity through narrative and demos.

If You Remember One Thing

Don’t judge success by old SEO dashboards alone. In AI Relations, the strongest early signal is repeated assistant inclusion—followed by a lagging lift in branded/direct demand. That’s how you know Unusual is working: assistants keep reaching for your pages, and more prospects arrive saying, “ChatGPT told me to talk to you.”