Week 2 of AEO: We checked our AI traffic. It's not great.
Week 2 of Olark's AEO journey: we go looking for our own AI search traffic and find it's... not great. Here's what our tools do (and don't) tell us.
Quick recap: last week we admitted we’re basically clueless about AEO. This week: where do we actually stand? Short answer: our AI-sourced traffic isn’t great. Here’s what we found.
The paid tools rabbit hole
Rhi ran us through Profound, which put us #5 against Intercom and Zendesk. Interesting, but we don’t yet know how it calculates that, so it’s a data point, not gospel. Her verdict: “the expense is probably not worth it for the number of LLM tools it monitors.” For now, we turned to tools we already have.
What our existing stack tells us
- HubSpot’s AEO dashboard shows domain visibility, competitor comparisons, and top URLs. Its take: “Your brand visibility in AI searches comes down to one thing: tracking the right prompts.” Our score: under 1%. 😭
- Google Search Console has a “Generative AI features” report — but it’s Google-only (AI Overviews, AI Mode, Discover). Nothing on ChatGPT, Perplexity, or Claude. Still, it showed 6.5K impressions in 28 days, our biggest AI number yet — a clue for later.
Naturally, we asked: is there a “Claude Search Console”? Nope. Closest thing is claude.ai referral traffic in GA4. Last 30 days:
- 427 sessions from chatgpt.com
- 1 session from claude.ai
That gap looks huge, but it’s probably not real. Chatbots are supposed to pass “referrer data” telling our site where a visitor came from, but most don’t do it consistently, and Claude seems to be the worst offender. So a lot of that traffic just lands in our “Direct” bucket instead. We likely have far more Claude-driven traffic than “1 session.” We just can’t see it.
So why’s the traffic low in the first place?
A webinar’s argument: AI models source brands based on domain credibility, and credibility comes from consistent publishing, not occasional great content. We’ve always favored sporadic, thoughtful posts over steady cadence (small team!). That inconsistency might be exactly what’s holding us back.
One line that stuck with us: “Do one thing that’s working for you and measure it.” Simple, and a little less intimidating.
Next week: we build the action plan.
See you next week,
Team Olark (Ben, Mandy, Rhi, Atinder, Logan)