Our first post about AI visibility got removed by LinkedIn's AI content bot. The irony is not lost on us. They reviewed it, reinstated it, and apologised. We're choosing to take that as a sign we're saying something worth reading.
Now - back to the experiment.
Last week we published the most uncomfortable post we've ever written. We're a digital agency that specialises in Answer Engine Optimisation (AEO). We help clients show up in AI-generated answers from ChatGPT, Google Gemini, and Perplexity. And our own brand visibility in AI search was 0.21%.
We said we'd fix it publicly. Here's what happened.
We didn't just talk about it - we owned it. We pulled the roadmap together and got to work. Here's what we implemented:
1. We fixed who we are.
The most urgent problem was brand disambiguation. Ask ChatGPT about BLOOM Transitions and it would offer two possibilities - us, and a US youth nonprofit called BLOOM Transition. Same name, completely different organisations. AI genuinely couldn't tell us apart.
We fixed this by adding Organisation schema to our homepage - structured data code that tells AI engines exactly who BLOOM Transitions is, where we operate, what we do, and who founded us. We also rewrote the opening copy on our homepage and About page to answer the question "What is BLOOM Transitions?" Plain, unambiguous language. No marketing fluff. Facts.
Ask ChatGPT (or any AI search engine, to be fair) who BLOOM Transitions is today. You'll get the right answer. Tell us if we're wrong.
2. We built and published an AEO services page.
We created a dedicated AEO services page structured specifically for AI citation. Every section is written to directly answer a question a buyer might ask AI. The page includes FAQPage schema, Service schema, and eight Q&A pairs covering everything from "what is AEO?" to "does BLOOM work with Australian businesses?" AI gold right there.
3. We added FAQ schema across every service page.
Our HubSpot services page, Shopify services page, and our SEO services page all got a sprinkle of FAQ schema fairy dust. Each one now has machine-readable Q&A pairs that give AI engines clean, citable answer units to pull from.
4. We rewrote our HubSpot Marketplace listing.
One of the most authoritative citation sources in our category was rewritten to be AEO-optimised. Plain language, specific services named, regions stated clearly, and AEO mentioned explicitly. This matters because HubSpot.com is the single most cited domain in our competitive space.
5. We updated our language settings to English (New Zealand).
A small but super-important signal. Setting the site language to en-NZ tells AI engines and Google that our content is intended for a specific audience - consistent with our schema's areaServed field and our geographic keyword strategy. That's not to say we aren't global and flexi to wherever your business is based, but it was appropriate for this exercise. And, more importantly, it worked.
6. We published an AEO blog post.
Our first blog post explaining what AEO is and why it matters for NZ businesses went live - structured for AI extraction, with FAQPage schema and direct answers to the questions buyers are actually asking.
In total:
15+ roadmap actions completed.
$0 spent on paid media.
Several happy dances.
We expected results to take 8-12 weeks. That's what we've been telling clients to keep expectations realistic. And that's what the data generally shows. However...
From 0.21% to 0.45%. Before we've built a single new backlink, before we've pursued a single external citation, before we've asked a single client for a review.
We changed the structure and clarity of our own content. And AI engines noticed.
We're not getting ahead of ourselves. 0.45% is still well below where we need to be. But it tells us the foundations are doing their job. And it tells us something important about AEO that we'll be sharing with clients: getting the basics right moves faster than most people expect.
The foundation work is done. Phase 3 is about building the citation profile - getting BLOOM mentioned on the external sources AI already trusts and actively cites. It's definitely trickier, but here's what's we'll be working on over the next four weeks:
1. Citation building.
Identifying and pursuing the NZ directories, agency roundups, and peer platforms where AI is already looking for agency recommendations. If AI cites a source, we need to be on it.
2. Client reviews on HubSpot Marketplace.
AI heavily weights third-party validation. Three or more client reviews on our HubSpot Solutions Marketplace listing will directly influence how AI answers "best HubSpot partner NZ" type prompts.
3. A client case study.
One well-structured case study showing a real client result, built for AI citation. Case studies are high-trust sources. One good one outperforms ten blog posts for AI citation purposes.
4. Ongoing tracking.
We're measuring brand visibility, share of voice, citation rate, and keyword positions monthly. The data goes into the LinkedIn posts and the blog. No spin, no cherry-picking.
Follow our 90-day AI visibility experiment on LinkedIn to see exactly what we're doing and how it's working. We're sharing every step - the wins, the data, and the moments where we got it wrong.