Getting found in AI search, not just Google

The challenge

B2B buyers have moved a large part of their research into AI tools. In Forrester’s 2024 Buyers’ Journey Survey, 89% of B2B buyers used generative AI in at least one stage of their purchase. Buyers still shortlist brands they know, but AI answers are adding to that list: in G2’s 2026 research, 69% chose a different vendor than they first planned because of what an AI chatbot told them.

Ranking on Google wasn’t enough on its own. If a healthcare leader asked ChatGPT for a trusted technology partner in Australia, we needed to be in that answer, and described in the right way.

My approach

I built on the SEO foundations I already had in place. I had developed our content plans from keyword research across every stage of the buyer journey (awareness, consideration, and decision), and looked after on-page SEO (titles, headings, meta descriptions, alt tags, and schema markup) as well as off-page work: backlinks, review sites, and PR.

AEO needed a different starting point, because people use AI tools differently. On Google, people type short keywords. In an AI tool, they ask a full question and add context about who they are and what they need. Google has reported that searches in its AI Mode are around three times longer than traditional searches. So I started with the questions we wanted to be the answer to.

Manual prompt research. When these tools first appeared, I took our long-tail and short-form keywords and turned them into the questions buyers were likely to ask. For example, a keyword like "healthcare technology partner Sydney" became a prompt like "Who are the best technology partners for a hospital group in Sydney that needs secure, compliant systems?"

I ran each prompt in ChatGPT, Gemini, and Claude in incognito mode, then reviewed what came back:

• which companies were recommended, and what was said about them
• which sources were cited and linked
• what type of content was being surfaced: product pages, review sites, roundup articles, Reddit threads, or YouTube videos

I then assessed where we could realistically compete and what it would take. That might be a service page modelled on a competitor’s that was being cited, more reviews on a third-party review site, or a place on a "top 10" roundup article. It was hands-on, prompt-by-prompt work, but it gave us our first clear picture of how AI tools saw our market.

What I built

Daily tracking with HubSpot’s AEO tool. When HubSpot released its AEO tool, I set it up with the full context of the business: our services, our competitors, and the buyer personas we were targeting. Rather than me estimating prompts from keyword data, the tool generated the prompts most relevant to those personas and services, and tagged each to a stage of the buyer journey.

It then ran those prompts across ChatGPT, Gemini, and Perplexity every day, giving me a daily record of which brands and sources appeared for each prompt. For each one I could see:

• Visibility and share of voice: how often we were mentioned compared with our competitors
• Sentiment: whether we were described positively, neutrally, or negatively. Showing up only helps if what’s said about us is positive
• Citations: which pages, domains, and content types were driving our competitors’ recommendations
• Recommendations: a prioritised list of content to create, with the topic, format, and structure, and examples of competitors being cited for the same prompts

A content plan for management. I pitched each content direction to management with the prompt we wanted to win, who was currently being recommended, and the type of content earning those citations. The plan covered product and service pages, blog articles, review sites, community platforms such as Reddit, and video.

A healthcare focus. AI tools were already citing us for healthcare, and it was also where many of our enquiries came from, so we went deeper there. It built on the healthcare focus that had already worked in our paid search and ABM programs.

A contract copywriter. Up to that point I had written all of our content myself. To scale up, I hired and briefed a contract copywriter to produce a new piece each week, many of them shaped by the AEO insights. I briefed, edited, and published each piece, and made sure it met our SEO standards too, since search was still an important channel.

Original research. We also invested in higher quality pieces that would stand apart from the AI-generated content filling other sites. I worked with a PR agency to survey more than 500 Australian businesses on a topic central to our industry, and turned the findings into a downloadable research report. It gave us original findings to publish and pitch to other sites.

Results

We could now see which prompts mentioned us, how we compared with competitors, and which sources were shaping AI recommendations.

The program gave us:

• A daily view of AI visibility, with share of voice and sentiment for the prompts that mattered most, across ChatGPT, Gemini, and Perplexity
• A content plan built on evidence, showing the prompts we wanted to win, who was being recommended instead, and the content earning those citations
• A healthcare focus, where AI tools were already citing us and many of our enquiries came from
• A steady flow of new content, with a new piece each week and an original research report to publish and pitch to other sites

In Ahrefs’ study of 75,000 brands, mentions of a brand across the web were the factor most closely linked to visibility in Google’s AI Overviews, ahead of backlinks.

Benchmark sources: Forrester, The Future Of B2B Buying, G2 research via Demand Gen Report, Google Q4 2025 earnings via Search Engine Journal, HubSpot AEO, Ahrefs, AI Overview brand visibility factors.

What I’d do next

If I ran this program again, I’d:

• Set an AI visibility baseline from day one, then report share of voice and sentiment for priority prompts monthly, alongside SEO results.
• Plan for the whole citation mix. AI tools draw heavily on third-party sources such as review sites, roundups, Reddit, and YouTube, so I’d build that work into the plan from the start, alongside our own content.
• Track AI referrals through to enquiries, with a "How did you hear about us?" field and AI referral sources reported in analytics, so the business can see which conversations start in an AI tool.