How can tech companies get recommended in AI search?

Make it easy for AI to see which buyers your product suits and why, on your own website and on the other websites AI reads. Good Google SEO is the foundation, but it is not the whole plan. AI tools search more widely than one keyword, cite a mix of websites and change their sources often. Start from your buyers’ own words, not your feature list or a keyword tool: in our research, keyword-led plans missed the questions buyers actually ask.

Avouch’s free scan shows which companies AI recommends in your market.

This guide is for software and technology products. For service companies, see our guides for IT support and managed service providers and cyber security companies.

What our research shows for tech buyers

• AI Overviews are on almost every buying question. In our tracking, an AI Overview appeared on 96% of US and 92% of UK results pages for buyer questions about AI SEO services, and on 90% of US and 76% of UK pages for data protection and cyber security questions. See Do AI Overviews take your clicks?.
• Your own website supports a recommendation, but rarely alone. In our study of AI answers about IT support, cyber security and app development companies, suppliers’ own websites appeared beside most of their recommendations, but almost never as the only source. Directories and comparison pages were among the other websites cited. See the websites behind repeated AI recommendations.
• The sources change all the time. About half of the websites AI cited for a question were new from one day to the next. See how often AI answers change their sources.

Why a keyword list leads to the wrong content

A keyword list tells you what people type. It does not tell you what a buyer is trying to decide. In our study of a data protection market, a close neighbour of tech:

• 33% of the saved SEO phrases asked for a definition, such as “what is a dpo”. None asked about cost or contract terms.
• 25% of real buyer questions were about cost or terms, such as “How do solo devs handle a DPIA without breaking the bank?”.
• 21% of the searches in the client’s own Search Console were about cost or terms, including “how much does outsourced dpo cost for a 50-person saas startup?”.
• Buyer questions were longer: a typical buyer question had eight or nine words, and a typical keyword three or four.
• Fewer than 1% of the searches Gemini made matched the keyword list exactly.

A content plan built from the keyword list alone would have written definitions and left the buyer’s real questions, about cost, terms and fit, unanswered. Definition pages attract people learning the subject, and many of those people are not buying. See what SEO keyword lists miss about buying decisions.

1. Collect your buyers’ own words

Start with what buyers say, in their own words, before you open a keyword tool. We call this a voice harvest.

Where tech buyers talk. In our harvest for a data protection and cyber security market, the most useful places were Reddit, Hacker News, Spiceworks (a community for IT professionals), Trustpilot, Capterra, G2, Clutch, Stack Exchange, UK Business Forums and YouTube. Add what you already hold: sales calls, demos, support tickets and lost-deal notes.

Check who is speaking. Not everyone who talks about a subject is buying it. We checked 1,832 posts from that harvest that had first been marked as buyers. A second check found:

• 40% were buyers;
• 38% were practitioners, people who do or sell the work;
• 15% were other speakers, such as news and career questions;
• 6% were people affected by a data breach, not people buying protection.

If you write from all of them, you answer the practitioner’s questions, not the buyer’s.

What to record for each buyer comment:

Keep private conversations private. Remove identifying details from your notes, and ask permission before you publish a customer’s words.

Then use keyword data to size and prioritise the questions, not to choose them.

2. Turn the questions into pages

3. Find out what else AI searches for

When it builds an answer, an AI tool runs extra searches on related topics. This is called query fan-out. For “which project management tool works with our systems?”, the extra searches might cover integrations, data imports, permissions and price. Our guides show how to see these searches in Google and ChatGPT. Add the gaps you find to your list of questions.

4. Check the other websites AI relies on

Choose the buyer questions to track, ask them in each AI tool, and open the pages each answer cites. For software, these are often:

• Review sites such as G2 and Capterra. See G2 and Capterra reviews in ChatGPT.
• Best-of lists and comparison sites. See best-of lists and comparison sites.
• Comparison pages, including your competitors’ and your own. See comparison pages on your own website.

Record which companies each answer names and recommends, and which pages it cites. Then decide whether the next job is on your website or on one of these. Our guide to why AI recommends your competitors shows how.

5. Make sure AI can read your pages

Ask whoever runs your website to check that:

Add a short, factual AI information page, like ours: who you are, what you sell, who it is for and how to contact you. An llms.txt file is optional and comes after these.

6. Measure what changes

Track the same questions every week: how often AI names you, recommends you and cites your pages. Ask new customers how they found you. Our guides to AI visibility metrics and measuring the return explain how.

How Avouch can help

We start from your buyers’ questions and the websites AI relies on, then agree the work that fits: clear website and product pages, content that answers buying questions, reviews and industry profiles, PR and industry coverage and LinkedIn. Where it is part of an agreed engagement, we track your Recommendation Rate™ each week, with scope and price agreed before we start.

Want to know where you stand?

Start with the free scan. For a fuller review and a written plan, choose the paid Recommendation Index™ assessment.