The same page type produced leads at three companies, then AI search started retrieving it

The situation

This is a story about one page type, watched across three companies at the same time: a Dutch fintech lender, a workplace hardware company, and a SaaS platform for event businesses. Different industries, different markets, same gap.

In every one of these markets, the highest-intent searcher is not typing the product category. They are typing a competitor's brand name with doubt attached: "[competitor] reviews", "is [competitor] reliable", "[competitor] disadvantages". A buyer with a shortlist, looking for the deciding factor. None of the three companies had a single page answering that moment.

The backdrop matters, because it was not friendly. At the lender, the core commercial pages were losing rankings to the broad devaluation of commodity content. At the hardware company, paid budget had been cut by 85% and total lead volume was falling. This was not a rising tide. It was the opposite, which is what makes what happened next worth writing down.

The play

The method was identical at all three companies, and it was in each strategy document before any page was built:

The first page becomes the template, the rest of the cluster reuses it, and the money pages link to them. At the hardware company my research note said it plainly before any data existed: no one in this space has comparison pages. That gap was the bet.

Act one: intercepting competitor doubt

The lender's first comparison page went live in late October 2025. It had zero search presence before November, so its "before" is genuinely zero.

• 13 clicks in its first month, 290 a month by spring, holding page 1
• Ranking top 10 for the competitor's own brand name, which alone put the page in front of 105K searches in a single month, 428 of them clicking through
• Ranking for the doubt queries around that brand: the local equivalents of "is [competitor] reliable", "[competitor] disadvantages", "[competitor] complaints", "[competitor] high interest"
• 18 leads attributed to the page as the organic entry, and still compounding month over month

This is not a "[us] vs [them]" keyword play. The volume lives in the competitor's own brand demand, and the page intercepts the skeptical slice of it: buyers who are already in-market and already doubting.

The hardware company proved the same thing in a market a fraction of the size. Four comparison pages reached page 1 within 4 to 8 weeks of launch, and the lead page among them collected 11 attributed leads off single-digit monthly clicks. When the intent is this sharp, you do not need traffic. You need the right 30 visitors.

Act two: the same pages are what AI search retrieves

I ran the same play at the event-software SaaS, against three of its competitors, and through the spring the comparison pages climbed from around position 26 to single digits. Then the search data changed shape.

By June, monthly impressions on the lead comparison page had grown from 115 to over 2,000, and the queries stopped looking like keywords. They became whole questions: "best alternatives to [competitor] for inventory management", full natural-language comparison prompts, the kind of thing nobody types into Google but everybody types into an assistant.

That is exactly what they are. When an AI assistant answers a comparison question, it fans out searches of its own, and those machine-issued queries land in Search Console like any other impression. The comparison pages were ranking for them.

So I tested it from the other side. I asked a live, web-searching AI assistant the same questions the buyers ask. Asked for alternatives to the competitor, it recommended my client in its shortlist. Asked the direct "which is better" question, it recommended my client and cited the client's own site in the answer. The fan-out queries it fired match the queries in Search Console almost word for word.

Here is the part I care about: nothing in this engagement was done "for AI search". No GEO checklist, no llms.txt, no retrieval hacks. AI search is still user search: a buyer asks the assistant a question, the assistant runs searches on their behalf, and whatever ranks for those searches is what gets retrieved and quoted. The pages rank for the questions real buyers ask, so assistants retrieve them. The boring, revenue-first work wins the new channel too.

What I'd tell you honestly

The lead numbers are attributions, not closed revenue. The 29 leads are recorded against the comparison pages as the organic entry page, from lead sheets the clients maintain. I verified the attribution model against the analytics setup before quoting them, and I would still call them directional. Neither client tracks pipeline euros against pages yet.

Comparison pages were a counter-move, not a rescue. At the hardware company, total organic leads fell across the engagement because paid budget collapsed and brand demand with it. The comparison pages produced leads against that current. They did not reverse it, and I will not pretend they did.

The AI answers mostly cite review aggregators, not the comparison pages. When the assistant answered the "which is better" question, it cited G2, Capterra and the client's own site. The provable win for the comparison pages is ranking for the queries assistants issue, and being the reason the client makes the shortlist at all. If someone sells you "AI citations" for your comparison pages, ask them who actually gets cited in the category first.

What I would do differently: validate the competitor shortlist against demand data, even when it arrives with confidence. At the hardware company, the competitor came from the business as a top target. The research behind that call was theirs, and I trusted it matched the markets we were focusing on. The data later showed most of that brand's demand sits somewhere else: the page converted anyway, but it also pulled in leads from countries the client does not serve. Using the business's competitor input was right. Skipping my own check of where that competitor's demand actually lives was the miss, and that check is now a standard step before any comparison page gets built.

The next test is on me

Twelve days before publishing this, I shipped the same play on my own site: a four-way comparison table on the fractional SEO page, my own honest "when I am not the right fit" section included. It is too fresh to have data. When it does, the receipts land here, whichever way they point.