How GEO helped Merge drive 7x more demo requests from AI search

Merge grew demo requests from AI search 7x by tracking hundreds of buyer prompts, studying which competitor pages the engines cited, and rebuilding its own content to match. LLM search traffic rose 10x and conversion from AI search ran 6x higher.

Merge sells unified APIs and agentic tooling to B2B SaaS companies, Fortune 500 organizations and frontier LLM providers. It is a crowded market, and every competitor wants the same answer slot when a buyer asks an AI assistant which integration platform to use.

The challenge

Merge already had some traction in AI answers. Its content was showing up, and Jon Gitlin, Content Marketing and SEO Manager, could see LLM traffic arriving in his existing SEO tool. What he could not see was why.

The old tool reported the traffic but could not explain what drove it or how to get more. "AEO was still a black box," he explains. "We knew it was a promising growth channel, but we needed guidance on how to capitalize on it."

Three gaps stood between early traction and a real channel:

• No clear view of mentions. How often did LLMs reference Merge, and in what context?
• No competitive benchmark. In a saturated market, Merge needed to know where it ranked on every prompt tied to its two product lines, Merge Unified and Merge Agent Handler.
• No content direction. Which formats and topics did the engines actually favor?

What they changed

The team set up an AI visibility program built around prompts, not keywords.

They tracked hundreds of prompts across the major engines. Answers from ChatGPT, Perplexity, Claude and Gemini were monitored continuously. Every prompt was tagged by product line and sub-topic, so results could be read per product instead of as one blended number.

They benchmarked against seven competitors. Merge compared overall mentions and sentiment against each rival. But most of the value came from prompt-level data: exactly where Merge ranked on each prompt it targeted.

They reverse-engineered the most-cited pages. For every tracked prompt, the team could see which URLs the engines cited most. Gitlin studied what those pages had in common and built those traits into Merge's own content.

The clearest example is a piece on the "best unified API platforms." Before writing, he analyzed the headings, listicle format and section structure of the top-cited articles on that topic. He then published a tightly structured article that made Merge's own value props easy to extract. It became Merge's most LLM-surfaced page, with 2,600 mentions in a single month.

"Being able to compare ourselves to competitors on such a granular level is huge," Gitlin says. The work lifted Merge on many non-branded prompts, the queries where buyers have not yet picked a vendor.

The results

Since the program began, Merge reports:

• 7x increase in demo requests from AI search
• 10x boost in LLM search traffic
• 6x higher conversion rates from AI search

According to the source, Merge now has higher LLM visibility than any other company in its market. The traffic that visibility sends is high intent, which shows up in signups, demo requests and pipeline, with direct attribution from the channel.

What you can copy

Tag prompts by product line. One visibility score across a whole company hides where you win and where you lose. Group prompts by product and sub-topic so each team sees its own number.

Pick a fixed competitor set. Merge benchmarked against seven named rivals. A stable set makes week-over-week movement meaningful.

Let cited URLs set the brief. Before writing, pull the pages AI engines cite most for your target prompt. Match their structure (headings, format, sections), then add what only you can say.

Target non-branded prompts. Branded queries mostly confirm a choice already made. Non-branded "best X for Y" prompts are where the shortlist gets built.

Measure down to demos. Visibility is the leading signal. Track AI-referred traffic, conversion and demo requests so the channel earns budget on revenue, not mentions.