By Samuel Thimothy · Samuel Thimothy
Published Jun 24, 2026
8 min read
AEO
If your service page can’t answer a question in the first 50 words, an answer engine will quote someone who can. We rewrote 12 pages to test that, and the results were louder than we expected.
For the past 41 days, we’ve been running a controlled experiment on a real B2B brand — a mid-market industrial software company with about 2,400 monthly organic sessions and roughly 60 indexed service pages. The hypothesis was simple, and frankly a little stubborn:
The pages that win citations inside an LLM aren’t the pages that rank #1 on Google. They’re the pages that answer a question before they sell.
So we picked 12 service pages — six in our test group, six matched as a control — and rewrote only the test group with what we’re now calling the four answer-first blocks. Everything else (URLs, internal links, schema, page speed, even meta titles) stayed identical. Then we tracked 80 high-intent prompts across ChatGPT, Perplexity and Claude for 41 straight days.
This is not a Google ranking story. The pages in the test group barely moved on traditional SERPs. What moved was citation behavior inside the AI answer layer — and that turned out to be the more interesting variable to chase.
The rewrite wasn’t a redesign. We didn’t change the visual layout, the hero image, or the CTAs. We just inserted four content blocks — usually in this order — and removed anything that contradicted them.
A single paragraph, between 35 and 55 words, that answers the page’s primary question in plain English. No marketing language. No callouts. Just a clean, declarative answer that an LLM can lift verbatim. Think of it as the “featured snippet” for the AI era.
One sentence that defines the core noun on the page (“What is a CMMS?”, “What is rapid prototyping?”) — followed by two sentences of context. Citations cluster around this block surprisingly often.
3 to 7 short steps, each starting with an imperative verb. We tested both ordered lists and prose paragraphs. The list form was cited 2.3× more often.
Two to four rows, three columns max. LLMs are oddly good at pulling structured tabular content into answers, and our test pages with a comparison table picked up citations on long-tail comparison prompts within 9 days.
The headline number — 7.4× share-of-voice lift — is real but slightly cheating. It includes a few prompts where the control group simply wasn’t being cited at all, which makes the multiplier dramatic. So let’s look at the absolute numbers instead.
Add Block 1 (the 50-word answer) to your top 5 pages this week. In our data, that single block was responsible for the biggest absolute lift — about 60% of the citation gain came from it, even when the other three blocks were skipped.
Request an AI Visibility Audit and we’ll show you which of your pages are getting cited — and which to rewrite first.
We’re publishing what didn’t work, too — partly for honesty, partly because the wrong assumptions are usually the more interesting story.
We expected FAQ schema to be a fast win. It wasn’t. Pages with and without FAQ schema picked up citations at statistically identical rates. The schema mattered less than the answer underneath it.
We added linked author bylines (with structured author data) to six pages and saw no measurable change in citation frequency. Author signals may still matter for E-E-A-T inside Google, but they didn’t move LLMs for us in this 41-day window.
Counter-intuitive: when we added freshness dates to two pages, citations dipped briefly. Our hypothesis is that the LLMs may briefly de-rank pages that look like they’ve been recently changed, until the content stabilizes. We’re testing this in isolation next month.
41 days is a short window. The freshness-penalty finding is the kind of thing that needs replication before you should trust it. We’ll publish a longer follow-up at Day 90 with a wider sample.
Read on for the copy-ready playbook — the same eight steps we’ll be teaching in the AEO Foundations track next month.
If you want to replicate the experiment on your own pages, here is the operator’s checklist. It’s the same one we’ll be teaching in the AEO Foundations track next month.
Day 41 is a milestone, not the finish line. The next two updates will cover (a) whether the citation lift is durable past 90 days, (b) whether the “freshness penalty” finding holds up in isolation, and (c) what happens when we apply Block 1 alone to a much larger sample of 200 pages on a separate brand.
If you want to see where your own pages stand, request an AI Visibility Audit at the bottom of this page — we’ll benchmark your citations and flag the pages to rewrite first.
AEO is the practice of structuring content so that AI answer engines — ChatGPT, Perplexity, Google AI Overviews and others — quote your brand as a source. Unlike traditional SEO, the goal isn’t a blue-link ranking; it’s being cited inside the generated answer.
SEO optimizes for ranking in a list of links; AEO optimizes for being selected and quoted in a synthesized answer. In our testing, the pages that win citations are the ones that answer a question in the first 50 words — not necessarily the pages that rank #1 on Google.
In our 41-day experiment, FAQ schema alone did not change citation rates. What mattered was the answer underneath it — a clean, declarative 50-word answer placed directly under the H1.
It varies by engine. Perplexity moved fastest in our tests (citations within ~4 days), ChatGPT was slower (around Day 11), and Claude was the most volatile day-to-day. Plan for a 30–90 day window before drawing conclusions.
Request an AI Visibility Audit. We’ll show you where the major answer engines cite you today, benchmark your share of voice against competitors, and prioritize the pages to rewrite first.
Samuel Thimothy has deep expertise and experience in online marketing, demand generation and sales. He helps businesses develop and execute marketing strategies that will improve their lead generation efforts and drive business growth. He serves as the VP at OneIMS, an inbound marketing agency and co-founded Clickx, the digital marketing intelligence platform that eliminates blind spots for brand marketers and agencies.
Request your AI Visibility Audit and we’ll show you where ChatGPT, Perplexity and Google AI Overviews cite you today — and the pages to fix first.
TL;DR We rewrote 12 B2B service pages with four “answer-first” blocks and tracked 80 prompts for 41 days. LLM citations rose +184% vs. a ...
TL;DR We rewrote 12 B2B service pages with four “answer-first” blocks and tracked 80 prompts for 41 days. LLM citations rose +184% vs. a ...
TL;DR We rewrote 12 B2B service pages with four “answer-first” blocks and tracked 80 prompts for 41 days. LLM citations rose +184% vs. a ...
TL;DR We rewrote 12 B2B service pages with four “answer-first” blocks and tracked 80 prompts for 41 days. LLM citations rose +184% vs. a ...
