The question I kept seeing in every Copilot advertising discussion: does optimising for AI answer engines actually drive measurable traffic? Everyone had opinions. I decided to run the numbers.
Over 30 days, I restructured a set of 15 content pages specifically for AEO (Answer Engine Optimisation) while keeping an equivalent control set optimised purely for traditional SEO. Here's what happened.
Methodology
Test pages (AEO-optimised): 15 product category and guide pages. Changes applied:
- Added FAQPage JSON-LD schema with 6–8 Q&A pairs per page
- Rewrote opening paragraphs to directly answer "What is X?" in the first 50 words
- Added H2 headings written as complete questions ("How does Copilot Checkout work?")
- Submitted pages to Bing Webmaster Tools with IndexNow for immediate re-indexation
- Added structured "How Copilot Sees This" sections with direct answer text
Control pages (SEO-only): 15 equivalent pages. No structural changes, continued standard SEO maintenance (meta updates, internal linking).
Measurement: I tracked three traffic sources separately: Google organic, Bing organic, and "AI referral" (sessions from Copilot, ChatGPT, Perplexity identified via referrer or no-referrer with known AI UAs). All measured via Microsoft Clarity + UTM parameters on Copilot-referred links where possible.
What the Data Actually Showed
The headline number — 340% increase in AI referral traffic to AEO-optimised pages — is real, but needs context. The baseline AI referral traffic was very low, so percentage increases are dramatic. In absolute terms, AI referral traffic went from accounting for about 2% of total traffic on test pages to roughly 9% by day 30.
The more interesting finding: Bing organic also improved by 18% on AEO pages, while remaining flat on control pages. This aligns with what we know about Copilot's index: it draws heavily from Bing, and better-structured content for AI consumption also appears to be better content for Bing's own ranking signals.
Google organic was essentially flat on both sets, with a marginal -4% on AEO pages that I attribute to noise rather than any structural impact from the changes.
What Specifically Drove Results
Looking at the data by specific change applied:
- FAQPage schema: The highest single-impact change. Pages with FAQPage schema started appearing in Copilot AI overview-style answers within 7–10 days of Bing re-indexation.
- Question-format H2s: Second most impactful. Pages with question-format headings were cited more often in full-sentence AI responses.
- Direct-answer opening paragraphs: Strong signal for pages where the user query was closely matched to the page topic. Weaker effect on broader category pages.
- IndexNow submission: Accelerated Bing pickup by approximately 3–5 days vs pages without it.
Important Caveats
Small sample, short timeframe. 15 pages over 30 days is not statistically rigorous. AI referral traffic measurement is imperfect — some AI traffic arrives with no referrer and is indistinguishable from direct. The 340% number should be treated as directional, not definitive.
What I'm confident about: the changes are net-positive across all measured channels, with no meaningful negative impact on Google organic. The risk/reward of implementing AEO optimisation is strongly favourable based on this data.
What You Should Do
Based on 30 days of data, here's the priority order for AEO implementation:
- Add FAQPage schema to every guide, category, and product page — 6–8 Q&A pairs, targeting the actual questions your audience asks about that topic
- Set up Bing Webmaster Tools and enable IndexNow — free, 20-minute setup, accelerates Copilot pickup of all your changes
- Rewrite the first 100 words of your top 10 pages — make them direct answers to the page's primary question
- Convert at least some H2 headings to question format — "How does X work?" performs significantly better in AI citations than "About X" or "X Overview"