Client
Icons of Real Estate, which produces podcasts for real estate clients and turns those episodes into blog content on their behalf.
Business Objective
Adapt the blog content built from client podcasts so it has a real chance of being cited or surfaced by AI answer engines like ChatGPT, Gemini, and Claude, on top of ranking in traditional search.
Challenge
The existing workflow turned podcast episodes into solid, keyword-targeted blog posts. That was enough for traditional SEO, but a growing share of searches never touch a results page at all, they go straight through an AI answer engine instead. Ranking well in Google doesn’t guarantee an AI system will read your content, extract it cleanly, or trust it enough to cite it.
The posts had a few structural gaps that mattered more for AI visibility than they ever did for classic SEO:
- The answer to the reader’s question was usually buried a few paragraphs in, after scene-setting and context, which works fine for a human reader but gives an AI system nothing clean to extract.
- No FAQ sections, so there was nothing shaped like the actual questions people type into a chat interface.
- No tables, no specific figures, and no outbound citations, which meant nothing on the page signaled thoroughness or trustworthiness the way AI systems seem to reward.
- No consistent schema markup, and no real system for keeping older posts fresh, both of which matter more to answer engines than most teams assume.
My Role
I owned the AI answer-engine optimization strategy for the blog content, on top of my usual content and web development work at Icons of Real Estate.
Strategy
I treated this as a layer on top of existing SEO practice, not a replacement for it. The core idea across every change: write so a machine can lift a self-contained chunk of the page and have it still make sense on its own.
A few principles drove most of the decisions:
- Every post now opens with a short key-takeaways or TL;DR block, a direct, complete answer in roughly 40 to 60 words, before any of the usual context or storytelling. The test I used was simple, if an AI system extracted only that one block, would it still fully answer the question without the rest of the article.
- There’s a documented “equalizer effect” in how AI answer engines treat citations. Pages that reference credible outside sources tend to gain more visibility than pages that don’t, and this effect tends to help lower-ranked pages catch up more than it helps pages that already dominate. That made deliberate citation a genuinely useful lever for client sites competing against bigger, more established real estate media brands.
- Specific statistics consistently outperform vague generalizations in how AI systems select and cite content, so wherever a post made a claim, I pushed for a real number instead of something like “many buyers.”
- Since these posts already come from recorded interviews, pulling direct quotes from the actual guest into the blog post was a natural fit. Quoted expert voices tend to perform especially well for narrative and industry-commentary content, which describes most of what these podcasts cover.
- Research on this consistently shows there’s no single universal GEO tactic. Citing sources works best for factual, informational content, statistics carry the most weight for opinion and regulatory topics, and quotes perform best on narrative, people-focused content. I matched the approach to each post instead of applying one formula everywhere.
- Keyword stuffing, oversimplifying real information, padding posts with filler, and leaning on persuasive language with nothing behind it either did nothing or actively hurt, so cutting those out mattered just as much as adding what worked.
Execution
Working through the client sites, I rebuilt the standard blog template and post-production checklist around a few concrete requirements:
- A TL;DR or key-takeaways block at the top of every post, written to stand alone.
- Headers phrased as the actual questions readers ask, not generic keyword phrases, with one clear topic per section.
- An FAQ section on every post, built from real queries pulled from tools like AlsoAsked and Search Console rather than guesses, since actual AI prompts tend to be longer and more specific than a typical search keyword.
- Tables for anything comparative, since AI systems extract structured data far more reliably than they extract the same information written as prose.
- FAQPage and Article schema added to match what’s actually visible on the page, plus a technical check to confirm AI crawlers weren’t being blocked by robots.txt or CDN-level bot rules, and that no content was hidden behind JavaScript, accordions, or logins.
- A refresh cadence for cornerstone posts, with genuine “last updated” edits rather than a cosmetic date change, since freshness carries real weight for AI answer engines, especially on buying-stage content.
- A consistency pass across the client’s Google Business Profile and other third-party listings, since AI systems appear to build more confidence in facts that stay consistent across multiple sources rather than living only on one site.
Results
This work is still early, but the initial signal has been encouraging:
- Within the first two months, client posts began appearing in Google AI Overviews for roughly one in five of the target queries we track, up from close to zero beforehand.
- Referral traffic from AI platforms such as ChatGPT and Perplexity grew from negligible to a small but real share of blog sessions on the restructured posts.
- Organic traffic to the restructured cornerstone posts rose by roughly 20%, a reminder that most of these changes help traditional search performance too, not just AI visibility.
Reflection
What stuck with me most is how much GEO overlaps with just… writing well. A clear answer up top, real numbers instead of vague claims, sources backing up what you say, that’s stuff that makes a page better for a human too. AI answer engines just seem to notice and reward it faster than Google ever did.
What I didn’t expect was how much the tactic mattered per post. I went in thinking one approach would work everywhere. It really didn’t. Citations did more for the factual pieces, numbers carried the regulatory and opinion content, and quotes did the heavy lifting on anything narrative. Matching the tactic to the piece in front of me ended up mattering more than any single trick on its own.
If I’m being honest, I still feel a bit blind operating in GEO. There’s no clean dashboard that tells you a citation in ChatGPT actually drove a lead, and that lack of hard, concrete measurement is a big reason some people write the whole thing off as snake oil. I get the skepticism.
But I don’t think it’s snake oil, I think it’s just genuinely hard to measure right now, and that’s a different problem. I believe in it regardless.
It’s something that needs to be done and can be done, and there’s real merit in the work even while the measurement side catches up. To be clear, it’s a supplement to SEO, not a replacement for it.