If you have read this far in our research series, a reasonable question is bubbling up. The signals that AI bots use to decide what to cite are not secret. We have published them. The technical specifications for JSON-LD are publicly available at schema.org. The relevant robots.txt directives are documented by OpenAI, Anthropic, and Google. Every step is open.
And yet only 0.4 percent of the brands in our dataset have crossed the AEO score 8 threshold that AI bots reward most. The other 99.6 percent have not done the work. Why?
The short answer is that AEO work is the kind of work that loses to almost every other priority on a marketing or development backlog, even when the team agrees it should be done. The slightly longer answer is more interesting, and it has three parts.
Implementing AEO is mostly invisible to the team that ships it. The schema is in the HTML head. The llms.txt is at the root. The robots.txt rules are buried at the bottom of a config file. None of these surface in the browser. A marketing leader who walks the site after AEO work is complete sees nothing different. The site looks the same. The colors are the same. The copy is the same.
Compare that to a redesign, a new landing page, or a brand campaign. Those are visible. People can point to them. They get screenshots in board decks. AEO does not. It is the kind of work that produces results only in dashboards and traffic logs months after it ships, which makes it hard to fund and easy to defer.
Brands that get to AEO score 8 usually have one of two things going for them. Either the owner personally cares about being cited in AI answers and overrides the marketing-vs-engineering tension to ship the work, or the team has a credible measurement loop that lets them see the results before the budget cycle ends. Most brands have neither.
Correctly implementing JSON-LD schema across an entire site is not difficult, but it is unforgiving. A single misplaced comma in a JSON-LD block invalidates the entire markup. A missing required field on an Article schema causes Google to silently drop the structured data without an error. Most CMS plugins emit only a subset of the schema that AI engines actually evaluate.
We see this constantly in audits. A brand has Yoast or RankMath installed, the plugin says "schema enabled," and the actual HTML output has Organization schema but no FAQPage schema, no Article author bylines, and no Product schema for the e-commerce SKUs. The team is doing the work, but the output is incomplete in ways that are not obvious from inside the plugin's UI.
The fiddliness compounds with the second-order problem: schema validation tools exist, but most teams do not have them wired into their build pipeline. So errors that would be caught by a CI check go unnoticed for months. By the time someone audits the site, the structured-data signals have been broken since the last theme update and nobody knew.
AI bots update their crawl and citation behavior on their own schedules. OpenAI added a new ChatGPT-User bot in 2024. Anthropic introduced Claude-Web. Google split Googlebot and Google-Extended to separate AI training from search indexing. Apple shipped Applebot-Extended. Each of these changes requires a corresponding update to a brand's robots.txt or it silently regresses.
A brand can score 8 today, ship no new code for six months, and find itself back at score 6 because an AI bot operator updated their crawler identity and the brand's robots.txt now blocks the new bot's allowed counterpart. Ongoing monitoring is not a nice-to-have. It is the difference between sustained AEO performance and a one-time push that decays.
Each of these reasons on its own is a friction. Combined, they produce a structural barrier to AEO adoption that does not exist for traditional SEO. Most brands will not do the work. Some will do part of the work and find that part of the work is worse than no work at all because broken schema is sometimes a stronger negative signal than missing schema.
The brands that get to score 8 have either decided to fight all three frictions internally or have outsourced the work to a service that handles them. Either path is a small minority of the market. That minority is the 0.4 percent at the top of the AI crawl distribution. Their advantage is not that they know something the rest of the market does not. Their advantage is that they did the work the rest of the market keeps deferring.
If you are reading this and recognizing your own brand in the 99.6 percent, the question is not whether to do the AEO work. The data says you should. The question is whether to do it yourself or hire it out.
If your team has a clean build pipeline, structured-data validation already wired in, and a person who personally cares about AI visibility, doing it in-house is rational. The cost is a focused week and an ongoing 30 minutes a month for monitoring.
If your team does not have those things, the math usually points to using a service. Engagemii's free AEO score gives you the audit and the fix list. The $29 audit gives you the branded PDF with the complete plan. The $99 Fix-It Kit deploys the schema, llms.txt, and robots.txt updates for you, plus a month of citation monitoring. The pricing reflects how much of the friction we are eating on your behalf, which is most of it.
The market window for being one of the 0.4 percent will not stay open forever. Eventually large enterprises will catch up. Eventually the structured-data work will be a default rather than a differentiator. But that has not happened yet, and the data says it will not happen this year. The brands that move now own citation slots they will keep for the next several years.
This article is the long-form treatment of Section 3.5 of Engagemii's research brief. The 0.4 percent figure is the share of crawled brands in our directory snapshot that score AEO 8 or higher. Full methodology at engagemii.com/research/aeo-crawl-drivers.
If you want to cite this article, the URL is engagemii.com/blog/why-most-sites-wont-fix-aeo.
Ready to find out if AI can cite your brand?
Get Your Free AEO Score