AEO Explained

How the AI Recommendation Score is calculated: the five categories explained

We touched on this briefly in the post comparing the AI Recommendation Score to Google reviews, but it's worth a full explanation on its own, because the weighting behind the score isn't arbitrary. It's built around one core idea: the outcome that actually matters is whether AI names your business today, and everything else in the score is a leading indicator of that outcome, weighted according to how strongly it tends to drive it.

The five categories

The score is built from five weighted categories, combining into a single 0 to 100 number and letter grade.

CategoryWeightWhat it measures
AI Recommendation Presence35%Does AI actually name the business across real, customer style prompts, tested directly across ChatGPT, Gemini, Claude and Perplexity
Website AI Readiness20%Can AI parse what the business does, where it operates, and who it's for, from the site itself
Brand Authority20%Reviews, press mentions, directory listings and citations AI treats as trust signals
Content Quality15%Depth and specificity AI can cite as evidence of genuine expertise, not generic marketing copy
Technical Health10%Baseline crawlability, page speed, and whether the site is properly indexed

Why AI Recommendation Presence carries the most weight

At 35%, this is the largest single category, and deliberately so. It's the only category that measures the actual outcome rather than an input that tends to produce it: does ChatGPT, Gemini, Claude or Perplexity name your business by name when a real customer style question is asked, across a fixed panel of prompts covering your category and area. A business can score reasonably well across the other four categories and still underperform here, which is exactly the gap the score is designed to surface rather than average away.

A business with excellent Website AI Readiness and strong Brand Authority but weak AI Recommendation Presence isn't a contradiction. It usually means the underlying signals are solid but something specific, a missing content angle, an exclusion pattern on one platform, a gap versus a competitor on the exact prompts customers use, is stopping those signals from converting into an actual named recommendation. That gap is diagnosable, and it's usually the highest priority fix.

Website AI Readiness and Content Quality: related but distinct

These two categories are often confused with each other. Website AI Readiness is structural: can a model actually parse the site at all, is content locked behind JavaScript rendering or booking widgets, is schema markup present and accurate, is core information like pricing and hours in readable text. Content Quality is about what's there once a model can read it: is it specific and factual, or generic and adjective heavy. A site can be technically readable and still score poorly on Content Quality if the copy itself doesn't give a model anything concrete to cite.

Brand Authority is broader than review ratings

Reviews sit inside this category, but they aren't the whole of it. Press coverage, accreditations, directory and citation consistency, and third party mentions all contribute. This is also the category most connected to the traditional SEO and reputation work most businesses already do, which is why a business investing seriously in reviews and PR often sees this category move first, even before any AI specific work begins.

Why Technical Health is weighted lowest, not ignored

At 10%, this is the smallest category, but it functions as a floor rather than a growth lever. Genuinely poor crawlability or a broken indexing setup can suppress every other category's potential, because a model that can't reliably access the site can't act on strong content or accurate schema either. Most businesses clear this bar without dedicated effort, which is why it carries less weight than the categories that differentiate a strong result from an average one, but it's still worth checking rather than assuming.

What this means for where you focus

The weighting is a guide to prioritisation, not a strict formula for effort. A business with a badly broken Website AI Readiness score should generally fix that before investing further in Content Quality, because the second category's improvements are wasted if a model can't parse the site to find them. A business already strong on the structural categories but flat on AI Recommendation Presence should look directly at what real customer prompts return today, because that's the category actually measuring the result that matters.

See your own score, broken down by category.

Our full report shows exactly where you stand across all five categories, with real AI conversation evidence and a ranked fix list. Start with a free AI Snapshot.

Get my free AI Snapshot