How We Score AI Visibility: Pondral's AEO Methodology Explained
Pondral breaks the score into five signals across five engines and shows the math behind each one.
A single number with no way to tell if 62 is good is not much use. A methodology is only valuable if you can see inside it. Pondral's AEO methodology is built on five measurable signals, tracked across five AI engines, and designed to tell you exactly what's working and what needs attention.
Why AEO methodology matters
If you can't see how a score is computed, you can't trust it or improve it.
If you don't know what's being measured, you can't improve it. You're optimizing blind. The difference between a black-box score and a transparent methodology is the difference between knowing you're losing and knowing where and why.
Transparency builds trust. When you can see the formula, the signals, and the reasoning, you can make confident decisions about your AI visibility strategy. You're working with a framework you understand and can validate against your own data. (If you're still sorting out where AEO fits alongside SEO, start there. The methodology makes more sense once the distinction is clear.)
That's why we built Pondral's methodology to be completely open. Each signal and weight is documented, and each score links to the stored answer it was graded from.
The five signals we measure
Pondral's AEO score is built from five signals, each measuring a different dimension of how AI systems perceive and present your brand.
Presence
Does your brand appear in the AI response at all? This is binary, yes or no, but it's the foundation of everything else. You can't win on prominence or citations if you're not mentioned at all. We track presence per query, per engine. The more queries where you show up, the broader your coverage across AI engines.
Prominence
Where you appear in the response matters. The AI mentions you in the first sentence? That's strong. Buried in the supporting details? Weaker. An answer names a limited number of sources, so position in that list is competitive. We also separate being named outright from being referred to only indirectly. We measure first-mention position and weight earlier, explicit appearances more heavily than later or implied ones.
Context
Being mentioned isn't enough if the engine damns you with faint praise, or gets the facts wrong. Context grades how well your brand is described: how it is portrayed (recommended outright, mentioned neutrally in passing, or actively warned against), whether what the engine says about it is factually accurate, and whether it is placed in the right category for the question. A glowing recommendation built on stale claims, or a brand framed in the wrong category, is capped accordingly. An LLM rater reads the text around each mention, judges all three, and combines them into a five-point grade from recommended to criticised.
Citation Link
Being named is good. Being cited, with a URL linking back to your content, is better. When Claude answers a question about your category and links to your own pages, that's a citation link. When it just names your brand without a link, that's a mention. A link points the reader to your own pages, so we score a link to your own domain higher than a bare mention. To be exact about what this factor does and does not do: it checks for a link to a domain you own. It does not detect quotes, so being quoted with attribution but no link scores the same as a bare mention.
Competitive Presence
How much of the brand conversation is yours within a given AI answer. If a response to a category question names five brands and one of them is you, your competitive presence in that answer is low, even if that single mention is strong. We measure your share of the brands named in the same answer, against the competitors that appear alongside you, then aggregate across your full query set. Competitive presence tells you whether you're the default reference or the also-ran.
| Signal | What it measures | Why it matters | Weight |
|---|---|---|---|
| Presence | Brand appears in the response (yes/no) | Foundation for all visibility; coverage across queries | 20% |
| Prominence | Where in the response you appear, named outright vs. only implied | Earlier, explicit mentions carry more weight; limited citation real estate | 25% |
| Context | Whether you're described well: portrayed favourably, accurately, and in the right category | Checks that a mention frames you correctly, not just that it exists | 20% |
| Citation Link | A link to your own domain vs. just a name | Shows whether the engine points readers to your own site | 20% |
| Competitive Presence | Your mention share vs. competitors in the same answers | Tells you if you're dominant or marginal in your space | 15% |
How we calculate the AEO score
The score isn't one number. It's a weighted composite of five signals, calculated per query, aggregated per theme, then rolled up to your overall AEO score.
On paid plans, every time we run a query across ChatGPT, Claude, Gemini, Perplexity, and Grok (the free check runs two of them), we measure all five signals for your brand. That gives us a per-query signal set. We weight them according to the public rubric: prominence counts the most, at 25 percent, with presence, context, and citation links each at 20 percent and competitive presence at 15 percent. That produces a score for that query on that engine.
Next, we aggregate across all queries in a theme. Say your theme is Global Positioning and Scale. We look at all the queries tagged with that theme, calculate the median or mean performance, and that is your theme-level score. Themes roll up to overall.
Visibility varies by theme and engine. You can lead one theme and trail another, or do well on ChatGPT and poorly on Perplexity. The granular breakdowns let you optimize where it matters. The overall score gives you a headline number for board meetings, but the real strategy is in the details.
What the score actually tells you
Your AEO score ranges from 0 to 100. But what does that range actually mean?
0-30: Critical Gap. You're barely visible. Most competitors show up more frequently, get cited more, or rank higher in responses. In competitive spaces, established players may already be ahead. There is room to move, and the work is real.
30-50: Emerging. You're in the conversation, but not yet the go-to. You're showing up in some queries, getting some citations, but not consistently across the board.
50-70: Established. You're a recognized player. Most major queries mention you. You're getting cited regularly. Competitors know about you. You're likely sharing the top spots with one or two others. This is the professional tier: you're taken seriously by AI systems and the humans relying on them.
70-85: Strong. You're the default reference. When someone asks a question in your space, your brand shows up first or second. You have strong citation presence. Most competitors rank below you.
85-100: Dominant. You're the authority. You're cited in the majority of responses. You appear across all major queries and themes. Competitors are chasing you. This requires sustained investment and ongoing content strength.
The score itself isn't the point. The breakdown is. Pondral shows you which signals are holding you back. Maybe your presence is strong but your citation links are weak: that means your content isn't getting linked. Maybe your overall score is rising but your context is turning critical: that's a red flag that needs investigation. The breakdown shows which signals are weak.
How we're different from other scoring tools
A methodology you can't see is a black box: a number with no explanation of how it's calculated. That doesn't help you act on it. We show you exactly what we measure and why. You can disagree with the weights or the signals, but you know what you're looking at.
On paid plans we track five engines: ChatGPT, Claude, Gemini, Perplexity, and Grok (the free check runs two). Why? Because your visibility in ChatGPT doesn't guarantee visibility in Claude. Your ideal customer might be using Perplexity. Tracking all five gives you a real picture of your presence across AI engines.
A single headline score hides the detail. We break it down by theme, by competitor, by engine, by signal. You see exactly where you're winning and where you're weak.
And we show you the raw responses. You can read exactly what the AI said about your brand, in context, with citations visible. You're not trusting our interpretation. You're seeing the data yourself.
We show the math so you can validate it against your own data.
Frequently asked questions
How often should I run an AEO audit?
It depends on whether you are measuring once or continuously. Any single run is noisy: ask the same question twice and the answer can differ, so one run is a sample rather than a verdict. That is an argument for repeating the same question set on a schedule and reading the trend, not for measuring rarely. On paid plans, Pondral runs a full audit across all five engines weekly and charts the trend, which is what makes run-to-run variation visible instead of hiding it inside one number. If you are auditing by hand instead, quarterly is a sensible floor for catching content and competitive shifts, and monthly is worth it during a major content push. The thing that does not work at any cadence is reading a single run as a precise score.
Can I compare my AEO score to competitors?
Partly. Pondral scores your brand, not your competitors' brands. The Competitive Presence signal measures your share of the brands named in the same answers, and the competitor breakdown shows which competitors appear alongside you and how often. That shows you the answers where they appear and you don't.
What is a good AEO score?
That depends on your space, and we don't publish a benchmark for each category. The more useful comparison is with the brands that appear alongside you in the same answers, which is what the Competitive Presence signal measures. Your position relative to them matters more than the absolute number.
Does the AEO score account for different industries?
Yes, in the part that should change, and deliberately not in the part that shouldn't. The scoring rubric is identical for every industry on purpose: that's what makes a 60 mean the same thing for a bank and a law firm, so you get an apples-to-apples comparison. What we tune per industry is the questions we ask: which question types dominate, how many discovery questions we include (the ones where a buyer doesn't know your brand yet), who counts as a competitor, and which sources count as authoritative. A patient and a software buyer search in completely different ways. Same rubric, industry-specific questions.
How quickly can my AEO score change?
We don't publish a timeline, because we haven't measured one. AI answers vary from one run to the next, so a single week's change is not by itself evidence that anything you did worked: changes under 15 points are directional. Read the trend across several weekly runs rather than any one reading.
The methodology is the strategy
You can't improve a score you don't understand. That's why we built Pondral's AEO methodology to be completely transparent. Your strategy depends on knowing exactly what you're optimizing for.
The five signals (presence, prominence, context, citation links, and competitive presence) capture the full picture of how AI systems see your brand. The weights are published and fixed. The granular breakdowns let you identify gaps and opportunities. And the multi-engine coverage shows you each engine separately, so a strong result on one doesn't hide a weak result on another.
Ready to see how you stack up? Run a free check to see where you stand.
Philipp GroubiiFounder, Pondral
Philipp builds tools that help brands understand and improve their AI visibility. Background in SEO strategy, digital marketing, and SaaS product development. LinkedIn →
Stay updated
Get new articles and occasional updates on AI visibility delivered to your inbox.
No spam, ever. Unsubscribe anytime.