How the score is built
Answerlord measures AI visibility by asking ChatGPT, Perplexity, and Google AI Overviews 5 live buyer-intent questions about a domain's niche, then scoring 0–100 from four measured signals: whether the domain is cited at all, across how many reachable engines, how prominently, and how favorably. Engines we can't reach are excluded — never guessed.
The AI Visibility Score is not a black box and it is never guessed. It's a weighted blend of four signals we measure directly from what ChatGPT, Perplexity, Google AI Overviews actually say when asked 5 real buyer-intent questions about your niche.
Live answers, not a recycled corpus
Most AI-visibility products work from a fixed library of generic questions, replayed against the engines on a weekly or monthly cycle, and then let you search the stored answers. That's useful for trend lines — but the answer you're shown can be weeks old, and the questions were never written for your business.
Answerlord does the opposite: every grade generates 5 buyer-intent questions for your specific niche and runs them against the engines live, at the moment you ask. What you see is what a buyer asking today would actually be told — including which competitors the engines name instead of you. The trade-off is honest: we sample 5 focused questions live rather than thousands from a warehouse, because for a business owner the question that matters is “what do the engines say about my niche right now?”
Cited at all
30 pts maxAre you cited by at least one AI engine across the five buyer-intent prompts — the line between existing and not existing to a robot.
Engine coverage
30 pts maxWhat fraction of the reachable engines (ChatGPT, Perplexity, Google AI Overviews) cite you, counting only engines we actually got data from.
Prominence (share of voice)
20 pts maxWhere you land in the answer — being named first earns full credit; being named further down (after several rivals) earns less. If the rank can't be read from the text, we fall back to your share of voice rather than guess.
Sentiment
20 pts maxWhen an engine mentions you, whether it speaks positively, neutrally, or negatively about you.
The formula
Each signal is normalised to a value between 0 and 1, multiplied by its weight, and summed, then scaled to 100:
score = 100 × ( 0.3 × citedAtAll // are you cited anywhere? + 0.3 × engineCoverage // fraction of reachable engines citing you + 0.2 × prominence // how high up the answer you rank when cited + 0.2 × sentiment // tone toward you where mentioned )
Sentiment maps to: positive = 1, neutral = 0.6, negative = 0.2. Prominence is rank-based: named first = full credit, then it tapers the further down the answer you appear; when the order can't be read, we fall back to your share of voice rather than guess.
The letter grade
We also translate the 0–100 score into a plain letter grade, so you can read it at a glance. It's the same number, just bucketed:
- A90–100
- B80–89
- C70–79
- D60–69
- F0–59
The rules we never break
- No fabricated metrics. If an engine can't be reached, it's marked unavailable and excluded from the denominators — we never invent a value to fill the gap.
- No data, no number. If every engine is down, the score is unavailable, not zero and not a guess.
- Coverage counts only what we reached. An outage on our side never lowers your score — coverage is measured against engines that actually returned data.
- Gaps come from real misses. The 3 gaps we show point at a specific engine, prompt, or competitor — never generic advice.
- Results are cached 24h per domain + niche, to keep this free and to keep the engines from being hammered.