Methodology v1.0.0 · effective 2026-09-01
The AI Recommendation Score
One number, from 0 to 100, describing how visible a business is across AI recommendation environments. It is built from six weighted components, and every one of them is counted from recorded AI answers.
The governing rule is short enough to state in one sentence: the score measures observed AI outcomes only.
Completing technical work does not raise it. Adding schema does not raise it. Publishing content does not raise it. Those are inputs we control, and they are measured separately by the Entity Authority and AI Readiness diagnostics. The practical consequence is that a month of genuine work can end with a flat score — and a flat score is what gets reported.
The six components
Weights are fixed and published. They are not tuned per client, and they are not adjusted to make a result look better.
| Component | Weight | What it counts |
|---|---|---|
| Presence Rate | 40% | Of all observations in the period, the share in which the business was mentioned at all. |
| Top-1 Recommendation Rate | 20% | Of all observations, the share in which the business was the first business recommended. |
| Top-3 Recommendation Rate | 10% | Of all observations, the share in which the business appeared in the first three businesses recommended. |
| Citation Rate | 15% | Of all observations, the share in which the AI answer cited a source the business controls or is named on. |
| Accuracy Rate | 10% | Of mentions that contained checkable factual claims, the share in which those claims were correct. |
| Positive Recommendation Rate | 5% | Of mentions, the share framed positively or strongly positively rather than neutrally or negatively. |
Presence carries the most weight because everything else is conditional on it: a business that is never named cannot be ranked first, described accurately or recommended warmly. Full denominator definitions are in the methodology.
How the number is built
Each component is a rate between 0 and 1, multiplied by its weight, summed, and expressed out of 100. There is no curve, no adjustment and no discretionary component.
The example alongside is a worked calculation, not a measurement of any real business. Greenbug has not yet collected live observations, and we would rather show you the arithmetic than invent a result.
Score = (Presence × .40) + (Top-1 × .20) + (Top-3 × .10)
+ (Citation × .15) + (Accuracy × .10)
+ (Positive × .05)
Worked example — not a measurement of any business
42.8
/ 100
EstablishedMeaningful visibility but not dominant.
How the number is built
| Component | Rate | Weight | Points |
|---|---|---|---|
| Presence Rate | 50% | 40% | 20 |
| Top-1 Recommendation Rate | 12.5% | 20% | 2.5 |
| Top-3 Recommendation Rate | 25% | 10% | 2.5 |
| Citation Rate | 25% | 15% | 3.8 |
| Accuracy Rate | 100% | 10% | 10 |
| Positive Recommendation Rate | 80% | 5% | 4 |
| Total | 42.8 | ||
Based on a business mentioned in 120 of 240 observations, ranked first in 30, in the top three in 60, cited in 60, factually correct in every checkable mention, and framed positively in 96 of its 120 mentions.
Score bands
Bands are published with integer bounds, but scores carry one decimal. A band owns everything from its minimum up to the next band’s minimum, so 39.9 is Emerging rather than falling between two bands.
| Range | Band | What it means |
|---|---|---|
| 0–19 | Minimal | Rarely appears in tested AI recommendations. |
| 20–39 | Emerging | Appears inconsistently. |
| 40–59 | Established | Meaningful visibility but not dominant. |
| 60–74 | Strong | Frequently recommended. |
| 75–89 | Leading | One of the strongest businesses tested. |
| 90–100 | Dominant | Exceptional recommendation visibility across platforms and query types. |
What the score is computed from
A fixed benchmark, run the same way every cycle. The prompts do not change between cycles — that is what makes month-to-month movement mean something.
24
frozen benchmark prompts
Built from the language buyers in your category actually use
5
AI platforms
The full benchmark set, listed below
2
runs per prompt
Because AI answers are non-deterministic
240
recorded observations
Per business, per cycle, each stored in full
Benchmark platforms
- ChatGPT
- Gemini
- Perplexity
- Claude
- Google AI Mode
Partial coverage is not the score
The Greenbug AI Recommendation Score requires all 5 benchmark platforms. A measurement covering fewer produces a Platform Score — for example a Perplexity Recommendation Score. That is a real, defensible measurement of one platform’s behaviour, computed with the same weights and denominators.
It is not the Greenbug AI Recommendation Score and will never be presented as one. Any surface showing a partial measurement displays platform-by-platform coverage alongside it.
Current status: no live measurements published
What the score cannot tell you
A number this compact hides things, and the honest way to publish one is to say what it hides.
- It is a sample, not a census. It estimates recommendation behaviour from a fixed set of questions; it does not enumerate every answer a customer could receive.
- It says nothing about revenue. Being recommended more often is not the same as being hired more often, and we do not claim the link.
- It is bounded by the prompt library. A business could be highly visible for questions we did not ask.
- It can move for reasons that have nothing to do with you, such as a vendor shipping a model update. Known platform changes are logged against the period.
Find out where you stand
A baseline gives you the score, the six component rates, the competitors who appeared instead, and the evidence behind every figure.