Last updated: 16 July 2026
We score your AI visibility with one number, the SCR score (0–100). SCR stands for Surfaced, Cited, Recommended. We run at least 30 buyer prompts across five AI engines, score each answer, and weight recommendations highest. Then we fix gaps and retest. First real movement usually lands in six to eight weeks.
Why bother? G2 reports that half of B2B software buyers now start their research with AI chatbots (2026). Gartner forecasts that by 2026, search engine volume will drop 25% as users turn to AI chatbots and virtual agents. If the answer engines don't name you, you're not on the buyer's list.
An AI visibility audit checks whether AI engines mention, cite, and recommend you when buyers ask about your category. We run a fixed eight-step loop, from brand and entity checks through to a retest. It's measurement first, then targeted fixes, then proof the score actually moved.
The SCR score is a single 0–100 number for how visible you are in AI answers. SCR stands for Surfaced, Cited, Recommended. We score every prompt on each engine, turn each signal into a percentage of prompts, weight them, then average across all five engines.
For every prompt on every engine, one of four things happens:
| Signal | Meaning | Per-prompt score | Weight |
|---|---|---|---|
| Absent | You don't appear in the answer at all | 0 | — |
| Surfaced | Your brand is mentioned in the answer | 1 | 30% |
| Cited | You're named as a source the answer relies on | 2 | 30% |
| Recommended | You're in the top-3 "best X" shortlist | 3 | 40% |
Each signal becomes a percentage of your prompts. If you're surfaced in 60% of prompts, cited in 40%, and recommended in 20%, those three percentages feed the formula. Recommendation carries the most weight, because being on the shortlist is what wins the deal.
SCR = (Surfaced% × 0.30) + (Cited% × 0.30) + (Recommended% × 0.40)
We calculate that for each of the five engines, then average the results into one 0–100 score. Higher means you show up more often, in the ways that matter most.
Share of Voice is the share of your money-prompts where your brand shows up at all, measured against named competitors. If you appear in 12 of 30 buyer prompts, that's 40%. It tells you how much of the AI conversation you own, and who's taking the rest.
SCR tells you how well you show up. Share of Voice tells you how often, compared to the other names in your category. We track both, because you can have a decent SCR on the few prompts you win while a competitor quietly owns the other twenty. Seeing them side by side is how you decide where to fight.
We measure five engines: ChatGPT, Perplexity, Gemini, Google AI Overviews, and Copilot. For each buyer segment we use at least 30 money-prompts, the real questions people ask before they buy. Everything is run, not scraped by a tool, so we see the exact answer a buyer sees.
Money-prompts aren't keywords. They're full buyer questions like "best [category] tool for a 20-person sales team" or "[your brand] vs [competitor] for SOC 2 compliance". We write them from your actual sales conversations, not from a volume report. Thirty is the floor; deeper segments get more.
Our SCR™ engine puts the prompt to the answer engine live and captures what actually comes back, including the sources it leaned on. A person then reads it before it becomes a number. Automated trackers report a cached or generic answer, which is how a brand can look present in a dashboard and be absent in the room.
First real movement usually shows around six to eight weeks in. Engines re-index at different speeds. In our experience Perplexity refreshes fastest, ChatGPT is slower, and Google AI Overviews is slowest of the three. So we measure at the start, fix, then retest on that cadence rather than chasing daily noise.
These are our observations, not guarantees. Some fixes land faster, some slower, and engines change their behaviour without warning. That's exactly why we retest with the same prompts on the same engines: it's the only honest way to tell whether the number moved.
Four things, measured the same way at both ends: where you started, where you are, how long it took, and how you sit against the competitors we agreed to track. Same questions, same engines, same method. One table your CMO can read in a minute.
No. Engines change, competitors move, and nobody controls the model. We guarantee the work, not the numbers: the agreed volume of work, and a full re-measure across all five engines at the end, run exactly the way the first one was.
No. SEO asks where you rank on a page of links. SCR asks whether an AI engine mentions, cites, or recommends you inside its answer, where there's often no list of ten blue links at all. The work overlaps in places, but the target is different.
No. We run our own SCR™ engine on the back end and a person reads every answer it collects — there's nothing for you to log into. It's one engagement: the audit that measures, and the ninety days that move it. We don't sell the audit on its own.
Yes. That's the point of publishing it. Ask on the call and we'll walk you through our current SCR, what we're testing this month, and how the same loop would apply to your category.
The fastest way to know where you stand is a short conversation about your category, competitors, and the prompts your buyers use. Book a 15-minute call