We test real prompts, not imaginary benchmark queries
The loop starts from the commercial and category queries your buyers actually use, then compares your brand against the competitors being named instead.
Methodology
How the weekly loop works.
The methodology answers a simple question with repeatable evidence: why are AI systems recommending other brands instead of yours — and is that changing?
We check infrastructure before recommending more content
Blocked crawlers, missing Bing indexing, weak schema, rendering problems, or contradictory directives can stop visibility long before content quality becomes the issue.
We separate extractability from authority
Some brands are technically accessible but hard to extract. Others are clear on-site but absent from the off-site source pool that AI systems rely on for corroboration.
We state uncertainty directly
AI citations are non-deterministic. The loop improves the inputs that raise citation probability, but it does not promise a fixed ranking position or guaranteed inclusion.
Why fixed prompts matter
Same prompts, every week. That's what makes movement measurable.
If you change the questions between measurements, you can't tell whether your brand moved or the questions did. Fixed prompts are what turn weekly re-measurement into a trend rather than a series of unrelated snapshots.
30 prompts
Buyer-intent queries defined with you at onboarding
5 platforms
ChatGPT, Claude, Gemini, Perplexity, Google AI Overview
Every week
Same prompts, same platforms, on a fixed schedule
Prompt set stability
The 30-prompt set is held stable for comparability. It is reviewed quarterly, or when your category, products, or positioning change materially. Any change is logged and the affected baseline is reset — trends are never silently broken.
What We Check
For a channel-specific example of the same access, understanding, and authority workflow, see our guides to diagnosing Gemini visibility gaps and optimizing for Perplexity Search. If you are building your own baseline first, use the AI search visibility tracking playbook and the Perplexity SEO checker workflow before comparing movement week over week. Movement is reported using two metrics: Share of Model (any brand appearance) and citation share (URL-attributed citations only).
Access
AI crawler directives, Bing indexation, sitemap health, rendering behavior, and machine-readable signals.
Understanding
Whether pages make your category, use case, proof points, and qualification cues easy to extract and cite.
Authority
Which third-party pages, reviews, forums, and comparison sources are doing the recommendation work in your category.
Measurement and what we promise
The baseline is frozen. The trend is directional. The outcome is not guaranteed.
Frozen baseline
Week 1 is preserved unchanged as the comparison point. Every subsequent re-measurement is read against that frozen snapshot, so movement is always relative to where you actually started.
Directional trend
AI answers are variable — the same prompt can shift week to week on its own. We report the trend over multiple weeks, not single-week swings, so normal AI noise isn't mistaken for movement.
No outcome guarantee
We commit to output metrics: prompts tested, reports delivered, assets produced. Outcome metrics — citation rates, appearances, share of voice — are observed and reported, not guaranteed. AI citations are non-deterministic.
What counts as movement: a change between the frozen baseline and a later re-measurement in whether or how your brand appears across the fixed prompt set — appearing in responses where it previously did not, being cited as a source, or being named alongside or instead of a competitor. Movement is reported directionally over the repeated prompt set, never as a guaranteed or precise ranking.
See how the loop starts
Week 1 is the baseline audit. The loop runs from there.