Methodology

Blind measurement

We ask AI assistants for recommendations in a niche without naming the brand or its domain in the query. Only when a model surfaces a brand on its own do we count it as an organic citation.

This is intentionally stricter than tools that ask AI directly about a brand. A brand that only appears when prompted by name has no real AI visibility – it has presence in the model's vocabulary, which is a different thing.

Each keyword is queried multiple times. We measure consistency, not single answers – a model that names you once in ten attempts is not the same as one that names you all ten.

The AI visibility score

The daily score (0–100) is composed of six factors. The weighting may change as we gather more data – the factors themselves and their definitions are stable.

• Citation Frequency – share of queries that mention the brand at all.
• Citation Breadth – share of AI providers that mention the brand. Measures distribution rather than volume.
• Recommendation Strength – how actively a citation is framed as a recommendation, from neutral mention to explicit top pick.
• Position – where in the response the brand appears. Earlier mentions count more than later ones.
• Sentiment – tone of the mention (positive/neutral/negative).
• Trend Momentum – change in citation rate against the previous period.

Quality scaling

Recommendation, Position and Sentiment are scaled by the square root of the citation rate. This prevents brands with one very positive mention from dominating the score. Less data produces a more cautious score – not a better one.

Hallucination filter

AI models invent brands. Before competitor profiles are stored, we run them through two independent checks: does the domain exist, and is the website reachable?

Profiles that fail both checks are dropped. This reduces but does not eliminate the visibility of fabricated competitors – domains that exist but have nothing to do with the niche can still slip through. We are working on it.

Search Console fusion

Brands can connect their Google Search Console to Achtung.app. We read clicks, impressions, CTR and position per query. The connection is optional and separate from tracking – a brand can be tracked fully without GSC.

We use GSC for one thing that does not work without it: detecting AI Overview signals. Queries with high impressions but low CTR are often queries where Google itself is producing a generative answer – and therefore queries where classical visibility no longer suffices.

GSC data never leaves the system. We do not publish brand-specific GSC numbers, neither inside another team's brand portal nor in the public field notes.

Market signals: news and competitors

Achtung.app collects news daily via Brave News and X/Grok for the brand and its detected competitors. Each article is scored by an LLM for relevance and potential business impact before it surfaces as an alert.

The score is an approximation, not a verdict. An article with a high relevance score is one we believe deserves attention – not a guaranteed business threat.

What we do not yet measure

This list is intentionally public. It changes when we add something – and it is the most honest answer to how far the product is today.

• Forecasting future visibility. We observe trends, we do not predict them.
• Direct revenue attribution. We do not automatically connect visibility to clicks, leads or revenue. It is on the roadmap.
• Cross-brand market reports. We cover individual niches, but do not yet produce vertical market reports.
• Recommendation workflows. Insights today are text, not trackable tasks.
• Complete hallucination detection. The filter catches obvious inventions but not all of them.

What we commit to

• No number without a sample size. If we name a percentage on a public page, the sample size sits next to it.
• No invented statistics. Our automated generators reject numbers that do not come from real snapshots.
• Published field notes are not silently overwritten. We correct, we do not delete.
• Methodology changes are documented here, not buried in a release note.