Reading your Sonar results

Data is useless unless you know how to read it. Sonar dashboard packs a lot of information into one screen: here's how to extract the insights that matter.

The visibility score card

The big number at the top of your dashboard is your AI Visibility Score: a weighted composite that factors in SOM, recommendation rate, model coverage, and sentiment. It's scored 0-100.

This differs from raw SOM. When interpreting results, distinguish the mention rate, recommendation quality, sentiment, and engine coverage. A hypothetical mention rate alone does not explain a composite score or prove customer value.

Think of the visibility score as your "GPA" and SOM as one of the grades that feeds into it.

Model breakdown

Compare sampled answers by engine and check successful sample counts. Uneven mention rates are observations to investigate; they do not prove that one engine's training or retrieval has caught up with another.

Keyword performance table

Use the keyword table to identify questions to investigate. Read a sampled rate in its own context rather than applying universal high, medium, or low tiers.

A zero sampled rate does not mean no source material exists. A middle rate does not predict the highest return from citation work. Choose a practical action using verified findings, relevance, and effort, then review results you can show at a comparable checkpoint.

Mentioned vs recommended

The results table distinguishes between mentions and recommendations:

• Mentioned means the AI included your brand name somewhere in the response. It might be a list of options, a passing reference, or a comparison.
• Recommended means the AI specifically endorsed or suggested your brand. Phrases like "I'd recommend," "a great option is," or "you should check out" indicate a recommendation.

Recommendations are worth more than mentions because they drive action. A user who sees "check out Brand X" is far more likely to visit your site than one who sees "companies in this space include A, B, C, and X."

Track your recommendation rate (recommendations / total mentions) as a quality metric alongside SOM as a quantity metric.

Understanding sentiment

Every brand mention is classified as positive, neutral, or negative:

• Positive: "Brand X is excellent for this" or "I've had great results with Brand X"
• Neutral: "Options include Brand X, Brand Y, and Brand Z"
• Negative: "Brand X has had reliability issues" or "I'd avoid Brand X for this"

Negative mentions are rare in AI responses (models tend to be cautious), but they happen, especially if negative reviews or critical press articles are in the training data. If you see negative sentiment, investigate the source and address it through review responses or PR.

Reading individual results

Click any result row to see the full AI response. This shows you:

• The exact prompt that was sent
• The complete AI-generated answer
• Where your brand appears in the response (highlighted)
• Which competitors were mentioned
• Source URLs cited (for Perplexity and Google AI Overviews)

Reading full responses is time-consuming but valuable: it shows you exactly how AI models position your brand relative to competitors. Do this for a sample of 5-10 results per monitoring run.

The weekly review rhythm

Here's how to use Sonar efficiently:

1. Monday: Check overall SOM trend (up, down, flat)
2. Tuesday: Review keyword performance: identify biggest gaps
3. Wednesday: Read 5-10 full responses to understand positioning
4. Thursday: Create action items: citation targets, press angles, entity fixes
5. Friday: Execute or queue those actions in the relevant modules

In Outrigger

The full results dashboard with model breakdown, keyword performance, and individual result detail is all in the Sonar module.