Methodology
How SeeRanked measures AI visibility
SeeRanked measures where a business appears in AI answers, for the questions its customers ask, in each city it serves, on five platforms. Here is how a check runs and how the score is built.
What we measure
For the questions your customers ask, in each city you serve, on ChatGPT, Claude, Gemini, Grok and Perplexity.
Whether an AI assistant names your business when a customer asks
Its position among the businesses the answer names
Which competitors are named next to you
How warmly the answer describes you
How a check runs
Prompt by prompt, city by city, platform by platform.
One check is one prompt, in one city, on one platform. A campaign with 10 prompts, 3 cities and 5 platforms is 150 checks.
- 1
You set up the campaign. Your business, the cities you serve, the questions your customers ask AI (prompts) and the competitors you watch.
- 2
We ask every prompt in every city on every platform. The city is part of the question, the way a local buyer would ask it.
- 3
Every answer is read and scored. Whether your business is named, its position among the businesses the answer names, which competitors are named, and how warmly you're described.
- 4
The results become your report. The AI Visibility Score, your ranking on each platform, your position against competitors, sentiment and share of voice for every prompt, the answers themselves, and recommendations on what to fix first.
The two report modes
How the AI answers are collected, by plan.
Deep Research
Every paid plan- Answers collected
- Real browser sessions on each platform
- Review
- A human analyst reviews the report before it's published
- Delivery
- Emailed when it's ready
Quick Report
The Free plan- Answers collected
- Each platform's official API
- Review
- Automatic, with no human in the loop
- Delivery
- Ready in about 2 minutes
Every scheduled refresh is a Deep Research run.
How the city is applied
A city is a setting, not a GPS location. The city is part of every question we ask, and where a platform lets us set the asker’s location, we set the same city. So each answer is the one a buyer in that city would get, not an answer from nowhere.
How the score is built
The same maths on every report.
- Position in one answer1st place scores 100. 10th place, or not named at all, scores 0. Places in between fall on a straight line: 2nd is 89, 3rd is 78, 5th is 56.
- Per platform, per cityYour average position across all prompts, turned into that 0 to 100 visibility. Not named counts as 10th.
- City scoreThe average of the platform visibilities, over the platforms that answered.
- AI Visibility ScoreThe average of your city scores, with the change since your last run.
- CompetitorsEach competitor's position is read from the same answers and scored the same way, so you can see who is ahead of you, platform by platform.
- Share of voiceFor one prompt, the share of the answers that name each brand. A brand named by four assistants owns more of the conversation than one named by one.
- Sentiment0 to 100: how warmly the answer frames you when it names you. 100 is an enthusiastic recommendation; around 60 is a neutral mention.
Sources
We show the answer each AI gave, with any links or sources it cites. We don't add sources the AI didn't cite, and we don't rank them.
How often it refreshes
Starter every month, Growth every 2 weeks and Scale every week. The Free plan runs once, on demand.
Limits, so you read the numbers right
Answers change. The same prompt can get a different answer an hour later. Compare runs over time, not one answer.
A city is a setting. We ask the way a local buyer would, and set the location where a platform allows it. No method can stand on a street corner in your city.
Names are matched by spelling. Variants of one brand are merged where we can tell; an unusual spelling may count separately.
Not every AI product. We don't track Google AI Overviews, Google AI Mode or Microsoft Copilot.
Measured, not promised. SeeRanked measures where you appear and recommends what to fix. It doesn't sell placements or promise rankings.
Frequently asked questions
How is the AI Visibility Score calculated?
From your position in each answer: 1st place is 100, 10th or not named is 0. We average across prompts per platform, across platforms per city, and across cities for the overall score.
Do you use real AI apps or APIs?
Both, by plan. Paid plans use Deep Research: real browser sessions on each platform, reviewed by an analyst. The Free plan uses Quick Report: each platform's official API.
How do you check a specific city?
The city is part of every question, and set as the asker's location where the platform allows it.
Why is my score different from last time?
AI answers change between runs, and your competitors' pages change too. The report shows the change since the last run, and keeps every answer so you can see why.
What counts as "recommended"?
Your business is in the top 3 businesses the answer names.