Shruwd
Shruwd

The AI visibility tracker that tells you when the number is real.

Weekly tracking of what ChatGPT and Google's AI Overviews say about you, with a range on every rate and a rule for when a change counts.

Start free See pricing

Two engines. Both measured the same way, every week.

ChatGPT

What ChatGPT tells people when it searches the web for an answer.

Google AI Overviews

The AI answer Google shows with its results, usually at the top.

Answers come through a data API, never from someone's logged-in session. They stand in for what a person sees, and Shruwd labels them that way.

The same questions, asked again, on a schedule.

The free plan is a single snapshot of your questions on both engines, not a schedule. Weekly tracking starts on Starter.

Four numbers, and what each one can and cannot tell you.

Mention rate

34%27%–41%

Mention rate

How often AI answers name you. Shown with a range, because it is measured, not guessed, and a measurement has a margin.

Share of voice

18%13%–24%

Share of voice

How much of the conversation is yours, against the competitors you track. Always shown next to mention rate, because share of voice also rises when a competitor falls, and that is not the same as you improving.

Prominence

62%

how early you appear

Prominence

Where you appear in the answers that name you, not just whether. Named first is worth more than named last.

Citations

Which sources the answers cite, and whether any of them are yours. That is what turns "you are not named" into "you are not named, and here is the page that got cited instead".

Citations feed your findings rather than becoming one more number to watch.

One thing you will not find

There is no blended visibility score. A single number that mixes four things cannot be argued with or debugged, and the components are what tell you what to do.

Too few answers? You get a message, not a number.

When there is not enough behind a number yet, Shruwd says so instead of printing a percentage that could mean almost anything.

A tool that shows "30%" from a handful of answers is not being more helpful than one that declines. It is just being confidently wrong.

Single questions take a few weeks to fill in. Your question set as a whole is ready much sooner.

Whole question set · mention rate

34%27%–41%

One question · week 2

Insufficient data

A few more runs needed

"This fix worked" is a claim somebody will check.

Before Shruwd calls something movement, the change has to be:

  1. 1

    Bigger than the noise. AI answers wobble from week to week on their own. A change has to stand out from that, even with all your questions checked at once.

  2. 2

    Big enough to matter. A tiny shift can be technically real and still not worth your time. Those are not reported as wins.

  3. 3

    Yours, not the model’s. If the AI provider changed its model in between, the result is flagged rather than credited to you.

Recheck
Before18% · 13–25%
After34% · 27–41%
Moved
  • Clear of the week-to-week noise
  • Big enough to matter

Rechecks

When you mark a fix applied, Shruwd gives the AI engines time to catch up, then measures the affected questions more thoroughly than a normal week. Checking the next day would only tell you how fast they re-crawl.

What you can track,
per plan.

Search Console, on Pro. On Pro you can connect Google Search Console and Google Analytics 4. Search Console adds Google's own impression and click data next to what the answers say, which is exact rather than sampled. Starter and the free plan measure answers and crawler logs. Crawler log ingest is exact too, and it is on every plan.

StarterStart here

Measure weekly, fix what matters

$49/ month

Billed monthly

Measurement

Engines
Google AI Overviews and ChatGPT
How oftenEvery prompt is asked on every engine, spread over a day.
Every week
PromptsThe questions your buyers ask AI assistants. One pool for all your brands.
25
RepetitionsAI answers vary between runs, so each prompt is asked more than once.
3 per prompt

Diagnosis

ChecksAccess checks whether AI crawlers can reach your pages. Structure and citation checks need weekly answers.
All
Findings shownA plan with a limit shows its most severe findings in full and counts the rest.
All
RechecksAfter a fix, the affected prompts run again at nine repetitions to show whether the number moved.
1 a month
HistoryHow far back the charts go. Changing plans deletes nothing.
90 days

Workspace

Brands
1
Team membersEveryone with access, the owner included.
5
Crawler log linesServer log lines read each month to see which AI crawlers reach your site.
500K a month
Search ConsoleSearch demand per page, which two more diagnostics need.
Not included

Every plan is a hard cap. There is no usage billing and no overage charge on any tier.

Questions and answers

Something not covered here? Ask us

How often does it check?

Weekly on Starter and Pro, with every question asked several times on each engine. The free plan takes a single snapshot when you sign up.

Does it track Perplexity, Claude or Gemini?

Not today. Shruwd tracks ChatGPT and Google AI Overviews.

Are these the answers a real person sees?

They are collected the same way every week, which makes them good for tracking change, though never identical to what one signed-in person sees. Where the answers come from

Why is my dashboard showing "insufficient data"?

There are not enough answers behind that view yet. It fills in as the weekly runs come in, and single questions take longer than your whole set.

Can I see the questions behind a number?

Yes. Each question has its own numbers on each engine, and the answers behind them, with who was named and which pages were cited, are available in the API.

What should I ask a tracker before buying one?

Our AI visibility guide has the checklist, and the sections above are Shruwd’s answers to it. The checklist

Find out what the answers say about you.

One measurement, your own crawler data, and the three access findings that matter most. Free, no card, about two minutes.

Start free