Rank. Spend.Sessions.AI searchstarts there.
MantiSearch reads Google’s page.MantiDark estimates your AI traffic.MantiGEO counts your AI mentions.
| Reading | W12 | State |
|---|---|---|
| Google’s page | Rank 2 | Measured |
| Google Ads | $7,490.00 | Measured |
| Google Analytics 4 | 1,840 | Measured |
| AI Overview citations | 19% | Measured |
| AI mentions | 40 of 82 answers | Measured |
| Estimated AI | 3,070 | Estimated |
Chart of demo data for On in W12. Three measured points: rank 2 on Google’s page, $7,490.00 past spend in Google Ads over 29 Aug–27 Sep 2026, and 1,840 Known AI sessions in analytics. AI Overview citations and AI mentions are solid lines. Estimated AI is a dashed line. Each small numeral is On’s rank for one tracked keyword. Dots fill the part that is not measured.
Move the pointer to take a reading. The ruler sets the week.
Tap the chart to take a reading.
- Measured
- Estimated
- Not measured
- A number you own
- A figure MANTIserp counts
- On’s rank for one tracked keyword
How to read this page
Our view: a budget bet on an AI visibility score you cannot check is a shot in the dark. So every line on this page says how its figure was made.
- Not measured
- Nothing was read here, so the page draws dots and no figure: a gap stays a gap.
- Estimated
- An estimate is dashed, and the count it starts from is always drawn beside it.
- Measured
- A solid line is read from a page or counted from stored rows: Google’s page, a spend report, a session, an answer.
A rank is a fact.Start there.
You report rank every week. MantiSearch reads 4 surfaces of Google’s page, and one of 24 rules gives the verdict. “Rank without the AI answer” is one. You act on it.
One keyword, read
Demo data · Running shoes · US · W12 · Desktop
- ASICS official store | Marathon running shoes
- Nike official store | Marathon running shoes
The best marathon running shoes pair cushioning for long distances with a light, responsive build. ASICS and Brooks are named for cushioned shoes that last through the training weeks. Many runners keep a lighter pair for race day.
- ASICS | Marathon running shoes
- On | Marathon running shoes
- Brooks | Marathon running shoes
- Marathon running shoes, tested over 300 miles
- Nike | Marathon running shoes
- How to choose marathon running shoes
- Marathon running shoes: this year's picks
- Marathon running shoes compared by weight and drop
- A buyer's guide to marathon running shoes
- Marathon running shoes: what runners ask
- On marathon running shoe
- ASICS marathon running shoe
- Nike marathon running shoe
- Brooks marathon running shoe
- Ads
- Ads · ASICS and Nike
- AI Overview
- AI Overview · cites ASICS and Brooks · On is not cited
- Organic
- Organic · On at rank 2
- Shopping
- Shopping · On, ASICS, Nike and Brooks
Demo data. Titles and snippets are invented.
The rank is read off the page Google served for this keyword; pick another keyword or device, and read the facts behind its verdict.
Every keyword, followed
Every tracked keyword runs from Google’s page to “How urgent?”, and its AI Overview step is read from that same page; pick a slice to see what to work on first.
You and your competitors
The six figures for your weekly slide, each with its change, your place among your competitors and a 12-week line; pick a week on the ruler.
What you can save
The keywords you pay for and already win show here as money already spent, read from your own Google Ads account.
Turn it off: the money figure leaves. Your wins stay, and paid wins are counted from your ad on the page, because MantiSearch does not guess your spend.
Estimated is not measured.We show which.
In your analytics, some sessions come from an AI service. Other AI sessions arrive as Direct. MantiDark estimates those, and you report the two figures apart.
Known AI
Start with the sessions whose source in your analytics is an AI service: a count, so the line is solid.
Cited pages
Next come the pages AI answers cite, starting with the ones above: the base of the estimate is the cited pages where, that week, no session has an AI service as its source.
Estimated AI
Only now does a dashed line appear: the estimate, with the band it is checked against, and a flag in the week it leaves that band.
- W12 · Known AI 1,840 · Measured
- W12 · Estimated AI 3,070 (Estimated) · Estimated
- W12 · Direct 212,400 · Counted, not as AI
Demo data
| Week | Known AI | Direct sessions | From cited pages (Estimated) | Estimated AI (Estimated) | Check band |
|---|---|---|---|---|---|
| W01 | 860 | 188,000 | 964 | 1,824 | Inside the band |
| W02 | 910 | 189,500 | 982 | 1,892 | Inside the band |
| W03 | 980 | 191,200 | 1,006 | 1,986 | Inside the band |
| W04 | 1,040 | 193,000 | 1,036 | 2,076 | Inside the band |
| W05 | 1,120 | 221,000 | 1,580 | 2,700 | Outside the band |
| W06 | 1,210 | 196,400 | 1,072 | 2,282 | Inside the band |
| W07 | 1,290 | 198,100 | 1,098 | 2,388 | Inside the band |
| W08 | 1,380 | 200,300 | 1,126 | 2,506 | Inside the band |
| W09 | 1,490 | 203,000 | 1,156 | 2,646 | Inside the band |
| W10 | 1,600 | 206,500 | 1,180 | 2,780 | Inside the band |
| W11 | 1,720 | 209,800 | 1,204 | 2,924 | Inside the band |
| W12 | 1,840 | 212,400 | 1,230 | 3,070 | Inside the band |
- Known AIMeasuredSessions whose source is an AI service
- 1,840
- Direct sessionsMeasuredCounted, not as AI
- 212,400
- From cited pagesEstimatedEstimated from Direct sessions on cited pages with no session from an AI service
- 1,230 (Estimated)
- Estimated AIEstimatedAn estimate: only Known AI is counted as AIInside the band
- 3,070 (Estimated)
Known AI, 1,840, plus the part from cited pages, 1,230, gives Estimated AI, 3,070. The last two are estimates.
In W05 the estimate is outside its band. It carries a warning, and the figure is not changed.
| Page | Known AI | Estimated AI |
|---|---|---|
| /en-us/explore/marathon-running-shoesFollowed page | 411 | 411 (Estimated) |
| /en-us/explore/cushioned-running-shoes | 368 | 368 (Estimated) |
| /en-us/explore/running-shoes | 344 | 344 (Estimated) |
| /en-us/explore/trail-running-shoes | 291 | 291 (Estimated) |
| /en-us/explore/waterproof-running-shoes | 236 | 236 (Estimated) |
| /en-us/stories/how-to-choose-running-shoes | 190 | 190 (Estimated) |
| /en-us/explore/running-shoes-for-flat-feet | 0 | 763 (Estimated) |
| /en-us/stories/marathon-training-plan | 0 | 467 (Estimated) |
The first row is the page behind the keyword you followed in MantiSearch.
Seven engines.Every answer kept.
Someone asks what AI says about your brand. MantiGEO collects answers from 7 engines each week and counts mentions and citations. You open a count and read its answer.
Demo data. Answers are invented for the demo.
Seven lines open from one prompt, one for each engine. 2 of the 7 engines with a stored answer cite an on.com page for this prompt. Demo data.
Tap an engine to open its answer.
Choose an engine to open its answer.
From the page MantiSearch read for this keyword.
Promptbest marathon running shoesThe keyword you followed above, as written
ChatGPT Cited
Prompt
best marathon running shoes
Stored answer
For a marathon, most testers start with ASICS for cushioning that lasts past 30 kilometres. On is the lighter choice at race pace, and Brooks and Nike follow for runners who want a softer or a faster shoe. Try each pair late in a long run before you choose.
Brands in this answer
Quote
On felt light at race pace, and the fit stayed secure.
Quoted from the stored page, word for word.
Source
runnersbench.example
Cited pages
- runnersbench.example/marathon-running-shoes
- www.asics.com/us/en-us/marathon-running-shoes/
- www.on.com/en-us/explore/marathon-running-shoesThe same page, aboveGoogle rank 2Known AI 411
On is cited by 2 of 7 engines with a stored answer for this prompt.
| Engine | On | Named in this order | Cited pages |
|---|---|---|---|
| Google AI Overview | Not mentioned | ASICS, Brooks | runnersbench.example/marathon-running-shoes, www.asics.com/us/en-us/marathon-running-shoes/, www.brooksrunning.com/en_us/marathon-running-shoes/ |
| ChatGPT | Cited | ASICS, On, Brooks, Nike | runnersbench.example/marathon-running-shoes, www.asics.com/us/en-us/marathon-running-shoes/, www.on.com/en-us/explore/marathon-running-shoes |
| Claude | Mentioned, not cited | ASICS, Brooks, On | thelongrun.example/marathon-running-shoes |
| Gemini | Not mentioned | ASICS, Nike, Brooks | runnersbench.example/marathon-running-shoes, www.asics.com/us/en-us/marathon-running-shoes/ |
| Perplexity | Cited | On, ASICS, Brooks | runnersbench.example/marathon-running-shoes, www.on.com/en-us/explore/marathon-running-shoes, www.asics.com/us/en-us/marathon-running-shoes/, www.brooksrunning.com/en_us/marathon-running-shoes/ |
| Grok | Not mentioned | Nike, ASICS | thelongrun.example/marathon-running-shoes |
| Google AI Mode | Mentioned, not cited | ASICS, Brooks, On | thelongrun.example/marathon-running-shoes, www.asics.com/us/en-us/marathon-running-shoes/, www.brooksrunning.com/en_us/marathon-running-shoes/ |
| On | Mentioned in 40 of 82 answers | Cited in 19 of 82 answers |
|---|---|---|
| ASICS | Mentioned in 44 of 82 answers | Cited in 23 of 82 answers |
| Nike | Mentioned in 46 of 82 answers | Cited in 12 of 82 answers |
| Brooks | Mentioned in 49 of 82 answers | Cited in 21 of 82 answers |
An answer here is close to what a person sees in the assistant, not identical.
Made once.Quoted everywhere.
You take a figure to management. It was computed once and stored. The screen, the API, Eve and the apps under “Connected AI apps” read that row. You quote it once.
A run copies its settings when it starts, so a later change does not reach an earlier week by itself.
Facts are written once. A partial run never publishes silently, and a failure is never stored as an empty result.
Eve quotes stored numbers. She is built not to work one out.
A change an AI proposes is a preview first, and nothing is written until you approve it. In Eve you press “Confirm”: a typed yes is not approval.
Quoted fromKeyword detailWhat you can save
Notmeasuredyet.
Access to MANTIserp is by invitation. It is for brands that already measure search and pay for Google Ads. The request tells us if that is you.
- Step 1 of 4: Your brand, current
- Step 2 of 4: What you measure today, not started
- Step 3 of 4: Your question, not started
- Step 4 of 4: You, not started
- Check and send
An empty chart sheet. Your brand appears on it as you answer.