Type one word and see what people actually type after it.
One word reads best. Full sentences belong in the box next door.
Tap to fill
This is an example.
Sample
28 phrases came back.
We pulled "dental implants" ourselves on 2026-08-10 and recomputed it here with the same function. Type a word above and this block becomes your result.
What people add
- dental implants
- dental implants near me
- dental implants cost
- dental implants in mexico
- dental implants turkey
- dental implants uk cost
- dental implants procedure
- dental implants before and after
- dental implants cost per tooth
- dental implants cost full mouth
- dental implants best reviews
- dental implants best price
16 phrases we left off the screen
We counted 28 and printed 12. The other 16 stayed on our side. That number is measured, not rounded.
We pick which ones to measure first. Then we put them to the engines and read back whose name comes out.
The first paid rung is $399.
What they want to find out
We only look at which words sit in the phrase. We do not read meaning, so "not recommended" still counts as choosing. Anything matching no list counts as learning.
Phrases that came out as questions
- are dental implants safe
- how long do dental implants last
Read one of these as a question →
What to fix first
Take the three most common additions and make them your subheads. The page then answers three real questions.
What you saw here / what the paid tier adds
| Free ends here | On the paid tierOne pass, then it is done | Which product opens it |
|---|---|---|
|
|
|
What splits them — How wide we open it, and whether we measure
Here is what the paid readout looks like — The numbers in the locked rows go in once we have measured them.
Free ends here
Opening the list costs us nothing, so you get all of it. Putting those phrases to real AI answer surfaces to see who gets cited costs us money on every run.
The first paid rung is $399. See the diagnostic ladder →
What this tool cannot see
- No search volume and no cost per click. Autocomplete returns phrases, not counts.
- This is a reading of right now. Type it again tomorrow and the list can differ.
- Google autocomplete only. Bing, YouTube and Amazon run their own lists.
- We do not store the word you typed.
Paste one question and see what kind it is, and which questions stand next to it.
Type it the way a person would say it. A question mark is fine.
Tap to fill
This is an example.
This is an example
Wants to choose
This one is picking between options.
Words that matched — worth it
How we sort — we check whether certain words appear. We do not read meaning. When several kinds match, the most specific phrase wins.
Below is the first example sentence read by the same code. Neighbouring questions appear once you run it, because that is when we call autocomplete.
We have not counted this kind on its own yet, so this row shows the whole panel.
Of those 50 questions, 32 were buying questions, 13 were meaning questions, and 5 carried a company name.
One line matters here. Landing on page one of search and getting cited in an AI answer are two different jobs. A citation is the source an engine leaned on while writing its answer.
What shape your answer needs
Put one comparison table on the page. Rows and columns are what an engine lifts.
What you saw here / what the paid tier adds
| Free ends here | On the paid tierRe-measured every month | Which product opens it |
|---|---|---|
|
|
|
What splits them — Sorting it versus measuring it
Here is what the paid readout looks like — The numbers in the locked rows go in once we have measured them.
Where this stops
Putting your sentence to the engines and reading back who gets cited is paid. Every run costs us real money. Free gets you what kind of question you are holding.
The first paid rung is $399. See the diagnostic ladder →
What this tool does not tell you
- It does not say who gets cited for your question today. We have never published a per-question citation list, so it does not exist, and we will not pretend otherwise.
- The sorting is word matching. Irony and negation slip through, which is why we print the words that matched.
- The company-name check knows our own names only. Someone else's brand needs a name list first.
- Both gauges were measured in Korea and Japan. Each one carries its scope.
- Autocomplete is a live value. Ask again tomorrow and the lines move.
Put two domains in and see whose door the AI crawlers can walk through.
Separate them with a comma. Three or more is a paid view.
Tap to fill
This is an example.
This is an example
Nothing is in the box yet. Below is the first example above, run for real. The numbers are real; only the input is an example.
Every door is open on both sides.
So the difference is not the door — it is what sits behind it. Which of you gets cited for the same question shows up in the other panels.
- Door-by-door comparison
- Open only for them
- Open only for us
- When and what we read
When and what we read
ahrefs.com · semrush.com
19 crawlers
| Crawler | What it feeds | You | Them |
|---|---|---|---|
| GPTBot | training | open | open |
| ChatGPT-User | user_fetch | open | open |
| OAI-SearchBot | retrieval | open | open |
| OAI-AdsBot | user_fetch | open | open |
| ClaudeBot | training | open | open |
| Claude-User | user_fetch | open | open |
| Claude-SearchBot | retrieval | open | open |
| anthropic-ai | training | open | open |
| PerplexityBot | retrieval | open | open |
| Perplexity-User | user_fetch | open | open |
| Applebot | retrieval | open | open |
| WRTNBot | retrieval | open | open |
| Google-Extended | training | open | open |
| Applebot-Extended | training | open | open |
| CCBot | training | open | open |
| Bytespider | training | open | open |
| cohere-ai | training | open | open |
| meta-externalagent | training | open | open |
| Meta-WebIndexer | retrieval | open | open |
Check the closed doors
First check whether each closed door was closed on purpose. A door you did not mean to close is the cheapest thing you will ever fix.
What the paid rung adds
- Not one rival but several, side by side
- How the two of you split the citations on one question board
- A record of doors opening and closing over time
- Page by page, not domain against domain
The first paid rung is $399.
Which product opens it
- Panorama$2,599
- Competitive Signal Program$10,470 a quarter · billed $3,490 monthly, three times
What the door check misses
- This panel only reads whether the door is open. An open door does not mean you get cited.
- It reads what robots.txt says. Blocks applied at the server are invisible here.
- The example shown first is measured in advance. A domain you type is measured on the spot.
Enter one domain and see where it sits on the citation ladder. The US panel is not measured yet — Korea and Japan are.
The Korean and Japanese ladders read panels we already measured. The US one is on the list, not on the page.
Tap to fill
This is an example.
This is an example
Nothing is in the box yet. Below is the first example above, run for real. The numbers are real; only the input is an example.
We have not measured the US market deeply enough to put a list here yet.
The lists standing right now are JP, KR. US goes up here the moment we measure it the same way.
- The ten rungs
- Your rung
- The entry threshold
- The denominator and the observation window
Put one page next door
Citations are built by everything published on that domain. Put one page into the box next door first and see if it is liftable.
This box is free end to end. Nothing is held back, so we drew no lock box. Below is the plan this tool sits in.
Which product opens it
- Competitive Signal Program$10,470 a quarter · billed $3,490 monthly, three times
What the ladder cannot show
- One panel, one point in time. What happened afterwards is not in it.
- Being off the list does not mean zero citations. It means outside the top places of this observation.
- The list does not say how many you have today. What it can give is a ceiling: at most this many more.
Give us a brand, a domain and the questions you care about, and we show how many surfaces carry that question and where this market stands today.
What lands here is what we know before the run. Your own numbers come from the weekly subscription.
Tap to fill
This is an example.
This is an example
Nothing is in the box yet. Below is the first example above, run for real. The numbers are real; only the input is an example.
What we know right now
Every question you give us goes to 6 AI answer slots. Right now that is 6 observations.
Counted on what you typed. Free covers up to 3 questions.
- Name shown or not, slot by slot
- Spread across repeats
- Against competing brands
- The observation denominator
What we measure and send
We run them, read back which slots carried your name, and repeat the same question to see how much the answer moves. It lands in your inbox.
Decide what to ask
Settle the questions first. Get the question wrong and every number after it measures something else.
What the paid rung adds
- Every slot we watch in this market, measured
- The same question repeated, so you see the spread
- The sources each answer carried
- Competing brands placed next to yours
The first paid rung is $399.
Which product opens it
- Radar$399
- Compass$1,199
- Competitive Signal Program$10,470 a quarter · billed $3,490 monthly, three times
What we cannot see before measuring
- What is on this screen is what we know before measuring. Whether your name comes back is not measured yet.
- AI answers move between runs on the same question. That is why we repeat and print the spread.
- Whether an AI crawler can reach your site is not what this box looks at.
Paste something you wrote and see, paragraph by paragraph, which parts an AI can lift.
We read the text you paste. The rules we grade it by are on this page.
Tap to fill
This is an example.
This is an example
Nothing is in the box yet. Below is the first example above, run for real. The numbers are real; only the input is an example.
Evidence density 100% — 1 of 1 measured paragraphs carry it.
Every measured paragraph carries evidence. What separates drafts now is kind, not density — one angle answers one kind of question.
- Evidence density by paragraph
- Which of the four evidence kinds are there
- Where the answer closes
- Where the brand lands
Which kinds are in there
Present: procedure
Missing: definition, number, comparison
At least 2 kinds. With one kind, an AI has one angle to lift from.
Does the answer close in the first window?
The opening answer closes at 61 characters — inside the 200 character window.
| Paragraph | Evidence found |
|---|---|
| 1 | procedure |
Start where the evidence is missing
Start with the paragraphs carrying no evidence. One number, one comparison, or one step is enough to turn a paragraph around.
This box is free end to end. Nothing is held back, so we drew no lock box. Below is the plan this tool sits in.
Which product opens it
- Refit Sprint$15,900
What the paragraph check misses
- It reads only what you paste. Whether the page gets cited is not visible here.
- It does not check whether your facts are right. It checks shape.
- Short lines are not counted as paragraphs. We print how many we skipped.
The places we watch
These are the screens where AI writes the answer, plus the search results underneath. We watch them all. That is not a promise to put you in them.
- AI answers 10 places
- ChatGPT
- Perplexity
- Claude
- Grok
- Gemini
- Google AI Mode
- Google AI Overviewswe read it on the Google Search page
- Bing Copilot
- DuckDuckGo Search Assist
- Meta AI model answers
- Search 4 places
- Google Search
- Yahoo Search
- Bing Search
- DuckDuckGo Searchwe read it on the DuckDuckGo Search Assist page
- News 2 places
- Google News
- Bing News
- Video 2 places
- YouTube
- Google Short Videos
- Shopping 1 places
- Google Shopping
- Social 1 places
- X search
Where the answers cite
These are the places the answers link out to. We count them separately from where we ask.
- News & PR
- businesswire.com
- onrec.com
- prnewswire.com
- newswire.com
- Built-in AI
- Google AI Overviews
- Social
- Social & News
- Social, News & Reviews
- Social, Video & News
- TikTok
- Social & Communities
- Wikis
- en.wikipedia.org
- Communities
- medium.com
- Academic
- arxiv.org