Compare
Dashboards measure. Eidomai does the work.
Good tools in this category hand you a number. Eidomai hands you the number, the next piece of content, and the peace of mind that someone is on it. Here is where each option fits.
Free AI check Free. ChatGPT and Claude. No call required.
The short version. Flip it and we show our work: the method, the numbers, the sources. Site-wide, or one section at a time.
Q: Best physical therapist for runners in Encinitas
Held by: Coastal Motion PT (Perplexity, Gemini, AI Mode). Sources: coastalmotionpt.com/runners, yelp.com, a 2025 Encinitas Advocate piece. Content to write: a running injury and gait program page, with the two clinicians named, return-to-run protocol described, Encinitas and Carlsbad both stated. Must not claim: Medicare figures, any outcome percentage.
Sample data
Why it is different
This is not another AI visibility score.
A score tells you a number went down. Eidomai tells you which question you lost, who took it, and what to publish to compete for it.
Some businesses have a dashboard an agency or a developer assembled from a tracking tool. It shows how often the engines mention you for the questions somebody typed into the tool. Three things to ask of it: are those your buyers’ questions, can you read it without a walkthrough, and what does it tell anyone to do on Monday. The map is built to answer all three.
How content gets written, checked and publishedFacts3/3
Reads human91
LiftableYes
ShippedSept 14
Both scores are advisory. A person still decides.
Sample dataNorthshore Physical Therapy is fictional.
Real answers
We save what each engine actually said about you, word for word.
Real competitors
See who gets recommended when you do not, and why.
Real work
For the questions worth working on, a next step built from facts you confirmed: improve existing content, add new content, or correct a fact. Not one piece of content per lost question.
Real proof
Same questions, same engines, next month. What moved, and what did not. A person checks every fact before it goes live.
Dashboards measure. Eidomai closes the loop. Tools hand you numbers and leave. Agencies do work you cannot inspect. Eidomai is a loop you can watch turn, and it is identical for a four-location practice or a community foundation: the questions change, the loop does not. The same eight stages, every month, for any business. The whole process is published on How it works.
Can I do this myself? Yes. The process is on the portal and on this site. Most owners look at the eight stages and choose done for you, but the choice is yours and the price is different. See pricing and who cooks.
How do I know it works? Trend, never truth. Single-month moves are noise; flips take 60 to 90 days after indexing. The ledger shows every movement with the query named, and the outcomes report puts AI conversions next to what you pay Google Ads for the same conversion.
The landscape, in one table
What Eidomai does versus what they do.
Strong products for the audiences they serve. This table exists so you can pick the right kind of help, including when that is not Eidomai.
| Dimension |  EidomaiBuilt by Nimbletoad | GEO agencies | |||
|---|---|---|---|---|---|
| What it is | A growth engine: the measurement is wired to a content pipeline. The questions worth working on become briefs, a brief becomes content or a fix, it ships, and next month’s measurement grades it. The ledger, the fact file and the held questions compound every turn. | A monitoring platform. You see the numbers; the work happens elsewhere | A content editor with an AI tracker attached. The tracker and the editor are two products with a link between them | A monitoring dashboard. Charts of mentions, nothing to do next | A retainer. Work you cannot inspect, measured however the agency measures it |
| Built for | Local and specialty brands in health and medical, therapy and mental health, wine, beer and spirits, law, investment advice, securities brokerage, real estate, and hospice and home care, each with its industry's advertising rules built in. Every other business gets the FTC's truth-in-advertising rules | Fortune 500 and enterprise brands with in-house teams | Content teams and SEO writers; AI tracking added onto a content optimization suite | Marketing teams and SEO agencies that want to run their own dashboards | B2B, SaaS, and enterprise content programs |
| AI engines measured | Eight: ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews, Google AI Mode, Copilot, Grok | Ten or more on enterprise plans; entry tiers track fewer | Five: ChatGPT, Google AI Overviews, Google AI Mode, Gemini, Perplexity. No Claude, Copilot, or Grok | Three to seven, capped by plan; some engines cost extra | Varies with the tooling the agency licenses |
| What you actually see | A battle map: every buyer question scored on every engine, held or lost, with the engine’s verbatim answer as the receipt | Prompt dashboards, citation analytics, prompt volume data | Visibility score, mention rate, average position, share of voice, sentiment, citation sources. Daily refresh on higher plans | Prompt lists, mention charts, source breakdowns | Typically a monthly deck summarizing activity |
| Who reads it | The client, on one page, in ten seconds | The agency or the marketing team, in a dashboard | The agency or the marketing team, in a dashboard | The agency or the marketing team, in a dashboard | The agency or the marketing team, in a dashboard |
| What it says to do next | A brief for every lost question, checked before it ships | A score that moved | A score that moved | A score that moved | A score that moved |
| Wrong-claim defense | Every factual claim AI makes about you is audited monthly on every plan. On Done for you, we write and send the corrections, and a weekly watchdog catches new claims between audits. | Sentiment and citation tracking; corrections are your job | Sentiment and mention gaps; accuracy auditing and corrections are not offered | Sentiment scoring on higher plans; corrections are your job | Not a standard deliverable |
| Works inside Claude and ChatGPT | Yes, on The Map and Done for you. Your AI reads your account, drafts only from your verified facts, and the draft check blocks a draft before it can be marked shipped. You choose read only or read and write. | An MCP server on the Enterprise plan. Its setup docs list Claude; ChatGPT is named in an October 2025 post. Brings your Profound data into the chat | Surfer MCP, in beta, for Claude and ChatGPT on Pro plans and up: AI Tracker data and the Content Editor | Otterly and Peec connect to both and bring your tracking data into the chat. Semrush has an official app for its SEO data; its MCP docs do not list the AI visibility toolkit | Not a standard offering. One GEO agency we found offers a site audit you can run from the chat |
| Dimension | EidomaiBuilt by Nimbletoad | GEO agencies | |||
|---|---|---|---|---|---|
| Who does the work | Your choice: you cook from the map, or we cook: we write, you approve every word, and we publish. Agencies can run the map for all their clients under their own name. | You operate the platform; content tools on higher tiers | You operate the editor and write (or generate) the content yourself; “Optimize from AI Tracker” links a losing prompt to their editor | You operate the tool and do all optimization yourself | Agency does the work on retainer, usually enterprise-priced |
| Getting your name right | Your brand and its variants resolve to one entity, and so do your competitors. “Northshore Physical Therapy” and “Northshore PT” both count as you. A founder’s name is measured separately and reported beside your mention rate, never folded into it: we found that error in our own data and fixed it, because an answer that names the person who invented your field is not the same as an answer that names your organization. | Entity handling is not documented publicly; enterprise onboarding defines the brand set | Entities are auto-extracted per prompt. In a live account we ran, one client appeared as two competing brands, the top competitor was split sixteen ways, books and directories were counted as brands, and the founder was tracked as a rival to the institute she founded | Usually a single brand string you type in; variants are missed silently | Depends entirely on who set up the tracking |
| Whether the number holds still | Every engine gets every question multiple times each month, and any cell where the samples disagreed shows the count on the tile. Any cell that split, or that changed hands since last month, is asked again until it is at least five samples deep. A move only counts as confirmed when the answer held across that deeper set; anything shallower is shown as provisional and never counted as proof. In September that re-sampling retired 24 percent of the flips a three-sample read had claimed. | Enterprise sampling methodology is not published | Prompts are tracked over time; the per-run sample count and the spread are not surfaced in the report | Typically one pull per prompt per run, presented as a ranking | However the underlying tool does it, usually unstated |
| Whether you can act on it before next month | Paste a draft on the Briefs tab and get two scores before it ships: how human it reads and whether an answer engine can lift it. Mark it shipped and the date lands on your trend chart. Spend a credit and the questions it targets are asked again that day on all eight engines. Pick five questions and they are re-asked every Monday. | Daily tracking on enterprise plans; content checks are a separate product | Content editor scores drafts against a ranking model; whether an engine named you is a separate tracker | Scheduled runs only; no draft check, no on-demand re-measure tied to published content | Depends on the agency’s own process; rarely tied to a measurement |
| What counts as a mention | Defined and printed where you can read it: the methodology card on every battle map states the questions, the sampling, the scoring rule, and the panel version. A mention, a citation, and a recommendation are scored as three separate things, because being named in passing is not the same as being the answer. | Metrics defined in product documentation; definitions differ from other tools, so scores are not comparable across them | Visibility score and share of voice computed by their own rules; not comparable to any other tool’s number | Every tool in this category defines a mention differently, and few publish the rule | Inherited from whatever tool the agency licenses, usually without stating it |
| What a cited source even is | A citation is a page the engine read, so a search result page or a shopping card is never counted as one, redirect links are resolved to the real source, and the rule that decides is versioned with a dated change log printed on the methodology card. | Citation analytics by source; normalization rules are internal | Cited domains listed as returned; no published rule on what is excluded | Whatever the engine returned, including result pages when they appear | Inherited from the tool |
| Whether a mention is even real | Engines can invent a brand into an answer without retrieving a single page about it. Our measurement runs with web search on, so answers arrive with the sources the engine actually pulled, the verbatim answer is kept as the receipt, and every factual claim is audited monthly against a fact file your business confirmed. An invented mention is never counted as a win. | Citation analytics show which sources were used; validating individual claims is your job | Mentions are counted as the engine reports them; there is no fact-validation step | Raw mention counts; a hallucinated mention counts the same as a real one | Depends on the analyst reading the report |
| Where the questions come from | We build the panel from the questions your buyers actually ask, drafted from your market and your competitors, then show you every one and let you propose more. | Prompt sets configured during enterprise onboarding | You supply the prompts | You supply the prompts | Whatever the agency decides, often invisible to you |
| Asked from your market | The same question asked from two cities returns two different answers. Where your questions are ambiguous about place, the request itself carries your market, so a San Marcos clinic is measured against San Marcos, California, never Texas. Every battle map states the market it was measured from. | Regional options on enterprise plans; per-market methodology not published | Location handling was not surfaced in the reports of the live account we ran | Usually one region per workspace, often unstated | Rarely disclosed |
| Month over month | The same question panel re-measured every month, so movement is real: share of voice, mention rate, recommended-first rate, wrong claims resolved, with a movement report tying published work to flipped cells. Trend, never truth. | Dashboards update continuously; you build the narrative | Prompts rechecked daily or weekly; trend depends on your prompt hygiene | Charts update; trend depends on your prompt hygiene | Reporting cadence and rigor vary by shop |
| Proof it mattered | Google Analytics joined to the visibility numbers: AI referral sessions and conversions reported next to the map, labeled a floor, not a total | GA4 integration on enterprise plans | Looker Studio reporting of visibility metrics; not joined to conversions | Rarely connected to outcomes | Revenue case studies at enterprise scale |
| Human review | A person reviews every question panel, every draft, and every correction before anything ships | Software, self-serve | Software, self-serve; AI-generated articles ship as fast as you click | Software, self-serve | Depends on the agency |
| Price posture | Three plans. The Scan: your map and one brief and prompt a month. The Map: every brief and the full work list, and you write. Done for you: a person at Eidomai writes, checks and publishes, and you approve every word. | A self-serve entry tier, then enterprise by quote | Published tiers on annual billing, plus a standalone tracker | Low monthly tiers with seat and engine limits | Monthly retainers at enterprise scale |
- Competitor rows
- Public vendor sites and published reviews as of July 2026
- Eidomai rows
- The published process on How it works, which is also the contract
- What is not here
- Prices for anyone but us. Vendors change theirs; ours are on Pricing
- Bias
- We built one of the columns. The table exists so you can pick the right kind of help, including when it is not us
Where this table comes from
Competitor columns were read from each vendor’s public site and published reviews on the date shown. Vendors change their products; if a row is out of date, tell us and we correct it.
- Profound: tryprofound.com, checked July 2026
- Surfer: surferseo.com, checked July 2026
- Monitors: otterly.ai, peec.ai, semrush.com (AI toolkit), checked July 2026
- Claude and ChatGPT row: docs.tryprofound.com/mcp, docs.surferseo.com (Surfer MCP), help.otterly.ai, docs.peec.ai/mcp, semrush.com/mcp and semrush.com/kb/1618-mcp, checked October 2026
- Agencies: published service pages of GEO agencies named in the September 2026 measurement, checked September 2026
- Eidomai: the published process on How it works, which is also the contract.
The engines we ask, every month
ChatGPTPerplexityClaudeGeminiGrokGoogle AI OverviewsGoogle AI ModeCopilot
One question at a time
The comparisons, written out.
Software against agency, tracker against tracker, The Map against Done for you. Each one says where it was measured and when.
Profound vs Peec vs Otterly for a small business, or should I just hire someone?The three tools the engines name most for AI visibility, what all three have in common, and the question a small business should answer before buying any of them: who writes the content after the tool reports the loss?AI visibility software vs hiring an agency, which is better for a medical practice?The engines answer this with tool names. The real question is who writes the content, who checks the facts, and who fixes the source when an engine gets your practice wrong. A plain comparison, with the case for each.
Start where every engagement starts: a measurement.
Free check on ChatGPT and Claude, results by email, no call required.
Free AI check Free. No call required. You keep the results.
It starts with an eye. It ends with AI. Eidomai, by Nimbletoad.