AI visibility

What AI search answers say about your brand, question by question.

Answers from the OpenAI and Google Gemini search APIs to your selected questions: brand mentions, cited URLs, competitor appearances and the collection status of every answer. Raw answers are kept.

Available now Create an account and enter your brand, domain, market and questions. You start it yourself once your questions are confirmed; collection usually takes a few minutes. Results appear in your private workspace. The first check is free.

Check your AI visibility View a sample report

Engines

Two search APIs, recorded the same way.

OpenAI API
Responses with the web search tool enabled. The model decides whether to search; each answer records whether it did and which URLs it cited.
Google Gemini API
Responses with Grounding with Google Search enabled. Each answer records whether it was grounded and which sources it returned.

These are API observations. They are not the consumer ChatGPT or Gemini apps, and not Google AI Overviews or AI Mode. Other engines are not offered until their collection is verified.

Questions

Brand questions and category questions, kept apart.

Category questions: a buyer is choosing and does not name you, such as “What payroll software fits a 75-person team?”
Brand questions: the question names you, such as “How much does your company charge?”
Your own questions come first: sales objections, enquiries and FAQ
Questions drafted from a template or by a model stay labelled as drafts until you confirm them, and are never counted as search demand

Question templates

What a result shows

Each signal is counted on its own.

What each signal in a result means
Signal	What it means
Brand mention	A successful answer names your brand or an alias you confirmed. A mention can be positive, neutral or negative.
Positive recommendation	Judged by a person from the actual sentence, with the reason recorded. Never inferred from a mention.
Own domain cited	The answer shows a URL on your domain as a source.
Competitor appearance	A competitor you listed appears in the same answer.
Possible description error	An answer differs from a fact your team confirmed. Shown side by side with the answer passage for your review; never confirmed by an AI alone.
Cited sources	The URLs the answer showed as sources, which you can open and read.
Comparable change	A difference between observations whose conditions match.
Collection status	Whether each planned answer succeeded, failed, is waiting or was not collected.

Collection status

A failed answer is not a missing mention.

Collection status of each planned answer
Status	Meaning	In the numbers
Waiting	Planned and not yet collected.	Not in any rate yet.
Succeeded	An answer came back. It is then either mentioned or not mentioned, with citations or with none.	Counted in the rates for its engine.
Failed	An error, a timeout or an empty answer after retries.	Excluded from the rates and shown in coverage. Never counted as "not mentioned".
Not collected	Stopped, or the requested condition is not supported for this measurement.	Shown as missing. The condition is never swapped for another one.

Measurement limits

What the observations cannot tell you.

How we measure

API samples are not counts of consumer-app users, impressions or search volume.
Questions a model or template drafted are not counted as real market demand.
A separate check restricted to your own site is never mixed into category-question observations.
Search being enabled is not the same as a search happening in that answer. We record which.
A list of search results is not the same as a citation shown in the answer. We record which.
Results from different questions, models, markets or modes are not merged into one trend.
A change after an update is not proof that the update caused it.

Scheduled monitoring

The same questions, observed again under the same conditions.

Early access Apply for early access

Whether a new competitor or source appears for questions you already track
Whether an old description remains after your product information changed
Whether mention or citation patterns changed for the questions you marked as important
What was observed after a change you recorded
Alerts limited to the questions you mark as important and changes that hold across repeats

Usage counts questions × engines × repeats × observations. For example, 50 questions × 2 engines × 3 repeats × 4 observations a month is 1,200 planned answers. That is a count of answers, not a cost estimate.

Sample report

What a result looks like.

Prepared from the FwdLink site’s B2B software example, October 2026. The company is “your company”; other companies are written as Competitor A to H.

Sample report · B2B software example AI visibility check · Your company 5 questions × 3 repeats per engine · search APIs · English · United States

Sample report. Per engine: successful answers out of planned answers, then, for product-choice questions that do not name the company, answers that mentioned it and answers that cited its own domain.
Engine	Successful / planned	Mentioned in product-choice answers	Own domain cited (product-choice)
OpenAI API	15/15	1/12	0/12
Gemini API	15/15	0/12	0/12

Brand question (names your company): mentioned 3/3 on both engines; own domain cited 2/3 (OpenAI) and 1/3 (Gemini). Read it for what the answer says, not for visibility.

Category What's a good tool to track tasks for a 30 person team? We're still on spreadsheets.
Category Recommend a task tracker under $10 per person a month.
Category Our tasks live in three spreadsheets and nobody knows what’s current. What do people use to fix this?
Category What are some alternatives to Competitor A?
Brand How much does your company charge per person a month?

Question 1 of 5 · Category question

“What's a good tool to track tasks for a 30 person team? We're still on spreadsheets.”

OpenAI API, repeat 1 of 3: Your company not mentioned · named: Competitor A, Competitor B, Competitor C, Competitor D, Competitor E
Cited pages: competitor-a.example competitor-b.example competitor-d.example
Improvement candidate: Review a draft that states the team size and per-seat price your team has confirmed on the pricing page. Hypothesis

Engines: OpenAI API (web search tool) and Google Gemini API (Grounding with Google Search)
Questions: 5: 4 category, 1 brand
Repeats: 3 per question per engine
Market and language: United States · English
Search: Enabled; whether each answer actually searched is recorded per answer
Collected: 30 of 30 planned answers succeeded

OpenAI API · Repeat 1 of 3 · Answer collected · Searched the web

What's a good tool to track tasks for a 30 person team? We're still on spreadsheets.

For a team of 30 that's outgrowing spreadsheets, these five are worth a look.

Competitor A Free trial up to 30 seats Linked its site
Competitor B Board and list views Linked its site
Competitor C Flat $6 per person a month Named only
Competitor D Built for teams under 50 Linked its site
Competitor E Timeline view, no setup fee Named only

Your company: Not mentioned in this answer
Cited pages: competitor-a.example
competitor-b.example
competitor-d.example

Gemini API · Repeat 1 of 3 · Answer collected · Grounded with Google Search

What's a good tool to track tasks for a 30 person team? We're still on spreadsheets.

Teams that size usually land on one of these.

Competitor B Free plan for small teams Linked its site
Competitor F Made for 20 to 60 people Linked its site
Competitor C Flat $6 per person a month Named only
Competitor G Checklists and recurring work Named only
Competitor H Imports straight from spreadsheets Linked its site

Your company: Not mentioned in this answer
Cited pages: competitor-b.example
competitor-f.example
competitor-h.example

Sample report · B2B software example

Observed: Both answers to question 1 named five task trackers for a 30-person team. Your company was not named, while competitors’ team sizes and per-seat prices were.
Page check: Your pricing page does not state the team size it is built for or the price per seat.
Improvement candidate Hypothesis: Review a draft that states the team size and per-seat price your team has confirmed on the pricing page.
Follow-up: Record the date the change goes live, then observe the same questions again under the same conditions.

A hypothesis, not a cause. A missing mention alone does not show why an answer left a brand out.

Open the full sample report

Results reflect the selected questions and recorded measurement conditions. API observations are labeled separately from consumer-app checks.

A first check

What you provide, and what you do not need.

The first check is free: up to 10 questions × 2 engines × 3 repeats = up to 60 planned answers. You start it yourself once your questions are confirmed; collection usually takes a few minutes.

Company or brand name and domain
One sentence describing the product or service
The language and market to observe
Competitors to compare (optional)
Your buying questions, or drafts for you to confirm

Not needed: a CMS connection, an ad account or invitations to other people.

Next step

See what AI answers say about your brand.
Your first check is free.