Gap analysis

The Gap Analysis page looks at AI answers that name a tracked competitor but not your brand. We call these gap answers. The page shows how often that happens, which rivals take your place, and the prompts to fix first. These are your highest-leverage opportunities — the questions where buyers are getting recommendations and you're invisible.

What you see

1. Gap rate

The large number is your gap rate: the share of answers in the window that name a tracked competitor without naming you. Lower is better.

Beside it:

The change — the gap rate on the last day with answers minus the first, in percentage points. A fall is good news. It needs at least two days with answers.
A sentence — how many of the answers named a rival but not you, out of how many.
By model — the gap rate on each AI model, highest first.
Gap rate over time — a chart of the rate on each day with answers. Days without answers are hatched: they're gaps in the data, not a fall to zero.

2. Who takes your place

Up to four cards, one for each rival named most often in your gap answers. Each card shows:

the share of your gap answers that name that rival, and how many answers that is;
the category that holds the most of its gap answers.

One answer can name several rivals, so the shares overlap: added together, they can come to more than all of your gap answers.

Click a card to focus the next two cards on that rival. The page opens on the rival named most often. All rivals shows every prompt instead.

3. Where a rival replaces you, and the prompts it wins

With a rival picked, two cards sit side by side:

Where [rival] replaces you — that rival's gap answers by category, and a split by AI model. The model split is labelled with what it counts: today it takes the prompts in the sample that name the rival and shows how many had an answer on each model that left you out. Compare in Competitors opens the Competitors page with the same window and tags for the head-to-head detail.
Prompts [rival] wins — the prompts where that rival is named and you aren't.

With All rivals, the list is Prompts to fix: every gap prompt, with extra columns for the rivals named and the state on each model.

The prompt list has these columns:

Prompt — the question, with its intent and category. A Partial coverage tag marks prompts where some answers name you and others name a rival without you: the ones closest to winning.
Score — a gap score from 0 to 100, used to rank the list. It rises with the number of rivals named, counts three times as much for a transactional prompt and twice as much for a commercial one, and gets a boost when the gap was seen in the last seven days or shows on every model asked.
Gap rate — the share of this prompt's sampled answers that named a rival but not you.
Rivals named and Models with gap (with All rivals) — each model is marked Gap, Mixed, Named or No gap seen.

Above the list you can switch between All gaps and Partial coverage, and search prompts, rivals and categories. Click the Score or Gap rate heading to sort by it. The list shows twenty prompts at a time, with Previous and Next.

Open a row to see every rival named, each model's state, how many answers left you out and when the gap was last seen, and, where available, short quotes from what the models said.

The prompt list works from a sample of the latest answers in the window, so it can cover fewer answers than the gap rate at the top. The coverage note gives the sample's size.

4. Who replaces you, by category

A table with the rivals from the cards as columns and up to six categories as rows: the categories where those rivals are named most in your gap answers. Each cell shows how often that rival is named in your gap answers in that category, and MOST marks the rival named most in each category. When one rival is the most named in two or more categories, and in more of them than any other rival, a sentence above the table says so.

How to read it

For each prompt with a meaningful gap, ask three questions:

1. Why is the competitor showing up?

Open the prompt's row and read what the models said. To see the sources behind those answers, check Citations. Common reasons:

They have stronger citations (G2, Trustpilot, news).
They have specific landing pages matching the prompt's language.
They're cited in third-party comparison content.

2. Is this a prompt I should be in?

Not every gap is worth closing. Some are:

Prompts that don't fit your offering. If you don't sell to that segment, AI is right to leave you out.
Prompts about features you don't have. "What's the best CRM with built-in calling?" — if you don't have calling, that's not a gap to fix.

Skip these and focus on prompts where you genuinely should appear. Partial coverage prompts are a good place to start: some answers already name you.

3. What's the smallest change that could close it?

Closing a gap usually requires one of:

Earning a citation — getting added to a comparison post, refreshed in a directory, or mentioned in a review.
Publishing matching content — a landing page, blog post, or comparison guide that AI can pull from.
Improving your existing content — making your messaging on a relevant page more specific to the prompt.

Not all gaps close fast. AI updates its picture of the world over weeks and months. But cumulatively, closing a few gaps a quarter compounds. Recommendations groups related gap prompts into pages worth publishing.

Filters and controls

The controls sit beside the page title:

Coverage note — when the data was captured and how many days had answers. Open it for the window, the number of answers and gap answers, the size of the prompt sample and how each number is worked out.
Filters — Model, Category, Intent and Tags. They apply to every card on the page.
Date window — 7, 14, 30, 60 or 90 days. The default is the last 30 days.
Export CSV — the whole prompt list, with how big the sample was. CSV export starts on the Starter plan. See Exports and API.

A link from the dashboard opens Gap Analysis on the same window and filters. Changes you make here aren't saved in the page's address, so they reset when you reload.

What gap analysis doesn't tell you

Why specifically AI chose the competitor over you. AI doesn't explain its reasoning. Gap analysis tells you what — the why you have to investigate by reading the responses.
Whether closing the gap will move sales. AI visibility is a leading indicator. Closing gaps often correlates with pipeline lift, but it's not a direct revenue line.

Plan availability

Gap Analysis is on every plan, including the free trial. CSV export starts on the Starter plan.