AI gaps screen

AI → Gaps files every in a by where you stand on it against your competitors. Visibility tells you how you are doing overall; Gaps tells you which specific questions to do something about.

It is the same measurement as Visibility, read a different way. Nothing extra is measured, so narrowing the date range, topics or platforms changes this screen exactly as it changes that one.

Each channel has its own gaps screen in its own currency: prompts here, and queries on Search gaps.

The four standings

Every question the window answered lands in one of four groups.

Standing	What it means	What it is for
Missing	A competitor is named here and you are not	The work list. These are questions you are absent from entirely.
Behind	You are both named, a competitor more often	Usually the quicker wins: you already appear, you are just not the first answer.
Ahead	You are both named, you more often	Questions you are currently winning.
Uncontested	You are named and no competitor is	Ground you hold alone - and the ground a competitor would gain most by taking.

Pick a standing to focus the whole screen on it: the chart brightens those dots and fades the rest, the competitor list is recounted over just those questions, and the table at the bottom narrows to match. Pick it again to see everything.

Two questions get no standing, and both are counted under the cards rather than hidden:

No tracked brand was named. The question was answered and nobody on your roster came up. Nobody is winning it, so there is nothing to be behind on.
Not asked in this window. Widen the date range, or wait for the next measurement.

Rates, not single answers

Both sides of every comparison are rates across the whole window, not one answer. A question only counts as Missing if competitors are named there reliably while you are not, so a single unlucky answer cannot move it.

That is also why the numbers are percentages rather than counts: assistants are asked the same question repeatedly, and how consistently a brand comes up is the thing worth acting on.

The workspace

Each dot is one question: how often you are named up the side, how often your strongest competitor is named across the bottom. Higher is better; further right is more contested.

The dashed diagonal is the dead heat. Anything above it is a question you are named on more reliably than a competitor; anything below it is still up for grabs. Dots against the bottom edge are the Missing ones - your competitors are being recommended and you never appear.

Taking questions off you

Which competitor is winning the most questions, counted rather than averaged. A competitor near the top of this list is the one to read next, and the name beside each row in the table below tells you which questions are theirs.

The list follows the standing you pick. On Missing or Behind it counts who is winning those questions; on Ahead it flips to who is closest behind you, which is who you are defending against; on Uncontested there is nobody to list, because no competitor is named at all - that is what makes those questions uncontested.

Every question

One row per question, widest gap first.

Standing is the verdict from the table above.
Assistants shows the same verdict on each assistant separately, as one dot per column, matching the marks in the heading. This is the most useful column on the screen: missing on one assistant and ahead on another is a content problem on one surface, while missing on all of them is a positioning problem. A hollow ring means that assistant was never asked this question.
You and Best rival are the two rates being compared, with the leading competitor named.
Gap is the difference in percentage points. A negative gap is your lead.
New marks a question that has moved against you since the previous window. These are worth reading first: a total that has not changed can still hide questions that swapped sides.

Click any row to read the answers it was measured from. That is where you find out why a competitor is winning - which of their pages the assistant used, and what it said about them.

Choosing who you compare against

By default you are compared against every competitor on the workspace's roster, so Missing means "somebody is winning this and it is not you".

Narrow the Brands filter to one competitor and the whole screen becomes a head-to-head: every standing is recalculated against that brand alone. This is the fastest way to answer "where is this specific competitor beating us?". The line under the cards always names who the comparison was against.

Your own percentages never change when you do this. They are always measured against every brand tracked in the workspace; only the comparison moves.

What to do with it

Start with Missing, sorted as it arrives - widest gap first.
Check the Assistants column. A gap on one assistant is usually a content or citation problem; a gap on all of them usually means the workspace does not associate you with that topic yet.
Open the row and read the answers. Look at which pages the assistant cited for the competitor it named.
Check Sources for the sites feeding those answers, and Strategy for the briefs that turn this into work.

Do not ignore Uncontested. Those questions are yours today, and they are the ones you would most notice losing.