Gap analysis

Overview#

Visibility tells you how present you are across your topics. Gap Analysis picks up where that leaves off: it lists the topics where AI answers without you, says which fix each one points to, and tracks the work you start until a measured run cites you.

A gap is one topic on one engine, and it is a gap only when you are missing from every question of that topic on that engine. You can be healthy on Perplexity for a topic and completely absent from it on ChatGPT, and since the two draw on different sources, collapsing them into one number would hide the detail that tells you what to do next.

A question counts as missed for the week when any check of it that week left you out. One sighting among several misses does not hide the gap.

Panels and controls#

The four numbers#

Each number at the top is a card, and selecting it opens the list it counts.

• Needs you counts the topics with at least one question that has nothing started, or only a suggestion. It matches the tab of the same name.
• Questions you’re missing counts the individual questions you are missing across every topic on the page, whichever tab the topic is in, including topics you only partly cover.
• Open gaps counts topic and engine pairs, and it is the last point of the chart at the bottom. Beside it is how many closed and opened since the previous measured week, compared only over the pairs measured in both weeks.
• Work in flight counts the topics that are drafted or published and still being re-checked.

All engines narrows everything to one engine at once: the numbers, the list, the card, the sources and the chart. Only engines your plan measures are offered.

The list#

The list is one row per topic, and the tabs follow the work on its questions.

A row shows the engines you are absent on, how many questions are waiting, how long the gap has been open without a break, and the fix. The small ribbon under it has one bar per week, most recent on the right, coloured by whether you were absent or cited that week.

Order changes how the list is sorted, and the rule in force is printed under the control. Best next move puts topics with a brief already written first, then the longest unbroken run, then the most engines. Longest open and Most engines are the other two. Search volume never orders the list. The first row under Best next move is marked Start here, and its card says why.

Nearly covered is a separate band below the gaps: topics where you are cited for some questions and missing on others. They are not gaps, they are not counted in Open gaps or on the chart, and they are usually the nearest wins.

The card#

Selecting a row opens its card. It leads with one recommendation, marked Recommended: what to do, one or two sentences on why, and the step itself with its button. The evidence is underneath.

The recommendation comes from the kind of page AI cites most across every missing question in the topic. Where no one kind leads, it recommends an article of your own, since that is the fix that always applies.

Under the step, Other ways to close this lists any other fix the cited pages support, each with what it rests on, such as the number of directories cited. Choosing one swaps the card to that fix, marked Your choice, and Back to the recommended fix returns.

Where questions in one topic point to different fixes, the row says how many.

For an article or a reply, the card lists every question of the gap the draft has to cover, with the Working title and the focus keyword, before anything is generated. You can leave a question out. Copy the brief copies it, and the button sends it to the editor. Nothing is generated until you approve it there, and the draft stays linked to the questions in its brief. The topic moves to In progress once every one of its missing questions has work started.

Three fixes end outside Seerly: a listing, a reply, an edit to your own page. Each has a button to say you have done it, which moves the topic to Watching and stops Content Scout suggesting content for it. From there Try another fix starts again if a later run still shows the gap, and Not done after all takes the mark back.

Strengthen your page asks Do you have a page that should answer this? If you do, pick it, compare it with the pages AI cites instead, and Audit this page opens Smart audit on it. If you do not, the card offers Publish an answer instead or putting the gap aside.

Not relevant puts a gap aside with a reason. It moves to its own tab and Content Scout stops suggesting content for it, but it still counts in Open gaps and on the chart, because AI still answers those questions without you. Put back in Needs you restores it.

Seerly reads the pages AI cited (sorting them into kinds and fetching their keywords) by itself after every analysis. Refresh evidence appears beside the Gaps and Sources switch only when there is something it has not read yet: a newer analysis, or cited pages that were never sorted. It reads them for every gap at once and does not re-ask the AI engines, so the gaps and the numbers stay as they are. It takes a few minutes, the button reads Refreshing evidence… until it finishes, and the list reloads by itself.

Content Scout works from the same gaps. It writes one suggestion for a topic, covering up to six of its missing questions, longest open first, and follows the fix on the card: an article where the card says Publish an answer, a reply where it says Join the conversation. It writes nothing for a topic whose fix is a listing or your own page. When it has already written a suggestion for a topic, the card offers Draft it and Read the brief instead. It is the same item you would find under Content suggestions.

Under the plan, What AI gave instead of you lists the most-cited pages, each with its domain, the kind of page, a badge marking it Competitor, Indirect competitor or Third party, and how many of the topic's questions it was cited for. The list is a sample and says so: it states how many pages were cited in total, and the fix is judged on all of them. What those pages all cover lists the terms those pages share, ordered by how many of the cited sites use each one. The dots beside a term show that share, and the key above the list says what one, two and three dots mean.

The life of a gap

your moveSeerly runs this

1. 1

   A gap opens

   A topic where every question misses you on an engine

2. 2

   The card names a fix

   Judged on every page cited across the topic's questions

3. 3

   You start it

   A draft, a reply or a listing, and the topic leaves Needs you

4. 4

   The next runs re-measure

   Published work sits in Watching until a run cites you

↺

a gap that stays open after publishing is telling you the fix missed

Sources#

Sources shows the same gaps grouped by the site winning them, limited to sites cited across more than one of your gaps. Sites are grouped by the move they call for: Get listed, Join the conversation and Read before you write. Each row names the gaps the site touches, and selecting one opens that gap's card. The button on a row opens the page AI cited, not the site's home page.

The chart#

Open gaps over time is one bar per measured week. The coloured part of a bar is the topic and engine pairs you were absent on, split by engine, and the green part is the pairs measured where you were cited, so each bar shows open gaps out of everything measured that week. The number on a bar is that week's open gaps; with a long history only the newest week carries its number. Pointing at a bar gives the count per engine, how many gaps opened and closed since the week before, and whether any weeks before it went unmeasured. Weeks where the set of engines changed are listed under the chart, because a move there may be the scope changing and not your visibility.

What changed names the gaps behind those counts: each one that opened or closed in the last 4 measured weeks, and the work still waiting on a re-check. Every row opens its gap. The line at the bottom points at the gap that has been open longest.

When your tracked questions change#

A past week on the chart counts the questions that were measured in it, even if you have since removed them. Changing your questions does not erase their history.

• A small change, such as one question swapped, is noted when you point at that week.
• A large change puts a dashed line on the chart at that week, and one line under the chart says when the questions changed and by how much. Bars before the line are drawn slightly fainter, because they were counted over different questions and are not directly comparable. A change is large when fewer than 8 in 10 of the previous week's questions were measured again, or the set grew by more than a quarter.
• A gap whose questions you removed is listed in What changed as removed, never as closed.
• New questions are measured in your next analysis. Until then the page says how many are waiting.

If you replace all your questions, the list is empty until the new ones are measured: Your new questions have not been measured yet. The chart keeps the earlier weeks and ends with an outlined bar for the week to come. The four numbers keep the last measured figures and say which week they are from. The list, the numbers and Content Scout only ever use the questions you track now.

Interpreting results#

How long a gap has been open matters more than how bad it looks. Engines vary between runs, so a gap that shows up once may be nothing at all, whereas one with a long unbroken run has settled into a position. The same logic applies to the pages cited in your place: a page cited for one question could be chance, while one cited for every question in the topic is what you are competing with.

Read the kind of page as carefully as the domain. If the pages winning a topic are articles and the only thing you have is a product page, you are looking at a format problem, and polishing the product page will not close it.

Check who is winning, too. A surprising share of the time it is not a competitor at all but a review site, a directory or a forum thread, and getting onto one of those is usually quicker and holds up better than trying to publish your way past it.

A gap can close with nothing published from it. When Seerly has no record of work on a closed gap the card says so, because an answer that changed on its own is worth knowing about before you spend effort on a similar topic.

Best practices#

• Persistent gaps over new ones. A long unbroken run is telling you something. A gap that appeared this week may be variance.
• Commercial weight over gap count. One open gap on a topic that brings in revenue deserves more attention than five on topics that never convert.
• Sources before pages. If the same site keeps turning up across several gaps, getting onto it closes more ground than any single article will.
• The fix the evidence points to. Read what the cited pages are before deciding what to write, and treat a topic marked as mixed as more than one job.

Next steps#

1. Open Needs you and start with the row marked Start here.
2. Read the plan on its card, including the count behind the fix.
3. Read What AI gave instead of you and note which pages are cited for every question.
4. Use Edit the brief first to check the question, the focus keyword and the pages to beat.
5. Start the fix from the card's button so the draft stays linked to the gap.
6. Open Sources to find the sites that win several gaps at once.
7. Come back after a few runs. Closing a gap is not instant, and one run either way will not tell you whether it worked. See Reading trends.

What to keep in mind#

Gap Analysis only knows about the topics and prompts you track, so a question nobody has configured cannot show up as a gap. That is the awkward part: an unmeasured gap and no gap look identical from here. See Prompts.

The count of open gaps follows the definition above. If you are comparing against figures you noted before October 2026, expect the newer counts to be somewhat higher: a question used to count as missed only when every check that week left you out, which hid gaps where you were seen once among many misses.

The fix is read from the pages AI cites, not from your own site. When competitors' product pages lead, the card asks whether you have a page that should answer the question, because that is something only you know.

It also tells you what is being cited, not why the engine chose it. You can see that a review site wins a topic every time. Whether that is down to its authority, its structure or its age is a judgement you still have to make.

Closing a gap is not something the product can do for you. It shows you the target, hands you the brief and tracks the result, and the work of getting cited still happens on your site and on other people's.