Keyword Gap: The Report That Shows What Your Competitor Answers and You Don't

Somewhere there is a list of every question your market asks and your site does not answer.

That list has a name. It is called keyword gap, and you can pull six months of content planning out of it in one afternoon.

What the report does

You enter your domain and up to five competitors. The tool crosses the keyword profiles and gives the overlap back, organized. ¹

Semrush splits the result into six categories. The name of each one is already the diagnosis: ²

Missing is the list of questions the whole market answers and you do not. Start there.

Picture an entrance exam where you got right everything you studied. The low grade comes from the questions you never read. This is called the coverage gap.

A real example, and it is not a client’s

I ran the comparison between estrategiaconcursos.com.br and grancursosonline.com.br, two large Brazilian test-prep sites. Neither one is a client of ours. And that is what makes the example useful. Both are public, and you can repeat the comparison right now. ³

What showed up in the Missing column was more interesting than the list of terms. Whole questions appeared, the kind someone types when they have not decided to buy anything yet.

That is the pattern. The gap is almost never in the commercial term, which everyone already fights over. It sits in the question that comes three steps earlier.

How to read the list without getting lost

The output has thousands of rows, and the common mistake is trying to attack all of it. The filter I use has three passes.

First, cut by intent. Separate the queries that describe a problem from the ones that describe a product. The problem queries go to content. The product queries go to a service page.

Second, group by subject, not by term. Twenty variations of the same question are one article, not twenty. One article that answers well covers the whole group.

Third, sort by distance. A subject next door to what you already publish climbs faster than a distant one, because the site already has ballast there.

The publishing order falls out of that: a neighboring subject, with volume, that every competitor covers and you do not.

The warning that has to come with it

Volume and difficulty in a third-party tool are not measurement, they are a model. The number is an estimate the tool calculates, and not data Google handed over. ⁴

That does not invalidate the report. It changes the use: it is for prioritizing, not for promising. The relative order between subjects is reliable. The absolute number of forecast visits is not.

Always cross it with Search Console after publishing. There the data belongs to the owner of the search engine.

The subscription limits

Worth knowing before you build a routine around it. The Pro plan returns up to 10,000 results per report, Guru up to 30,000, and Business up to 50,000. ⁵

In a large market, the Pro ceiling cuts the list short. The way around it is running by site section instead of running the whole domain at once.

Frequently asked questions

How many competitors can I compare at once?

Up to five domains besides your own. ¹

Which category should I look at first?

Missing, which gathers the queries where every competitor ranks and you do not. It is the clearest gap in the report. ²

Are the volume numbers exact?

No. They are the tool’s estimate, and not a Google measurement. They serve to order priorities, and after publishing the check happens in Search Console. ⁴

Do I need the most expensive plan?

It depends on the size of the market. The limit is results per report: 10,000 on Pro, 30,000 on Guru, 50,000 on Business. You can work around it by running section by section. ⁵

Notes and sources

1. Semrush, knowledge base, article “Keyword Gap”, semrush.com/kb/28-keyword-gap, consulted on September 1, 2026. Comparison of up to five competitors.
2. Definition of the report’s six categories: Shared, Missing, Weak, Strong, Untapped, and Unique. Same source as note 1.
3. Comparison run on September 1, 2026 between estrategiaconcursos.com.br and grancursosonline.com.br, two large test-prep sites in Brazil. Neither domain is a client of SEO Genome, and both were chosen for being public and reproducible.
4. Caveat stated by SEO Genome, and not by the tool: the knowledge base consulted carries no explicit estimate warning. Volume and difficulty in a third-party tool are the output of a proprietary model, and not data supplied by Google.
5. Limits by subscription: Pro 10,000 results, Guru 30,000, Business 50,000. Same source as note 1.