Semrush Keyword Gap, and the only bucket worth opening

Keyword Gap is the report I would keep if Semrush took the rest away. It is also the one people run once, screenshot, and never act on, because the interface offers four buckets and only one of them is a task.

Set it up at the right level

You add up to five domains in total: yours and four competitors. Pick real competitors, meaning sites that rank for things you want, not the biggest names in your industry. Benchmarking a five-page SaaS site against a media giant produces a list of things you will never do.

The setting most people miss is the comparison level. You are not limited to whole domains. You can compare a subdomain, a subfolder, or an exact URL against theirs.

That’s what turns this from a strategy report into a page-level tool. Point your /pricing/ page at their /pricing/ page and the gap becomes a specific list of terms one page is missing, which is a brief rather than a project.

Three buckets are context, one is the work

Missing is the deliverable. Every competitor ranks for it and you do not appear at all. These are either pages that don’t exist yet or pages that exist and are invisible, and the difference decides whether it’s a writing job or a fixing job. Check before assuming.

Weak is the fastest money. You already rank, just worse than everyone else. There’s a page, it has some authority, and the work is improving rather than creating. Sort Weak by volume and you have this month’s priority list without writing a word of new content.

Shared is a saturated table stakes list. Useful once, for confirming you haven’t missed something structural.

Strong is for the client report and nothing else. It is where you already win, and rereading it does not produce work.

Almost everyone spends their time in Shared and Strong, because those are the buckets that feel reassuring. The output lives in Missing and Weak.

Filter Missing before you believe it

The raw Missing list overstates the opportunity, for two reliable reasons.

Their brand terms. If all four competitors rank for their own product names, those show as keywords you’re missing. You are not missing them. Exclude branded terms first.

Their market. A competitor earning in a country you don’t sell in will show you a list of keywords in the wrong geography. Set the database to your market before reading anything.

After that, apply a volume floor and sort by it. What survives is genuinely actionable, and it’s usually a fraction of what the report opened with.

The onward step is the Keyword Magic Tool: take the ten best Missing terms and expand each into its cluster, because a gap keyword is rarely one page on its own.

Backlink Gap is the same trick, applied to links

Worth knowing about because it uses the identical logic on referring domains: who links to your competitors and not to you.

That output is better than any generic prospect list, because every site on it has already linked to something in your category. They are demonstrably willing. You are not cold-pitching somebody who has never linked to a company like yours; you are asking somebody who has done it three times already.

Run it after Keyword Gap, on the same competitor set, and the two lists tend to explain each other. A keyword you’re missing plus a domain that links to everyone ranking for it is usually one page and one email, not two separate projects.

The mistake that makes the whole report useless

Choosing competitors by reputation instead of by overlap.

The instinct is to enter the four biggest names in your industry. What comes back is a Missing list of several thousand keywords, most of them belonging to a company with two hundred employees and a decade of authority. The report is technically correct and completely unusable, so it gets screenshotted once and abandoned.

Pick competitors whose Authority Score is within roughly fifteen points of yours, and who rank for things you already half-rank for. The Missing list that comes back will be shorter and every row on it will be plausible. A gap you can close is worth more than a gap that’s accurate.

Then rerun it quarterly on the same set. The value compounds, because the delta between two runs tells you whether you’re gaining or losing ground on people your own size, and that is the only competitive metric that reliably means anything.