Does entity coverage work? A content score, measured in the open

By Mitch Flindell

Short answer: nobody has shown it, us included. A content score counts how many of the names, phrases and sections the top Google results mention are also on your page. Covering what a reader expects to find is sensible writing, but a higher count has not been shown to cause higher rankings, and Google doesn't publish a score like it. So we measure it. Every Dink that writes or rewrites a page gets a brief: a list of what the top pages all mention, and this page publishes how those changes did. Today the answer is "too few to tell": no Dink with a brief has been checked live yet. Checked 16 October 2026.

What a content score actually counts

Content tools take the pages that rank for a search, pull out the words and names they have in common, and score your page on how many it has. The idea is reasonable: if every page that ranks for "ai seo agency" explains what GEO is, a reader will expect yours to as well, and a page that skips it may look thin.

The catch is what the count can't separate. Pages rank on the same search because they're about the same thing, so of course they share words. That doesn't mean the shared words are why they rank. A score is also easy to raise without making a page better: paste the missing terms into a paragraph and the number goes up. Google's spam policies name keyword stuffing, and a page written to a score tends to read like one.

We use the brief anyway, because it's a fast way to spot something a reader would miss. We just don't trust it, which is why every change it touches is judged on what it did in search.

How Dinkum does it

This runs on its own every time a Dink writes a new page or rewrites a page's main copy.

1. Read the top five Google results for the search the page is meant to win, in the country the site sells to. If our own page is one of them, it's left out. A page that won't load is skipped, and the brief says how many pages it read.
2. List what most of them mention: names, phrases and section headings that at least three of the pages have. If fewer than three pages could be read, nothing counts as shared. It's a plain count, so some rows are filler.
3. Check our page against the brief before the change: "7 of 21". The writer is told to cover each item where it's true and useful to the reader, skip the ones that aren't, and say why. Then to spend the real effort on what none of the top pages have.
4. Check our page again a day after the change goes live, against the same brief. That's the "after" number. It's the live page, not a draft.
5. Judge the change at 14 and 28 days in Search Console. The page's clicks (or click-through, for a title change) are compared with the 28 days before, adjusted for how the rest of the site moved. It "worked" if the number rose 10% or more at 95% confidence. We also note the page's average Google position before and after.
6. Put it in the ledger with every other merged Dink, in one of three groups: the page came to cover more of its brief, it had a brief but didn't cover more, or it had no brief. The groups are pooled across every site Dinkum works on, because one site rarely has enough changes to say anything.

What a brief looks like for a customer, start to finish, is on SEO content briefs. The rest of how a Dink is chosen, made and judged is in how Dinkum works.

How to check what the top pages mention, by hand

You don't need a tool for this. A spreadsheet and an hour per page will do, and the judging part matters more than the counting.

1. Pick one page and the one search it should win. Search it in a private window, from the country your customers are in. Write down the top five ordinary results, skipping ads. If your own page is there, leave it out.
2. Read each page's headings and note what it names. Products, tools, standards, places, people, and the questions its sections answer. Your browser's find (Ctrl+F or Cmd+F) is enough to check whether a page mentions something.
3. Keep what at least three of the five mention. That's your list. One page mentioning something is that writer's idea; three is what readers of this search have come to expect.
4. Mark which ones your page already covers, and write the count down: "9 of 18". Also note the length of the top pages, so you can see that length isn't the goal.
5. Rewrite only where it's true and useful. Add a sentence or a section where a reader would look for it. Skip anything that doesn't apply to you and note why. Never paste terms in to raise the count. Then put most of your effort into something none of the five have: your own numbers, a worked example, a photo, an answer only you can give.
6. Write down the day it went live. Count your page again the next day, from the live site.
7. Judge it after four weeks, against the rest of your site. In Search Console's Performance report, filter to the page and compare the 28 days after the change with the 28 days before. Do the same for the whole site. If the whole site rose 20% and the page rose 20%, the change didn't do anything you can see.
8. Keep a ledger and don't judge from one page. One row per change: the page, the count before and after, the date, and the result. After a dozen or so, look at whether the pages that came to cover more did better than the ones that didn't. Even then, it's a hint, not proof.

A sheet to start from:

A real brief: "ai seo agency"

On 15 October 2026 Dinkum made a brief for one of our own pages, the guide to choosing an AI SEO agency, against the search "ai seo agency" in the United States. Our page wasn't in the top five. One of the five didn't load for our reader, so the brief is built from four.

Every row, as the dashboard shows it:

Read it the way a writer should. Some rows are worth acting on: all four top pages spell out generative engine optimization and mention Google AI Overviews, and the brief found neither on ours. Some are filler from a plain count: "optimization agency" and "Media" aren't things a reader is looking for. And the length gap is not a target. A buyer's checklist doesn't get better at 3,600 words.

This brief was made by hand on 15 October while we built the feature, not by a Dink, so it isn't rechecked and isn't in the ledger below. It's here because it's the first real one, with its real numbers.

The ledger, today

This is what Dinkum's report printed on 16 October 2026, across every merged Dink on the eight sites it works on. Site names are left out.

92 merged changes across 8 sites. No change with a brief has been rechecked live yet, so there's nothing to compare. All 92 went live without one. They all went live on or after 6 October, so none is 14 days old, which is why every one is still waiting for its first verdict.

Search Console verdicts

The latest verdict per change (28 days if scored, else 14). Worked means the main number (clicks, or click-through for a title change) rose 10% or more against what the rest of the site predicted, at 95% confidence. Inconclusive means too little search data to judge.

Rank moves

Median position change is the changed pages' average Google position in the 28 days before the change against after it. Page-one moves are searches that crossed onto or off page one and held for a week, credited to the latest change on that page in the 60 days before.

What's in each group

A rate, a median lift or a median position needs at least 5 measured changes in its group. Until then it says "too few to tell", and so does this page. We'll update these tables from the same report as changes with a brief reach their 14- and 28-day checks.

What this can't tell you yet

• It's observational, not a test. The groups differ in more than coverage: what was changed, when, and on which site. A page that was rewritten thoroughly may cover more of its brief and also be better in ten other ways. If the "coverage rose" group does better, that's a result measured alongside a rise, not proof the rise caused it.
• The samples are small. Five measured changes is the minimum for a rate, and five is still small. One more change can swing a small group's rate a long way.
• Search moves on its own. Seasons, Google updates and a competitor's new page all move clicks. Comparing each page with the rest of its site takes out some of that, not all of it.
• A brief is a machine count of four or five pages. It can't tell a useful name from filler, and the top five for a search change over weeks. A brief is a snapshot of one day.
• Inconclusive is common. A page with too little search data can't be judged either way. Those are counted but left out of rates.

Questions

Does a higher content score improve rankings?

It hasn't been shown to. The top pages share words partly because they're about the same topic. Use the brief to find gaps a reader would notice, then judge the change on clicks, not on the score.

How many top results should I compare?

We use the top five, and count something as shared when at least three have it. With fewer than three readable pages, nothing counts as shared.

Should my page be as long as the top pages?

No. Length follows from what the reader needs. In our example the top pages ran from 1,975 to 4,293 words; a checklist that answers the question in 1,500 words isn't worse for being shorter.

Isn't this just keyword stuffing?

Done badly, yes. Google's spam policies name keyword stuffing. Mention something only where it's true and useful, explain it the first time, and skip what doesn't apply.

When will the ledger say something?

When at least five changes in a group have a 14- or 28-day verdict. Each one needs 14 days live before its first verdict, so it's weeks away, not days.

Do it yourself, or have Dinkum do it every weekday morning

Everything above you can do by hand, and the ledger is the part most people skip. Counting is easy; keeping score on every change for months is the work.

Dinkum does it on its own. Each weekday morning it picks one change for your site. When the change writes or rewrites a page, it reads the top results for that page's search first. It writes the change as a pull request you approve, checks the live page the next day, and judges it at 14 and 28 days against the rest of your site. Every list and every verdict is in your dashboard. See how Dinkum works.