Why your action plan is so short: the five floors that keep noise out of it

A short list is the point

The first thing most people notice about their action plan is how little is on it. A tool they used before gave them two hundred items; this one gives them twelve. The difference is not that SearchDone looks less carefully. It is that it refuses to turn small numbers into findings, because a finding you cannot trust costs you the time to check it and the confidence to act on the rest.

Five floors do that work. Each one is a number in the product's configuration, each can be changed for your site, and each exists to prevent a specific false alarm. Here they are.

Floor 1: 50 impressions, or the row is ignored

Before any check runs, every page-and-search row with fewer than 50 impressions is thrown away. Google itself already hides the rarest searches, and what remains below 50 is too small to judge.

The false alarm it prevents: a page shown 20 times with one click has a 5% click rate. One more click makes it 10%. A single visitor doubling a rate is not a signal about your page, it is noise, and every check downstream would treat it as fact.

Floor 2: only pages between positions 3 and 20 count as within reach

The "so close to page one" check flags a page whose average position is above 3 and no worse than 20, and values it as the visits gained by reaching position 3. Above position 3 there is nothing to push; below 20 the page is not really in the running, and a "push" from position 40 is a rebuild, not a fix.

The false alarm it prevents: a page at position 35 with heavy impressions would show a huge theoretical gain, crowd out the pages you can actually move this month, and never deliver.

Floor 3: clicks at least 50% below what the position predicts

The "nobody's clicking" check only looks at pages in the top 10, and only flags one when its real click rate is at least 50% below what that position normally earns. At position 4 the curve expects 7.0%, so a page is flagged at 3.5% or lower.

The false alarm it prevents: the same page at 6% is a little under par, and a little under par is normal. Click rates wobble with the day of the week and with whatever Google puts above you. Flagging every wobble would fill your plan with rewrites of titles that are fine.

Floor 4: a page must hold 10% of a search to count as a rival

Two of your own pages competing for one search is a real problem: Google splits the credit and often neither ranks as well as one good page would. The check flags a search only when at least 2 of your pages each hold 10% or more of its impressions.

The false alarm it prevents: a search shown 1,000 times where one page took 940 and another took 60. Google briefly tried the second page and moved on. Without the floor, almost every search with an incidental second page would be reported as a conflict, and the genuine ones would be lost in the list.

Floor 5: a 30% drop, from at least 10 clicks

The "quietly slipping" check splits your date range in half and compares them. A page is flagged when its clicks fell by 30% or more from the first half to the second, and only if the first half had at least 10 clicks to fall from. The same check runs on its own across your whole site every week, so a page that starts slipping is reported while there is still time to fix it.

The false alarm it prevents: a page that went from 3 clicks to 1. That is a 67% drop and it means nothing. A page that went from 40 to 26 is a 35% drop and it is worth your attention.

How the survivors are ranked

What gets through the floors is sorted by the visits each fix would win, highest first, using the click curve described in the previous piece. Severity comes from the same number: high at 100 or more estimated visits a month, medium at 25 or more, low below that. Findings from your live pages (missing titles, broken links, images with no description, a wrong canonical) sit alongside them with a severity that scales with how many people see the page.

When you should change a floor

The defaults suit a site with a few hundred to a few thousand visits a month. If your site is much bigger, raising the impression floor keeps the plan focused on pages that matter at your scale. If it is much smaller, lowering it will show you more, at the cost of more noise; the honest advice there is to wait for more data rather than to read more into less. Every floor is a setting on your site, and every finding shows the numbers it came from, so you can always check the reasoning rather than take a score on trust.

The full list of checks, including the ones that read your live pages, is on the how the checks work page.