Why so few
An SEO tool that reports two hundred issues has told you nothing, because the reading is now your job. This one holds 10 at a time and holds everything else back — not because the rest was discarded, but because attention is the scarce resource and a list nobody finishes is a list that gets skimmed.
Qualified items that did not get a slot are still there and still ranked. Your dashboard lists them alongside the rest — they cleared the same bars, so nothing marks them out as lesser — and says in the heading how many are waiting; over MCP they come back with the reason each was held. Anything urgent bypasses the cap entirely: a budget that could swallow an outage would be a budget that hides the one thing worth seeing.
Three kinds of item
- Opportunity — demand you are close to and not capturing. A page ranking just off the first screen, a query your competitors hold and you do not.
- Risk — something earning today that is in danger, or has already started slipping. These carry the highest bar for traffic, because interrupting somebody about a page nobody visits is how a product teaches them to ignore it.
- Diagnostic — a measurement rather than a proposal. It says something true about the site and asks for no change. Diagnostics can never be approved or shipped; they are read and marked read.
What is behind one
Opening a suggestion gives you the whole of its reasoning, not a score:
- the measurements that produced it, with the window they were taken over;
- what the detector went looking for, and the exact thresholds every candidate had to clear before it became a finding at all;
- which bar this one cleared, when most open findings on a site clear none — a different question from the one above, and the one that decides whether you are interrupted;
- which independent detectors corroborate it, and which do not;
- what the remedy actually is — for internal links, the exact source page, target and anchor phrase;
- how confident the engine is, and what that confidence rests on.
On a diagnostic there is no remedy, and what stands in its place is the list of hypotheses that could each have produced what was measured — each marked as supported by another detector on your site, ruled out by one, or still untested, with the sentence that settled it.
Where a number is unavailable it says so rather than estimating. A projection of what a ranking change is worth returns nothing far more often than it returns a figure, and always says which of four reasons applied.
The four decisions
Approve
Work you intend to do. This records intent and nothing else — it does not mean anything changed. Reporting the change as shipped is the separate step, and it is the one that starts measurement.
Dismiss
Rejection, with a reason. This is the only negative signal the engine ever receives, and it is expensive on purpose: it suppresses this suggestion for 90 days, and 3 dismissals of the same kind of work switch that kind off for this site entirely.
That is the mechanism by which the engine learns your site rather than a general theory of SEO. It is also why dismissing something you merely have not got round to is the wrong control.
Snooze
The right control for “not now”. It comes back on its own after the days you name, re-scored against whatever the data says then — which is frequently a different suggestion, or none. Snoozing records no judgement about whether the work is worth doing.
Acknowledge
For diagnostics only: read and understood. A diagnostic cannot be dismissed, because dismissing one would be telling the engine its measurement was wrong when what you mean is that you have seen it.
Undoing one
Any decided suggestion can be reopened, which puts it back in the inbox and re-scores it against current data. Reopening an approval that was never shipped is the honest way to clear intent that did not survive contact with the week.
Things that resolve themselves
A finding whose cause disappears — the page was rewritten, the competitor dropped, the query stopped mattering — is marked resolved rather than dismissed. That distinction is load-bearing: a resolution is not a judgement anybody made, and counting it as one would teach the engine that you rejected work you never saw.