A breakdown we had never requested showed 184,133 impressions from translated results

We found this by requesting Search Console's search-appearance breakdown.

What was sitting there

Search Console carries a dimension describing how a result appeared on the results page. It is available in the interface, documented, and free. On this property it had never once been requested.

Requesting it showed 184,133 impressions in twenty eight days from translated results, at an average position of 8.37, producing 255 clicks. We had not examined these impressions separately before, although Search Console's overall totals for the property already counted them.

Two findings, and they point opposite ways

The first is scale: 184,133 impressions from translated results, on a property built for one domestic market with no translated content and no international targeting.

The second is that these impressions were clicked at 0.14 percent, at an average position of 8.37. Why so few were clicked is not measured here.

So this is not a discovery of demand waiting to be captured. It is a discovery of a real surface that performs badly, which is a different and less exciting thing, and it is what the number says.

Why there is no percentage here

The obvious next sentence is what share of the site that represents. No share here can be shown to be right, so none is given.

This number is a property-level slice by appearance type. It carries no URL breakdown at all, so there is no way to tell whether it includes the fragment-row impressions that our page-level denominator excludes. Three plausible denominators give 8.96, 8.85 and 8.73 percent, and choosing between them would itself be the mistake: a share needs a numerator and a denominator counted over the same population, and we cannot demonstrate that for any pairing here.

The counts are exact and they are the finding. 184,133 impressions, 255 clicks, 0.14 percent, average position 8.37.

A trap in this number

Search appearance is a separate breakdown of the site's search results, not a separate slice of them. With no URL breakdown, there is no way to tell which pages these 184,133 impressions belong to, so they may overlap the page groups reported elsewhere on this site, such as the three clusters in finding 01.

That is why they must not be added to those groups as though the two were separate populations: the sum could count the same impressions twice. It is an easy mistake because both figures come out of the same tool and look like the same kind of number. They are not.

Why it matters even though it was rarely clicked

Until it was requested, no summary, report or decision on this property had looked at translated results separately, although their impressions were always inside the totals. There is no way to know how many other breakdowns are in that state without requesting them.

How to check your own

Open Search Console and request the search appearance dimension explicitly. Then do the same for the dimensions you have never exported: country, device, search type, date at day granularity rather than aggregated. Each one takes a minute. On this property the first one produced impressions from translated results that had not been examined separately before.

Full method

Google Search Console search appearance dimension, requested for this property on 2026-09-12 over the same window as the page export. That aligns the window; it does not align the population, which is why no share is stated - see the limit below.