Week four on a domain that had no history in August. Search Console recorded 465 impressions in the seven days to 12 September, up from 285 the week before, and zero clicks. Not a low click rate. Zero.
This is the weekly public half of our Sunday review, and this week the interesting number is the one that did not move.
Where the impressions came from
413 of the 465 landed on one page, the free SEO audit report tool, across 28 queries that are all the same intent: seo analysis report pdf, free seo audit report pdf, free seo report pdf, seo audit report pdf, seo pdf report, technical seo audit report. Its average position across the week was 69.
Every one of those terms improved week over week, and impressions on each roughly doubled.
| query | impressions | last week | position | before |
|---|---|---|---|---|
| seo analysis report pdf | 70 | 17 | 65 | 69 |
| free seo audit report pdf | 54 | 30 | 71 | 75 |
| free seo report pdf | 46 | 25 | 74 | 79 |
| seo audit report pdf | 44 | 14 | 69 | 73 |
| seo pdf report | 35 | 18 | 87 | 89 |
| technical seo audit report | 23 | 8 | 82 | 85 |
| seo audit pdf | 19 | 8 | 56 | 67 |
Position 56 is page 6. Position 87 is page 9. The click-through rate at that depth is not small, it is effectively zero, and the doubling of impressions is Google showing us to more people who will never scroll far enough to see us. Rising impressions with no clicks is not traction. It is being auditioned in an empty room.
Who is in the room instead
We checked the first page for our three biggest terms, from a UK search, signed out, on 13 September. The pattern was the same on all three.
For "seo audit report pdf": Scribd first, then a Whatagraph report template, CrawlRaven, a SlideShare deck, an SEOptimer sample report, Brady Mills, SEOmator, a RoastMyWeb template and a Swydo template.
For "free seo audit report pdf": Brady Mills first, then Zadroweb, DgtlMart, SEOmator, a Scribd document, Whatagraph, SlideShare, Semrush and CrawlRaven.
Two things stand out, and neither is what we assumed when we built the tool.
Roughly two thirds of page one is a sample or a template: a document somebody uploaded to Scribd or SlideShare, or a blank report template from a reporting SaaS. The people searching these terms are not all looking for an audit. Many are looking at what an audit report is supposed to look like, often because they are about to produce one for a client.
The remaining third are agency lead magnets: a free audit form that emails you a report in exchange for your address.
We built the second kind. Our page offers a crawl of your site and emails a PDF. That is a reasonable thing to want, and it is not what most of this query family is asking for.
What we changed, and what we are not doing
We published the sample, on Sunday afternoon, about two hours after reading the report that produced these numbers. The full audit report is now printed as a page rather than hidden behind a form: every section, the findings table, the page by page crawl, the live index check, downloadable as a PDF with no email address, with the option to run one on your own site underneath it. It is at rank.fast/tools/seo-audit/sample/.
It is a real report about a real website, ours, and it is unedited. It scores our own site 98 out of 100 and then criticises our homepage for a vague call to action, which is a fair comment we left in. That is the one advantage we have over a blank template: a template shows you the shape of a report, and this shows you what one actually says.
What we are not going to do is chase the ranking with volume. We are not publishing ten variations of the same page, and we are not buying our way up this particular hill. If the sample page is the right answer, it will show up in this same report in a few weeks, and if it is not, we will say so here.
The rest of the week, briefly
Our entity page, published nine days ago, now sits at position 6 on the US desktop results for "rank fast" and position 5 for "rank.fast". Six companies with versions of our name still rank above us on the bare term. It has also started collecting its own queries, including someone searching for one of those namesakes by domain, which is exactly the disambiguation job it was written to do.
All 25 of our own pages are indexed.
In the guest post pilot, three of the five control placements are still not indexed, at day 13 to 16. All five treatment placements were indexed within a day. We will not read anything into that until the 30-day checkpoint on 27 September, but it is the number we are watching most closely, because it decides whether the product we are building has a reason to exist.
What this proves, and what it does not
It shows a new site being shown to more people each week for a coherent set of queries, from a standing start four weeks ago, with no clicks yet because the positions are too deep to produce any.
It does not show that we are ranking, and it does not show that the audience for these terms wants what we built. On the evidence, a good share of them want an example report, which is a different product decision, discovered by reading a report rather than by guessing.
Numbers here are from Google Search Console for the seven days ending 12 September, with the previous seven days for comparison. Search Console data lags about two days. The page-one results were checked signed out from a UK search on 13 September and will have drifted by the time you read this.