Dated, not remembered
Every entry in the timelines below is a timestamp taken from the production record. Where the interval is the point, we publish the measured interval rather than a rounded one.
Verified outcomes
Audit results, with
the receipts attached
Two sites, audited by SEOReport, told from the production record. Every figure here names the window it covers and what it was measured against — and the claims we cannot support are listed too.
Windows and denominators
Every rate, total and score states the period it covers and the thing it was measured against. A number without a denominator is a slogan, so you will not find one on this page.
The caveat travels with the number
Where a figure moved for a reason other than the obvious one, we say so beside the figure. A disclosure at the bottom of a page is a disclosure nobody reads.
Story one — Sphinx Capital
The audit that found the wrong site
In mid-2026, Sphinx Capital's apex domain quietly redirected every visitor to a mirror of the site held on its hosting provider's infrastructure. Nothing looked broken. The pages loaded, the content was there, and the address bar behaved. An audit pointed at the apex domain would have cheerfully scored the mirror and filed the result under Sphinx Capital's name.
That is the failure we most want to catch, because it is invisible from the outside: not a missing tag, but the wrong site being measured. SEOReport detected the offsite redirect on 14 August 2026, and the finding was confirmed two days later.
Rather than treat it as one site's problem, we shipped the detection. Offsite-redirect and mirror detection went into the product on 16 August 2026, alongside a site-ownership signal, so every audit since then follows the domain to wherever it actually resolves and says so when that is not the site you asked about.
| Date | What happened |
|---|---|
| 14 August 2026 | An audit detects the apex domain redirecting offsite to a hosting mirror. |
| 16 August 2026 | The finding is confirmed, and offsite-redirect and mirror detection ship into the product for every site we audit. |
| 24 August 2026 | With the domain resolving to the real site again, a full audit scores it 95 out of 100. |
Window: 14 to 24 August 2026, three dated points from the production record. The score is SEOReport's own 0 to 100 site score from a single full-site audit of the live domain on 24 August 2026. It measures one site at one moment on our scale; it is not a search ranking and it does not compare across sites.
Post-recovery audit, 24 August 2026
95 / 100
Sphinx Capital is named here with its approval. The story is about the product finding a problem, not about anyone failing to notice one.
Story two — the Activate website
43 contradictory dates, fixed in eight minutes
SEOReport audited the Activate website on 8 August 2026. The audit reported 43 pages whose last-modified date fell earlier than their publication date — a contradiction a search engine reads as a site that cannot be trusted about its own freshness. The finding named the pages, which is the difference between a warning and a work order. The fix was committed eight minutes later.
| Timestamp (UTC) | What happened |
|---|---|
| 8 Aug 2026, 23:00 | An audit reports 43 pages whose modified date precedes their published date. |
| 8 Aug 2026, 23:08 | The fix is committed — eight minutes after the finding landed. |
| 9 Aug 2026, about 01:17 | The fix reaches production. |
| 9 Aug 2026, 01:36 | A re-audit puts the content health category at 100 out of 100, with date integrity passing and no affected pages left. |
Window: 8 to 9 August 2026, four timestamps from the production record, with the middle of the chain independently confirmed by a separate run. Denominator: 43 affected pages before the fix and none after, and the content health category on SEOReport's 0 to 100 category scale, which moved from 0 to 100.
Three weeks earlier: the index nothing could crawl
On 15 July 2026, an audit of the same site found its article index was a browser-assembled skeleton — a page that builds itself after loading and offers a crawler no links to follow. The same audit flagged a duplicate canonical, a missing llms.txt, and sitemap last-modified values that Search Console had rejected.
The site's content programme had launched the day before. The server-rendering fix was committed 26 minutes after the audit and deployed two minutes after that. Until the audit ran, the programme it was publishing was invisible to the crawlers it was published for.
What the search footprint did next
Across the three months that followed, the site's indexed search footprint grew. We publish it as impressions per day, averaged over each calendar month, because raw monthly totals make a short month look like a decline.
| Month (2026) | Impressions per day |
|---|---|
| June | 4.4 |
| July | 12.0 |
| August | 49.3 |
Window: June, July and August 2026, one Search Console property. Values are impressions per day averaged across each calendar month, so August is 11.2 times June. Clicks over the same window went from 0 to 52.
Read this as footprint, not ranking
Over the same window the average position widened from 12.6 to 21.7. The site now appears for many more queries than it did, and the newer ones sit further down the page, which pulls the average the other way. That is footprint expansion, not evidence that any individual query ranks better. We publish it because leaving it out would let the impressions figure mean something it does not.
Who arrived through that footprint
On 24 August 2026, a high-volume user signed up to Activate after arriving from a Google search. The signup is the whole of the claim. It is not a statement about how much the product is used, it is not a statement about money, and the user is not named. Two things are in the record for that day — the date and the referrer — and those are the two things published here.
The number we are not leading with
The site's overall score also moved over those 80 days, from 63 to 100, and we are deliberately not putting it on a chart. The final step from 96 to 100 crossed an engine release that removed two false positives — false positives we discovered on this very site — so part of that last climb is our own scoring getting more honest rather than the site getting better. The claim that stands without an asterisk is the category one: content health, 0 to 100, in eight minutes.
The longer arc — the same site as story one
From 44 out of 100 to a $10,000,000 application
Before any of the above, in March 2026, a free audit scored Sphinx Capital's site 44 out of 100. The site was rebuilt on the back of that audit, and the new platform went live on 27 March 2026.
Thirty-six days after the audit, on 9 April 2026, the site received a $10,000,000 loan application from a visitor whose first recorded touch was a referral from Google. We publish the interval and the referrer because that is what the record holds. We are not claiming the audit produced the application, and the borrower is not ours to name.
Window: the 4 March 2026 audit to the 9 April 2026 application — 36 days. The 44 out of 100 is the site score from the pre-rebuild audit on SEOReport's 0 to 100 scale. The referrer is the first-touch referrer recorded against the session that submitted the application.
What this page does not claim
A proof page is only worth reading if it is also willing to say where the proof stops.
- We do not claim an audit caused the outcome printed next to it. What we publish is the dated sequence; the judgement is yours.
- Impressions are not rankings. A wider footprint usually drags average position the wrong way, and on the Activate website it did exactly that.
- A signup is a signup. Where we report one, we are not reporting usage, spending or an outcome, and we do not name who signed up.
- A score measures one site at one moment on our own 0 to 100 scale. It is not a search ranking, and it is not comparable between sites.
- Where a number moved partly because our engine changed rather than the site, we say so — see the score arc above.
- Figures we could not reproduce from the production record are absent from this page, including several that flattered us more than the ones we kept.
Run the same audit on your own site
The audit behind everything above is the free one. Enter a URL and read the findings for yourself.