Results · August 2026
One month of automated grunt work on a 3M-pageview tech publication.
The editors kept the judgment and the writing. The automation kept the calendar, found the gaps, and placed the links. Here is what changed from July to August.
- +90%
- affiliate click-out rate
- +242%
- affiliate earnings
- +21%
- pageviews
share of pageviews that click a product link, before and after
July to August, one network
July to August
One honest line on that last number. Traffic also rose in August because Google Discover sent more readers. The click-out rate is the automation's own result: the same readers clicked almost twice as often. The earnings rise is both together.
What the automation did
- Read every story as it published, and the back catalogue behind it.
- Proposed the links the product links and the related-story links each piece was missing, on rules the editor set.
- Waited for a yes every proposal sat in one dashboard until an editor accepted or rejected it. Nothing published on its own.
- Applied the accepted ones on the site through its own editor account, so every edit is tracked and can be reversed.
- Wired the guides it proposed which stories belong under one hub article and which should link to it.
By the end of August it had placed 1,429 links and covered about two thirds of all pageviews.
What the queue looks like
Add product link on the Poco F8 Pro review
Product · rule: first mention, in-stock only
Link the Xiaomi 17 launch story to the Xiaomi phones guide
Hub · 4 sibling stories already linked
Calendar: Snapdragon Summit, 23 Sep
Read from last year's coverage
What it did not do
- No AI writing. Not one sentence on the site was written by it.
- No change to what the editors chose to cover.
- No access to the editors' own accounts. It works through one editor user the publisher creates and can revoke at any time.
Want to see it on your stories? Reply to the mail that brought you here.
Measured on the publication's own Google Analytics and affiliate reports, July against August 2026. Click-out rate compares the ten days before the automation covered most pages with the last twelve days of August. Percentages only, by the publisher's choice.