A set of pages went from one internal link each to forty four. Fifteen days later, nobody had checked whether it worked.
The job was sound. Kruze’s removals pages were orphans, sitting on the edge of the site like a shed nobody has a key for, so we wired them into the junk-removal pages that already rank. The document that ordered the work even wrote its own verdict in advance. If linking alone does not lift the man-and-van page off page three, it said, the gap is authority.
That is a good test. Falsifiable, specific, written down before anyone knew the answer, which is the rare bit.
Then it sat there. Nobody put a date on it.
Here is the part that stings. Measured this morning, the pages had gone backwards. Average position 27.8 to 30.0. Impressions up 28 percent, so Google is showing them to more people, and those people are declining with impressive consistency. Click-through stayed nailed to 0.5 percent. Home removals slid from 31.5 to 41.7, which is less a dip than a resignation. Six of the biggest queries hold 945 impressions between them and have produced, in total, no clicks at all.
Not one removals page is in the top ten. Eight junk pages on the same site, same domain, same everything, sit at 10.5 or better.
So linking was never the missing piece. Internal links do not manufacture authority, they shuffle the authority you already have around the house, and if nobody outside is pointing at you there is nothing to shuffle. That answer had been sitting in Search Console for a fortnight, waiting for somebody to open the tab.
Writing the falsifiable test is not the hard half. It feels like rigour, it reads like rigour, and it is worth almost nothing until someone puts a date in the diary beside it. Until then it is not a test at all. It is a nicely-worded opinion in a document, ageing quietly.