301'd six expired domains into one site over a month. Two moved, four did absolutely nothing

I spent most of February telling byte_broker in his thread that 301s from expired domains were a 2015 tactic. Then I went and spent four hundred quid finding out, because being certain without testing is exactly the thing I accuse everyone else of.

Six domains, bought over January and February, all in or adjacent to one niche, all cheap - between £40 and £110. Redirected into a site of mine, one per fortnight so I could see them separately. Single hop, no chains, redirect to the most topically relevant page rather than the homepage in four cases and to the homepage in two.

Results after between eight and sixteen weeks:

Two of them moved something. One landed on a page that went from about position 22 to position 11 over three weeks, and it's held. The other produced a smaller lift across the section rather than one page, which I'd normally call noise except it happened in the fortnight after and not before.

Four did nothing measurable at all. Not down, not up.

What I can't work out is what separates the two. The two that worked aren't the two with the strongest profiles, they aren't the most expensive, and they aren't the ones with the closest topical match on paper. If anything the best domain I bought is one of the four that did nothing.

So I was wrong that it never works, and I'm not much closer to knowing when it works. Anyone got a variable I haven't thought of?

Appreciate you actually going and spending the money to argue with yourself. Not many people do that.

The variable I'd look at first: how long had each domain been dropped before you picked it up? Not registration age, dead time. My own hit rate correlates with that more than with anything the tools measure - the ones that worked for me were caught inside a few months of dropping, and the ones that sat parked for two years have never done anything.

My theory, and it is only that, is that a redirect works when the target still has some live signal attached to it, and that decays on a schedule that has nothing to do with the age of the links. Two years parked and it's been recrawled, found empty, and quietly written down, and no redirect brings that back.

Second variable: were the four that failed ones where you redirected to a deep page? Depends what you're running, but I've had better luck to a homepage than people say I should, and I've never understood why.

Both to the homepage worked. Four to deep pages, two of those were the smaller effects and two were nothing.

That's 2 for 2 versus 0 for 4 which on n=6 means nothing, but it's the cleanest split in the data and I hadn't looked at it that way.

That split is confounded and worth flagging before anyone runs with it.

A homepage typically has far more internal links pointing at it than a deep category page, gets crawled more often, and is already the strongest page on the site. If a redirect delivers a small amount of something, the page most likely to convert it into a visible ranking change is the one that was already closest to a threshold. You may be measuring which target could show an effect rather than which redirect delivered one.

FWIW the test that separates them is redirecting into a deep page that's already ranking around position 12 to 15, which is where movement is most visible. Redirecting into a page at 40 and seeing nothing tells you very little.

How long was each one dropped for? byte_broker asked and you answered the other question.

Fair. Pulled it up.

The two that worked: dropped about 5 months and about 3 months. The four that didn't: 14 months, 2 years plus, 8 months, and one I can't establish because the records are patchy but it's over a year.

So byte_broker's variable is a cleaner split than mine. Both winners are the two freshest drops and every failure had been dead a year or more.

so buy fresh drops. that's the thread

The dead-time split is the most interesting thing posted on this board in a while and I think it has a mechanism behind it, which is more than the tactic usually gets.

Consider what has to be true for a redirect to carry anything. Some system has to hold an association between the old URL and whatever value was assigned to it, encounter the redirect, and choose to transfer rather than discard. Every one of those steps is more likely when the record is fresh. A domain dead for two years has almost certainly been recrawled at least once, found to be a parking page or a timeout, and had its pages dropped from the index. There is no longer a record for the redirect to be attached to - the redirect is being evaluated as a new URL that happens to return a 301.

That would predict exactly what you found: a decay curve on dead time, more or less indifferent to the historical strength of the domain, because strength is irrelevant once the record is gone.

It would also predict something testable. If you take one of your four failures and put a live, indexed site on the domain for three or four months before redirecting it, you are re-establishing the record before you use it. That is the version of this tactic that people were doing in 2018, and it may have stopped being described because it takes six months rather than because it stopped working.

I have not run this. It is a prediction, and if anyone has one of these sitting around it is a cheap test.

I've got one from a year ago doing nothing. Happy to run that if someone tells me what to measure.

Two months on and I'll put down what I've got, since d_marchetti is running the interesting version and I'm not.

The page that went 22 to 11 is now at 9. Held through everything since. The section-wide lift has faded back to roughly where it started, which I'm now inclined to call noise after all.

So the honest final score on £400 and six domains is: one clear win, one probably-nothing, four nothings. That is not a tactic, it's a lottery with a decent expected value if you buy cheap enough, and the only edge I've found is byte_broker's dead-time rule, which I would not have found without spending the money.

I'll amend what I said in February. It isn't dead. It's a 2015 tactic that works about a sixth of the time and costs a sixth of what it used to, which is roughly break-even and nothing like the thing people are selling.