The archive keeps making the promise you stopped making

Repositioning is a decision you can take in a morning and a website you can change in a week. The archive takes a year. Two hundred posts arguing for the product you retired do not become wrong on the day you stop selling it; they stay persuasive, and that is the harder problem.

The old pages are not wrong. They are still convincing.

A page that sold the old product well goes on selling it well. It ranks on the queries it always ranked on, speaks to the reader it was written for, and closes with the argument you have just abandoned. Nothing about it is broken. Broken pages announce themselves. Persuasive ones do not.

What that produces is a gap between what the archive promises and what your sales team can deliver. A prospect reads three posts, arrives at the call with a clear and confident picture of something you no longer offer, and the first ten minutes go to correction. That cost appears in no dashboard. The pages look healthy and the pipeline quietly gets more expensive.

It is worse when the old positioning was the better-argued one, which it usually is, because it had years. The retired product has a mature archive and the new one has a landing page. For a while, the most compelling thing your company has ever published is an argument for something you no longer sell.

Sort by the promise, not by the traffic

The instinct is to open analytics, sort by sessions and start at the top. That sorts by how much a page is read, not by how much it commits you. A post with forty visits a month that names a retired product in its second sentence does more damage than a busy one whose argument came through the change untouched.

Sort by promise instead, and the archive falls into three groups. There are pages whose argument still holds under the new positioning. There are pages that make a promise you can no longer keep. And there are pages that were never about the product at all — the ones arguing for a way of working rather than a SKU.

That third group is almost always larger than anyone expects, and it is why the urge to clear the decks is wrong. A good archive is mostly worldview. Repositioning rarely changes the worldview; it changes what you charge for.

Most of the archive needs an edit, not a verdict

Between "still true" and "retire" sits the bulk of the work: pages whose argument survives but whose furniture does not. The product noun in the third paragraph. The example drawn from a customer segment you left. The closing line pointing at a page that no longer exists.

This is an editing pass, not a content project, and it is underestimated because it is boring rather than difficult. It also goes undone for the same reason internal linking goes undone — it requires holding the whole archive in mind at once, the one thing a busy marketing team cannot sustain and a machine does not mind.

Retirement is the smallest group and the most expensive to execute, which is why deleting a post is a last resort rather than a tidy-up. Google's site move guidance is explicit that redirecting many old URLs to one irrelevant destination, such as the home page of the new site, can confuse users and might be treated as a soft 404. Retiring twelve posts properly means twelve decisions about where each reader should land.

An engine can do the pass. It cannot draw the line.

The mechanical half of this is exactly what automation is good at and people are bad at sustaining. Which pages name the retired product, which links point at pages you are about to remove, which examples assume a customer you no longer serve: that is an inventory, and inventory across two hundred pages is machine work.

The line itself is not mechanical. Deciding that a claim is one you are still willing to defend is a commercial judgment with competitive and sometimes legal edges, and it usually lives in a founder's head for months before anyone writes it down.

The failure mode here is specific and worth naming, because it is ours. An engine re-arguing an archive from a stale brief produces pages that are fluent, confident and re-sell the old product in the new vocabulary. Those pages look migrated. They are worse than the untouched ones, because an untouched page is visibly dated and a confidently wrong one is not. Google's bar is whether a page is helpful and written for people, and as we have argued before, it does not care who typed it — a polished restatement of a promise you have withdrawn clears no bar at all.

So we ask for the new positioning in writing before a single page is touched: what you sell now, to whom, and what you are no longer willing to claim. The third is the one nobody has written down, and the only one the pass actually needs.

The archive turns slowest, so turn it first

If you accept this, the sequencing changes. The archive review starts with the repositioning rather than after the new site is live, because it is the slowest asset you own and the only one that keeps arguing while you wait.

Budget an editing pass over everything and a handful of genuine retirements, not a rewrite. Judge each page by the promise it makes, not the traffic it earns. And write down what you will no longer claim: that sentence turns two hundred ambiguous pages into three clear piles.