The press release didn’t die. It got rebuilt for machines that read

Somewhere around 2013, the press release stopped working as a search tactic. Google had spent years watching the same announcement appear on four hundred low-quality sites with the same keyword-stuffed anchor text pointing at the same commercial page, and it did what it does: it discounted the signal and made clear that links in syndicated releases were expected to be marked as non-following.

The industry read this as the death of the press release, and for the specific thing most people were using it for — buying links at scale — that reading was correct. Wire distribution as an SEO mechanism is over and has been for a long time.

What went largely unexamined is that the format survived the collapse of the tactic. And the format turns out to have properties that matter a great deal more now than they did then.

Why the format suits machine retrieval

Consider what a retrieval system needs from a source before it will use it in an answer. It needs to know when the information was published, because stale facts are worse than no facts. It needs attribution, because an unattributed claim is not citable. It needs concrete, checkable statements rather than adjectives. It needs to identify which entities the document is about — which company, which product, which place. And it benefits enormously from explicit structure rather than inference.

Now consider the press release. A dateline. A named issuing organization. A named spokesperson with a title, attached to a direct quotation. Specific facts: what launched, where, when, at what price, covering what area. An about-the-company block identifying the entity. A contact.

The format was designed in the nineteenth century for wire editors who needed to establish, fast, who was saying what and when. The requirements of a machine assembling an answer turn out to be strikingly similar. Nobody planned this. It is a genuine accident, and it is being almost entirely wasted.

What is actually being wasted

Most announcements published today are optimized for a distribution model that no longer functions, and carry the scar tissue to prove it.

The adjectives. "Leading provider." "Innovative solution." "Industry-leading platform." These are unverifiable and therefore worthless to a retrieval system, which cannot cite them and will not try. They also displace the sentences that would have been useful.

The buried lede. Wire-era releases often open with three lines of positioning before reaching the news. A system that extracts the first substantive passage as a summary gets positioning instead of information.

The missing specifics. "Available in select markets" is not a fact. "Available in Denver, Boulder and Fort Collins from October 1" is. The first cannot be used to answer anything; the second can answer several questions.

The absent structure. A release published as unmarked prose asks every consumer to infer what it is about. A release published with explicit markup states it.

What to do instead

Write the first forty words to stand alone

Assume the opening passage will be extracted and quoted without the rest of the document around it. It should contain who, what, where and when, and it should be true and complete in isolation. This is the single highest-leverage change available, and it costs nothing.

Replace adjectives with facts

Every unverifiable claim should be either removed or converted into something checkable. "Leading regional installer" becomes "installed 1,400 units across Colorado since 2019" — assuming that is true, which is the point. If a claim cannot survive being made specific, it should not be in the document.

Name a human

A quotation attributed to a named person with a stated title is citable in a way that a quotation attributed to "the company" is not. It also constrains the writing usefully: people say things, and organizations emit statements.

Mark it up

Explicit structured markup — article type, publication and modification dates, publisher identity, the entities the document concerns, and links connecting that entity to its other verifiable presences — converts inference into assertion. It is the difference between hoping a system works out what the document is about and telling it.

Give it a permanent home

A release that exists in four hundred syndicated copies and nowhere canonical is a mess for every consumer of it. One canonical URL on a stable domain, with any syndicated version pointing back to it, resolves the ambiguity.

The part that is still dead

None of this resurrects the link-buying model, and it should not be read as an attempt to.

Links inside syndicated announcements are expected to carry attributes marking them as such. Publishing paid third-party content on a site in order to borrow that site's ranking authority is a pattern search engines act against directly. Anybody selling announcement distribution on the promise of ranking improvement through links is selling a 2011 product with a 2026 price.

The honest value proposition is different and narrower: a dated, attributed, factually specific, structurally explicit document, published at a stable canonical address, that a retrieval system can find and use. That is worth something. It is just worth something for different reasons than the reasons the industry spent a decade claiming.

Why this matters more for small companies

Large organizations are already well represented in the sources retrieval systems draw on. They are covered by trade press, they file public documents, they are written about by people who are not them.

Smaller companies frequently have no such coverage. What exists about them is their own website and some directory entries. For that category, a well-constructed announcement is often the only dated, attributed, third-party-published document that exists about the business at all.

Which means the discipline described here is disproportionately valuable to exactly the organizations least likely to apply it — because they are also the ones most likely to have been sold the old model, found it did nothing, and concluded that announcements are a waste of money.

They were right about the model. They were wrong about the format.

The publisher of Action Global News operates marketing businesses that include announcement publishing among their services; see our editorial standards. Companies with news for our desk can use the news tips page, and our announcement publishing terms are on the newsroom page.