Five acronyms, and the four things any of them can actually change

Four proposals on one desk

Four proposals were open on a marketing lead's desk last quarter and I was looking at them with her. One was an SEO retainer up for renewal. One was an agency pitching generative engine optimisation. One was an internal request for a programmatic content programme. The fourth had arrived because her board had heard about answer engines and wanted to know what the company was doing about them.

Four acronyms, four numbers, and a reasonable person trying to work out whether she was being sold four things or one thing four times.

The honest answer is closer to the second, and it took me a while to be able to say it briefly. Most of what these five acronyms ask you to do is the same work, and the parts that genuinely differ are smaller than the vocabulary suggests.

What each one actually means

Plainly, without the marketing around them.

SEO is making a page findable and credible to a search engine that returns a list of links. It is the oldest and the best understood, and most of what it asks for is unglamorous: the page loads, it is crawlable, it is about something specific, other pages link to it, and it is not a duplicate of four other pages on the same site.

AEO is making a page usable by something that returns an answer instead of a list. A search summary, a chat assistant, a voice result. The shift is that being in the top ten is worth nothing if the thing reading you is producing one answer, so the question becomes whether your specific claim can be found, bounded and attributed.

GEO is the same idea aimed at generative systems specifically, and in practice the advice overlaps with AEO almost entirely. Where it differs is emphasis: more attention on whether your claims survive being summarised, and on whether the model can tell your product apart from a similarly named one.

pSEO is producing many pages from a data source rather than writing them individually. Integration pages, location pages, comparison pages. It is a production method, not a way of being read, which is why it belongs in a different category from the other four.

AIO is the broadest and the least settled. Most people using it mean making sure an AI system that summarises your company gets it right, particularly about availability, pricing and what you actually do.

The four things any of them can change

Underneath the vocabulary, everything these acronyms can ask of a page reduces to four properties. This is the framing I now use in that conversation, because it turns four proposals into a decision about which of four things to work on.

Structure. Whether a claim on the page has a boundary around it: a heading that answers a question, a table row, a short paragraph under a plain heading. Not whether the writing is good.

Answerability. Whether the page addresses a question somebody actually has, in the form they have it, rather than a topic somebody in marketing decided to cover.

Evidence. Whether the claims can be sourced. A number with an origin, a limitation stated plainly, a specific rather than an adjective. This is the one that decides whether an automated reader treats your page as reliable, and it is also the one that decides whether a human buyer does.

Maintenance. Whether the page is still true in a year, and whether it agrees with the rest of your site. This is the one nobody sells and it is the one that determines whether the other three keep working.

Every proposal you receive can be sorted into those four. Doing that sorting in the meeting is the most useful thing I know for these conversations.

One clarification on maintenance, because it is the property people assume is somebody else's job. It is not the same as keeping the site online or updating a plugin. It is whether the specific claim you structured last year still matches what the company sells, and whether the four other pages mentioning it were updated at the same time. Nobody sells that, everybody needs it, and it is the reason the other three decay.

Where they overlap, and where they genuinely differ

This is the table I draw. The last column is the one worth reading first, because it is where the money goes wrong.

The pattern is that four of the five are asking for structure, evidence and maintenance in different proportions, and one of them is asking you to produce more pages, which is a different kind of decision entirely.

pSEO is the odd one out and deserves separating

The other four are about how a page is read. pSEO is about how a page is made, which makes it a different sort of decision with a different sort of risk.

It works when the underlying data is real and each page answers a question somebody genuinely has. An integration page for a tool your buyer actually runs, with specific information about how that integration behaves, is a good page whether it was written by hand or produced from a record.

It fails when the page count is the goal. Forty near identical pages with a name swapped are forty things that have to be kept correct, produced by a team that did not grow, and they are also the clearest possible signal to any automated reader that the site is padding.

The question that sorts it is not about search at all. If a person landed on page thirty seven of the set, would they get something specific. If the honest answer is no, the programme is producing liability rather than coverage.

Why the vocabulary keeps multiplying

Worth understanding, because it tells you how to read the next one that arrives.

Each acronym appears when the way people find things shifts enough that the existing advice stops covering it. Search engines returned links, so SEO was about being in the list. Systems started returning answers, so somebody needed a word for being the source of one. Generative systems summarise rather than extract, so somebody needed a word for surviving that.

Those are real shifts and the words are doing legitimate work. What happens next is the problem: a word that describes a shift becomes a service line, then a budget line, then a department, and by the time it is a department the overlap with the previous one is nobody's interest to point out.

So the practical stance is to take the observation seriously and the packaging sceptically. When the sixth acronym arrives, and it will, the question is which of the four properties it is asking you to change. If it is asking for a fifth property, that is genuinely new and worth paying attention to. If it is asking for structure and evidence again, it is a rename with a price attached.

What none of them can do

Worth stating clearly because it is where the disappointment comes from.

None of them is a channel. You cannot buy AEO the way you can buy paid search. These are properties of pages you already own, which means the work is editing your site rather than spending on a platform, and that is a harder thing to get approved because it does not look like a campaign.

None of them can compensate for a site that contradicts itself. If your pricing appears differently on three pages, structure work makes those contradictions easier for a machine to find, which is worse than leaving them buried. That is the specific case where two of these disciplines actively make things worse before they make them better.

And none of them survives without maintenance. A page optimised for any of the five and then left alone for two years is a page that is now confidently wrong in a well structured way.

What to do with four proposals

Practically, in the meeting where you have four numbers in front of you.

Ask each supplier which of the four properties they are proposing to change, and for which specific pages. A proposal that cannot answer at the page level is a retainer for activity.

Ask what happens to the work in eighteen months. Nobody will have a good answer. The one who admits it is the one to take seriously.

Then check your own site for contradiction first, before commissioning any of it. If pricing disagrees with itself, that is a week of work you can do internally and it makes every one of the four proposals more likely to succeed.

And separate the pSEO decision from the other three, because it is a production commitment with an ongoing maintenance cost, and it should be approved on those terms rather than as another optimisation line.

• Which of structure, answerability, evidence or maintenance is this changing?
• Which specific pages, by URL?
• What happens to this work in eighteen months, and who owns it?
• Have we fixed our own contradictions first?
• Is this a production commitment rather than an optimisation?

Where teams get this wrong

Buying the acronym rather than the work. Four suppliers proposing overlapping structural work at four prices is a common situation and it is rarely presented that way, because overlap is not a good sales story.

Running them as separate programmes with separate owners. The same page ends up being edited by two teams to different specifications, and neither result is coherent.

Measuring on rankings when the mechanism has changed. If the thing reading you produces one answer, position in a list of ten is measuring something that is no longer how you are found. That does not mean rankings are worthless, it means they are one signal among several and they are the easiest one to report.

Doing structure work on a site that disagrees with itself. Boundaries around claims make contradictions easier to find, not less consequential.

Adding an FAQ block as the AEO deliverable. An FAQ that restates the page above it adds a second place for a fact to live, which is the beginning of drift.

What good looks like

A company handling this well does not have five programmes. It has one practice with a few properties it cares about, and the acronyms are vocabulary for talking to suppliers rather than departments.

Concretely: each fact is stated authoritatively in one place and referenced elsewhere. Headings on the pages that carry facts answer questions rather than naming topics. Claims have origins. Somebody reviews the pages that matter on a schedule. And the pages produced from data are produced from data that is genuinely different per page.

The observable difference is that when a supplier pitches the next acronym, the conversation is short. Either it is asking for one of the four properties, in which case you already know whether that work is done, or it is not, in which case you can ask what it is actually asking for.

How to measure this honestly

This is where I would push back hardest on most proposals, including ones I might otherwise agree with.

Attributing a change in traffic or ranking to structural work on a specific page is difficult on a marketing site with normal volumes, because too many other things move at the same time. Anybody promising a ranking outcome from this work is promising something they cannot control.

The measurements that are honest are smaller and more direct. Ask a few assistants the question your page is supposed to answer, before and after, and see whether the answer resembles your page. Count whether your sales team still answers the question the page now answers. Check whether the fact you made canonical still agrees with itself in six months.

None of those is a dashboard. All of them are checkable by a person in an afternoon, and they measure the thing you actually wanted rather than a proxy for it.

Where Creogen fits, and what it is not

Creogen is a website operations platform in private development, and I want to be careful about how it sits in an article full of acronyms that get sold as ranking tools.

It is not a ranking tool and it does not optimise anything. What it is aimed at is the fourth property, maintenance, which is the one none of the five acronyms sells and the one that decides whether the other three keep working. Which pages state the same fact, which of them disagree, what changed and whether it helped, and which pages nobody has reviewed in a year.

Creobot supplies the second property, answerability, by collecting the questions visitors ask that no page answers. That is the difference between a page addressing a topic somebody chose and a page addressing a question somebody had.

Neither addresses structure or evidence, which are editorial decisions that a person makes about a specific page. I would be sceptical of any tool claiming otherwise, including one of ours.

The short version

Five acronyms, four properties, and most of the work overlaps. Structure, answerability, evidence, maintenance.

Sort every proposal into those four and the conversation stops being about vocabulary. Separate pSEO out, because producing pages is a different commitment from making pages readable. And fix your own contradictions before commissioning anybody, because that week of internal work makes every subsequent pound more likely to do something.

None of this is a channel and none of it is buyable. It is your site, edited, by somebody who owns it.