Skip to main content

In 2020, an SEO project had a shape most people could picture. You ran an audit, researched keywords, fixed the technical issues, wrote the pages, built some links and reviewed the rankings. Plenty of businesses still budget for SEO that way.

That budget now covers roughly half of the job. The other half decides whether your brand appears in AI Overviews, gets named in a ChatGPT answer or gets passed over for a competitor that explained itself more clearly to the machines. If your search visibility feeds your pipeline, a plan built on 2020 assumptions puts real revenue on the line.

The good news is that the new work is learnable, measurable and manageable. It just has more moving parts, and those parts need looking after.

Keywords gave way to topics and entities

A keyword list used to be the foundation of a content plan. You picked a phrase, matched a page to it and aimed for a position on a results page.

Search engines and language models no longer read your site that way. They map the people, places, services, products and concepts you talk about, then judge how well you cover them and how consistently you describe them. A page about website migration is assessed alongside your pages on redirects, rebrands, domain changes and technical audits. The whole cluster tells the model whether you know the subject.

This changes how we plan. We start with topics your customers care about, define the entities involved, and decide which page owns which part of the conversation. Keywords still matter as language signals, but they now serve the topic rather than lead it.

Every page is written twice

A page used to be finished when the copy was approved and published. Today that is the midpoint.

The first version is for people: clear, useful, written in the voice of your business. The second version is for machines, and it lives in structured data. Schema markup tells search engines and language models what the page is about, who wrote it, which organisation stands behind it, and which entities it mentions. Without it, a model has to guess. With it, a model can quote you with confidence.

Have you ever updated a service page and forgotten the markup underneath? The visible text now says one thing and the schema says another. That mismatch weakens trust signals, and it happens quietly.

llms.txt now sits beside sitemap.xml

Your sitemap.xml has long told crawlers which pages exist. An llms.txt file does a similar job for language models. It gives them a clean, readable guide to your most useful content, your services, and the context they need to describe you accurately.

Treat it as part of your standard technical setup. We add it at launch, review it whenever the site structure changes, and keep it aligned with the entity map so the story stays consistent from one file to the next.

Schema and entity maps drift over time

Here is the part that catches teams out. Schema is not a one time deployment.

When a page changes, the schema often has to change with it. A new service means new entries for what the page is about. A new case study means new mentions, new relationships and sometimes a new author or publisher reference. A rebrand touches almost everything.

The entity map behaves the same way. It records the people, services, locations, products and concepts your brand should be associated with, and how they connect. Your business grows, your offer shifts, and the map needs revisiting. An out of date map sends mixed signals, and mixed signals are how competitors earn the citation you should have had.

Measuring more than clicks

For years the reporting question was simple: how many clicks did we earn? That number still matters, but it no longer tells the full story.

AI Overviews answer many questions directly on the results page. A buyer may see your brand cited, read the summary and remember your name without clicking at all. Another buyer may ask a language model for a shortlist and never reach a search results page. Traditional reporting records both of those moments as nothing.

So we track a wider set of signals:

  1. Rankings, organic traffic and conversions, as before.
  2. Whether your pages appear as sources in AI Overviews.
  3. How language models describe your brand when asked about your category.
  4. Which competitors are cited where you are absent.

This gives you a clear picture of visibility, not just visits. Our AI brand knowledge audits are built to answer exactly these questions.

Audits and sprints need a maintenance plan

The one off audit and the one off SEO sprint made sense when the landscape moved slowly. You fixed the issues, collected the gains and returned in a year or two.

Today the ground shifts under you. Search features change, models update, competitors publish, and your own site keeps evolving. A report that was accurate in March can be out of step by September.

Think about how you care for your website. You pay for hosting, updates, security checks and fixes because a site left alone degrades. SEO now deserves the same arrangement. A regular cycle of review, correction and refinement keeps your schema accurate, your entity map current, your llms.txt file useful and your content fresh. The cost of steady care is predictable. The cost of recovering after months of drift is not.

Traditional link building still earns authority. Links from respected, relevant sites tell search engines that others vouch for you, and that has not changed.

What has changed is the second job. Language models lean on sources they consider reliable when they assemble answers, so you also want presence in the places those models cite. That means industry publications, trusted directories, expert roundups, review platforms and well regarded reference sites in your sector.

The two goals overlap but they are not identical. A link that passes authority may never be cited, and a citation source may carry little traditional weight. We plan for both, and we pair them with strong internal link building so the authority you earn flows to the pages that need it.

AI speeds up tasks and lengthens the process

Many business owners assume AI tools have made SEO faster and cheaper. Individual tasks do move quicker. Drafting an outline, clustering topics or generating a first pass of markup takes minutes.

The process as a whole takes longer. Each page now needs topic research, entity decisions, human editing, schema, validation, internal links, citation planning and measurement across more than one surface. Every step adds a check, and every check protects the quality that models and search engines reward. Automation saves time on the parts, and the extra parts consume it again.

Which brings us to the practical question of what to do about it.

Where to start in 2026

You do not need to rebuild everything at once. A sensible order looks like this:

  1. Audit the foundations. Review your technical health, your existing schema and how AI platforms currently describe you.
  2. Build the entity map. Define the topics, services and relationships your brand should own, then date it so you know when to review it.
  3. Align content to topics. Consolidate thin or overlapping pages, and write new ones with a clear role in the wider structure. Our content creation for SEO service follows this approach.
  4. Add the machine layer. Deploy schema and an llms.txt file, then link both to your sitemap and entity map.
  5. Plan links and citations together. Choose targets that serve authority, citations or both.
  6. Set a maintenance rhythm. Review monthly or quarterly, and update schema, entity maps and reporting whenever content changes.

If a site migration or rebrand is on your roadmap, the stakes rise again. Entity signals, schema and URLs all move at once, which is why we treat website migrations and rebrand and domain moves as strategic transformations rather than technical tasks. Local businesses face their own version of these changes, and our local SEO services cover them.

Over to you

SEO in 2026 asks more of every business, and it rewards the ones that adapt early and keep their systems current.

So, what is the biggest challenge you face with your business right now? Is it visibility, a site that has outgrown its structure, or simply knowing where the new work should begin? Tell us, and we will show you where the biggest gains and the biggest risks sit.

Still have questions? Get in touch for a free strategy session.