What the SEO playbook should look like in 2026

11th June 2026

When I returned from maternity leave in 2024, after having my second child, I was confronted by two headlines: “SEO is dead” and “we’re in the SaaSpocolypse”. For an SEO consultant who specialises in SaaS (and hadn’t slept in six months) this was pretty horrifying.

So I think it is important to open up the question: what has actually happened to SEO in 2026? What SEO playbook can we safely use to get results from organic, and what does that look like for today’s modern business?

In short: don’t think SEO is dead. But I do think what we knew as SEO — being found on Google — has expanded well beyond Google and that’s the key thing new SEO strategies need to account for.

What’s changed in SEO: Search has moved beyond Google

When we say SEO, we mean “search engine optimisation”, but let’s be honest, we only really think of being optimised to be found in Google search. SEO should have encompassed being found in Google, Bing and Duck Duck Go at minimum, but most of us ignored that because Google gave us enough traffic.

What’s happened now is that as well as searching on Google, users are going to search elsewhere and I don’t mean doing the odd search on Bing. Search behaviour has fundamentally changed, and that matters.

Whereas people used to go and start their journey in Google, they are now starting in LLMs like ChatGPT or Copilot. In some cases, they are starting on platforms like TikTok. Search surfaces have expanded well beyond Google.

The four search surfaces every SEO strategy now needs to cover

The new SEO landscape should really cover four areas:

Your SEO strategy may not target all of those equally, but it needs to consider all of them because they are all search surfaces people are using to find information and make purchasing decisions.

1. SERPs: Google blue links are still relevant

SERPs, blue links and traditional Google search are still relevant. Most traffic is still coming from here.

The good news is that a lot of what you used to do to be ranked in SERPs will also help you be ranked in LLMs.

If you have had an SEO strategy and you have been working on SEO for a few years, you should already be seeing some traffic from LLMs.

When people come to me in a panic that they are not being found in AI tools, it is often because they did not have a good SEO strategy to begin with.

If you are starting from scratch, you probably still need to do some of the basic things that would have helped you be found in Google before you even start thinking about AI surfaces.

2. AI Overviews: somewhat good for brand awareness, but low click-through

AI Overviews are like an evolved version of featured snippets.

The returned information is often quite long. It might be a summary, a table, or a bulleted list. Google is trying to give the user a comprehensive answer to what they are asking.

When AI Overviews appear, they usually take up the entire section above the fold. That means users are not necessarily going to click beyond them because the AI Overview is there to give them the answer.

Only a very small percentage of users click on the links cited in AI Overviews. So it is quite difficult to gain meaningful traffic from them.But there can be brand awareness. If your product or brand is mentioned in an AI Overview, that can still have value. The challenge is that it is hard to prove attribution.

AI Overviews also differ by device and industry.

Mobile searches are more likely to trigger an AI Overview than desktop searches. So it is really important to know your customers and know where and how they are searching.

If the majority of your customers come through mobile, then AI Overviews may need to be a bigger part of your strategy. If your customers search mainly on desktop, you might slightly deprioritise them.

It also differs by industry. In sectors like healthcare and education, AI Overviews appear much more frequently. In ecommerce and finance, they appear less often.

3. LLMs: ChatGPT, Copilot, Perplexity and Claude

The main thing to know about LLMs is that the way they collect and decide which information to surface is very different from Google.

Google Search is primarily based on a keyword-matching exercise. Somebody searches for a keyword, Google finds a website where that keyword is in the title, the heading one and the body text, and it matches the two together.

LLMs work on a completely different model.

The second thing that is really important to note is where LLMs get their information from via the internet. They get it from Bing’s index or Google’s own search index.

So if your website cannot currently be crawled, indexed or discovered by Google or Bing, it is not going to be discovered by LLMs.

This is the thing people are missing. They are trying to skip over the Google and Bing step and go directly to the LLMs. But LLMs use Google and Bing’s search indexes anyway.

Why LLMs need topic authority, not just keywords

When somebody searches in ChatGPT, the tool does not just answer that one question.

If I ask, “What are the best flowers to plant in spring?”, ChatGPT may go out and look for sub-queries related to that prompt. It might look at flower types, seasonal plants, gardening tips, soil prep, sunlight, watering and climate.

And remember, LLMs often have an element of memory. ChatGPT may know I am based in the UK, so it will think about climate information for where I am based.

It then puts all of that information together to return an answer.

That is why, when we are thinking about creating an AEO strategy or an SEO strategy for LLMs, we need to think in terms of topics. We need to think about owning topics and owning related topics, because of this fan-out query method that LLMs use.

LLMs will also often rewrite your prompt. Even if I put a specific prompt in, ChatGPT might rewrite that prompt to something it thinks is better or more likely to surface good information.

So when you are thinking about your AEO strategy, do not focus too much on owning one specific prompt. Prompts are very fluid. You need to think about owning topics.

Which LLM should you optimise for?

There are so many LLMs out there that I would start by understanding which one your customer is most likely to use.

If you are selling business-to-business and your customers are mostly Microsoft organisations, they are highly likely to use Copilot.

If your customer is an everyday consumer, or a certain demographic, they may be more likely to use ChatGPT.

This matters because each LLM has a slightly different way of surfacing information. They use slightly different sources. Some have partnerships.

For example, ChatGPT was heavily trained on Reddit. You will see a lot of Reddit come up, so you might need to optimise for your company or product to be found on Reddit if you are targeting ChatGPT.

It is almost impossible to target every single LLM within your strategy. Understanding your customer will help you prioritise which one to focus on.

Which sources are cited most in LLMs?

The question I get asked at least once a week is: which sources are most cited in LLMs?

If you look at broad studies, you will see domains like Reddit, Wikipedia, Medium, Forbes, LinkedIn and YouTube.

That is valid for broad queries. But we have to remember that this is based on broad, everyday users. It is not necessarily based on your specific industry.

If you look at your industry, it might look completely different.

It can still help generally as a PR strategy to be surfaced on Reddit, Forbes or LinkedIn. But you need to prioritise based on the sources that matter in your industry.

4. AI Mode: Google’s more LLM-like search experience

AI Mode is what you can use when searching in Google, for example in Chrome. You can click on AI Mode and it works more like an LLM response than a traditional Google result.

You get a comprehensive answer, and the nice thing about AI Mode is that it links sources in a clearer and more compelling way than AI Overviews.

If you were one of the websites cited, you would probably be more likely to get click-throughs because the sources are shown much more clearly.

One thing to remember at the moment is that Apple does not have an AI Mode on Safari. So if your customers are heavily using iPhones and searching on mobile, there is currently no AI Mode there.

If Apple did decide to add an AI Mode, that could really change the amount of traffic you are getting or how well your products are performing.

So if that is your customers’ device of choice, it is something to keep an eye on and consider in your future strategy.

The future: AI agents and agentic search

A fifth area to be aware of is where we are heading: AI agents, or agentic AI.

If you have not heard this term before, the easiest way to describe it is this. Imagine you had a personal assistant and you said to them, “I want to go and have sushi on Thursday at 2pm.”

Your personal assistant would know where you work. They would know your location. They would look for restaurants, select the best one, make a booking for 2pm and put it in your calendar.

That is basically what we expect AI agents to be able to do on our behalf.

They will be linked to our systems, they will have information about us, and they will be able to take an instruction and execute something without us having much input.

For some people, that’s scary.

One of the industries this will really affect is ecommerce. If you work in ecommerce, or with ecommerce brands, I would start investigating what this means.

Google has already released its Universal Commerce Protocol, or UCP, which is basically a list of instructions for what you should do within your website and what attributes you should give your products to enable them to be found in agentic search journeys.

It feels like AI agents may not be broadly available within Google until 2027, but that is not very far away.

Eventually, there might be shopping within AI Mode where your AI agent makes a decision and purchase without ever leaving AI Mode or going to your website.

There are things ecommerce brands can put in place now to help products surface. There are caveats — for example, products with age restrictions or warnings may not be eligible — but generally, most products will be eligible.

This is definitely something to look into, because it is where AI is heading. It will be another surface we have to optimise for within search strategies.

A 12-step SEO and AEO strategy for today

So, with all of that said, this is the strategy I would use to come up with an SEO strategy today.

1. Make sure your website is technically robust

This is boring, and it is the step most people skip over. But you must make sure your website is technically robust.

If your website cannot be crawled, discovered and indexed by Google’s robots, it cannot be found by ChatGPT.

The mistake I see most companies make is that it is not really obvious, in obvious places, what they do and who they do it for.

Please go and look. It sounds really stupid, but I see this all the time.

Does your homepage say what you do? Do your headings say what you do? Are you talking about the category you are in? Are you using clear heading structures?

A quick-loading site has always been important for SEO, but it is even more important now because ChatGPT and other tools may just ignore your site if it is not fast-loading.

You also want your most important information to be in fast, static HTML and CSS pages, not hidden or wrapped within JavaScript, because LLMs are not great at reading that at the moment.

2. Create goals in the context of AI search

Start by creating some goals for your business in the context of AI.

Think through what each search surface actually means.

AI Overviews are good if they mention your product by name or mention your brand. But beyond that, they are quite informational and it’s hard to get click-throughs.

Prioritise which search surfaces are going to be most important for you based on what you know about your users.

If you can, do customer research. Ask your customers whether they’re starting their journeys in LLMs. Ask what they are doing. Get anecdotes, even if you cannot get perfect data.

3. Research your users, buyers and their buying journey

Start to gather more research around your users and buyers.

Understand what device they use. Understand what questions they’re asking at each stage of the buying journey. Understand what language they use to describe your product or company.

Sometimes this is not the same language that you would use.

That means you may need content on your website that is about solving a pain point, rather than just saying, “We are the best product for X.”

Sometimes people do not even know what they are searching for. This has become even more important.

4. Do some initial desk research

Start doing some simple desk research.

Take a prompt that you think your potential customer would ask and put it into Google. Does an AI Overview show?

Go into AI Mode. What happens?

Go into ChatGPT. What happens?

Start to get a feel for what is happening for your keywords and topics.

5. Determine your areas of topical authority

The next thing is to determine your areas of topic authority.

This is not that dissimilar from what you would have done for an SEO strategy in the past. But with LLMs, it’s even more important to cover your topics broadly.

Look at the category you fall into, the features, products or services you offer, the problems you solve, and the competitors you replace.

A lot of buying-journey prompts are compare-and-contrast prompts. “Is this better than X?” “Should I buy this instead of this one?”

You need to understand the topics you need to own, not just the keywords you want to rank for.

6. Understand your current visibility in LLMs

You should try to get an idea of your current ranking in LLMs.

There is no reliable tool right now that can show you every single prompt your company, brand or product is showing up for.

If ChatGPT tomorrow created its own Search Console, like Google Search Console, then yes, you could see every single prompt you rank for. But there are no signs they are going to do that at the moment.

There are tools out there, and some are very expensive, but just be aware there is no definitive tool.

Most of these tools do not have access to the APIs for these LLMs. What they are doing is scraping. They are running prompts and scraping data from them. It is never going to be one hundred percent reliable.

What I like to do is create a representative sample of what we think is happening.

Existing SEO tools like SEMrush and Ahrefs now have AI visibility tools. I tend to use SEMrush’s tool. There are also newer AI-specific tools. You can also do it manually by doing desk research.

And in GA4, if you go into New Users, then User Acquisition, and break it down by Source and Medium, you can start to see some clicks from ChatGPT and other LLMs.

But remember: some of your direct users may also be coming from LLMs. You just cannot see it.

7. Find your topic gaps, not just keyword gaps

In traditional SEO, we talked about finding keyword gaps. Which keywords are we not ranking for?

When you are thinking about LLMs and AI surfaces, you need to think about topic gaps.

When you look at prompts and the information coming up, what else does ChatGPT always talk about? That is where you will find topic gaps you also need to cover.

It can be a good idea to create a representative prompt library and do some analysis.

Which sources are being cited? Which related topics are coming up?

This will help you find the gaps within your own website.

But remember: prompts are not keywords.

Keyword position is fairly fixed. If you get onto page one of Google, unless you have a really bad penalty, you might move down a place or two, but you’re not likely to be completely kicked off.

Prompts are fluid. You could appear in a prompt one day and not appear the next. Even a slight change in wording, tone or what the user has said before in ChatGPT will change the answer.

So don’t fixate on appearing for one specific prompt. Focus more broadly on owning topics with real relevance and deep information.

8. Write AEO-optimised content

Once you’ve done all that, the next step is to write AEO-optimised content.

Focus on your top-of-page signals first. Make sure your homepage, category pages, product pages and use case pages are really clear about what you do.

Then think about how you structure your core pages and blog posts.

LLMs really love structure.

This is the main shift from how people used to write blog posts: two thousand words broken up by a few headings. LLMs need more structure to understand and take pieces from your information.

Tables are really good. Subheadings are good. Bullet points are good. As much structure as you can give your blog posts and pages will help LLMs pick out the information they need.

Specific detail also matters.

It is not good enough to write generic content. It was never good enough for Google, but it’s definitely not good enough now.

You cannot have a content writer who does not really know your customers, your product, your services or what you offer writing generic content.

It needs to be incredibly specific.

If you are selling a dress, is it lightweight? Is it soft? How does it wash? What do people say about it?

Those are the details that will get you found in LLMs.

9. Create the types of content LLMs like to cite

There are some forms of content that perform really well for LLMs.

Buyer’s guides are really good. For example, if you are a software company, you might write “The best software for managing restaurant accounts” and compare six or seven pieces of software, positioning yours as the best option.

I see that type of content coming up a lot in AI results.

Comparison blog posts also perform well: X versus Y, alternative guides, replacement guides, and “best for” content.

The About Us page is also more important than people think. Most of us write it and then completely discard it, but within LLMs and AI Overviews, the About Us page can provide really important context.

Go back to your About Us page and make it richer and more detailed.

Pricing information also matters, particularly for service-based businesses and product-based businesses outside ecommerce.

If you do not have pricing information on your website, you can be missed out of some sources. If someone is asking which option is most expensive or what the price points are, and you do not have pricing on your website, that may be a signal not to include you.

FAQs are another obvious one. Having FAQs as part of a page, or even a standalone FAQ page, can help your information be easy to understand and use.

Within your articles, useful elements include:

Anything that gives you more authority and context can help you be surfaced more.

10. Make sure new content is indexed

One thing I have definitely noticed is that, in the past, when you published a new blog post or page, Google would crawl it pretty quickly and pick it up.

Now I am seeing it can take a month, sometimes two months, for Google to pick up a new page.

I think that’s because the volume of content being produced and published is much higher now. Google only has a certain crawl budget. There are only so many pages it can crawl.

If you are publishing thousands of pieces of new content a week, this will not work. But if you are publishing one new page or one new blog post a week, I would go to Google Search Console, use URL Inspection and force it to index that page or blog post so it gets picked up quickly.

11. Target the sources being cited in your industry

Beyond your content strategy, it’s really important to find and target the sources being cited within your industry.

This will change depending on your industry.

You’ll need some kind of strategy around it — a PR strategy, seeding strategy or backlink strategy.

This isn’t that dissimilar to SEO. We used to talk about links and content. Now we still need content, but we don’t necessarily only need links. We need mentions in the sources being cited.

I’m seeing a lot of companies invest in PR again because it helps them get brand mentions.

There are also things you can do yourself depending on which sources you want to be targeted in.

For example, if you’re a hospitality brand and you know TripAdvisor is being sourced, you might do a big push to get reviews on TripAdvisor from your customers.

You need to work out which sources are most important and how you are going to obtain mentions there.

12. Track your results

Finally, you need to track results.

This is a new playbook we are writing, so you need tracking that is more robust than your boss going into ChatGPT every day to see whether you are showing up. Whether you’re using one of the AI visibility tools, an SEO tool, or GA4 to track what’s happening with your traffic, you need some measure of whether what you are doing is working.

Over to you

If you need help turning your current SEO strategy into one that works for 2026, get in touch today.