SEO · AI SEARCH
SEO Is Not Dead.Search Has Become Bigger Than Google.
A practical guide to building a website that can be found, understood and trusted across Google, AI Overviews, ChatGPT, Perplexity and Gemini.
For years, SEO had a fairly clear objective.
Understand what people search for, create useful pages around those searches, get those pages indexed, improve their rankings and turn organic traffic into leads, sign-ups or customers.
That system still matters.
What has changed is that Google is no longer the only place where people search for answers, products and recommendations.
A potential customer might search Google and get an AI Overview before reaching the traditional results. They might ask ChatGPT to recommend the best software for a specific use case. They might ask Perplexity to compare three companies, or use Gemini to research a purchase before ever visiting a brand's website.
The search journey is becoming distributed across platforms.
Sometimes an AI system sends a visitor to your site. Sometimes it cites your content. Sometimes it mentions your brand without linking to you at all. And sometimes it mentions your company but gets the product, pricing or positioning wrong.
This changes what businesses need to optimise and measure.
It does not mean traditional SEO is dead.
In fact, Google's own guidance continues to reinforce that many of the same fundamentals used for traditional Search also matter for AI-powered search experiences. Your pages still need to be accessible, crawlable, indexable, useful and understandable.
There is no magic AI SEO switch that replaces those fundamentals.
The opportunity is bigger than that.
Modern SEO is increasingly about making your company easy to find, easy to understand, easy to verify and easy to recommend wherever people are researching a problem.
The job of SEO has expanded
Traditional SEO normally asks a few familiar questions.
Can this page rank? Can it attract qualified traffic? Can that traffic generate leads, sales or sign-ups?
Those questions still matter.
But AI-led search creates another layer.
You also need to understand whether search and AI systems can correctly identify what your company does, what your products are, who they are for and which questions your pages can answer.
You need to know whether your company appears when someone asks an AI assistant for recommendations, whether the information being presented is accurate and whether your website is being used as a source.
This creates an important distinction.
A page can rank well on Google and rarely appear in ChatGPT. A company can be mentioned frequently in AI answers without receiving much referral traffic. An AI assistant might recommend your brand but describe an old version of the product or show outdated pricing.
Visibility is no longer just a ranking position.
It is increasingly a combination of discoverability, understanding, citation, accuracy and commercial impact.
That sounds like a new discipline, but much of the underlying work remains familiar.
Technical accessibility matters.
Clear content matters.
Authority matters.
Internal links matter.
External mentions matter.
Useful, original information matters.
The channels have expanded. The foundations have not disappeared.
1. Make sure search systems can actually access your website
Before thinking about AI citations, content strategy or backlinks, start with a much simpler question:
Can search systems properly read the website?
A page can look perfectly normal in Chrome while exposing surprisingly little information to a crawler.
This is particularly common on JavaScript-heavy websites.
A person visits the page, their browser downloads JavaScript, the application runs, and the content appears. But the first HTML response may contain almost nothing meaningful.
For example:
<div id="root"></div>
<script src="/app.js"></script>
The page heading, product details, price, FAQs and links may only become available after the JavaScript executes.
Google can render JavaScript, but crawling, rendering and indexing are not exactly the same process. Other search and AI crawlers may behave differently.
Where possible, important commercial and informational content should be available in the initial HTML through server-side rendering, static generation or another reliable rendering approach.
Check what your server actually returns
You do not need an expensive SEO tool to perform the first check.
Open a terminal and run:
curl -L -s https://example.com
Then check whether the HTML contains the page's important information.
For example:
curl -L -s https://example.com/pricing | grep -i "£29"
Or check the main heading:
curl -L -s https://example.com | grep -i "<h1"
You are trying to establish whether important information such as the main heading, product description, pricing, internal links, canonical tag, structured data and contact details is actually visible in the response.
Do not only test your homepage.
Your product, service, pricing, comparison and documentation pages are often far more important commercially.
A pricing page that exposes the names of three plans but hides the actual prices until JavaScript runs may technically exist, yet still make it harder for external systems to answer the most important buyer question:
How much does it cost?
Check HTTP status codes
Every important URL should also return the correct HTTP response.
You can check it with:
curl -I https://example.com/page
A working page should normally return a successful response such as:
HTTP/2 200
A permanently removed page should return an appropriate 404 or 410.
A page that has permanently moved should usually redirect using a 301.
One common mistake is showing an error page while still returning 200. This creates what is commonly called a soft 404.
Another is redirecting every deleted URL to the homepage.
Clean status codes make the website easier for crawlers to interpret and maintain over time.
Review robots.txt carefully
Your robots.txt file tells crawlers which areas of the site they can access.
You can usually find it at:
https://example.com/robots.txt
A rule such as:
User-agent: *
Disallow: /
blocks the entire site.
More commonly, businesses accidentally block important sections such as:
Disallow: /products/
Disallow: /blog/
Disallow: /docs/
AI platforms can also use different crawlers for different purposes.
For example, the crawler involved in surfacing content inside an AI search product may not necessarily be the same crawler associated with model training.
That distinction matters because businesses may want their public information discoverable in search without making the same decision for every other use.
The right approach is not to blindly copy an "AI robots.txt template."
Decide which public systems you want to access your content, review their current crawler documentation and periodically check your server logs to understand what is actually visiting the website.
2. Keep your website crawlable and indexable
Crawling, indexing and ranking are different stages.
A crawler can discover a page without indexing it.
A page can be indexed without ranking for anything meaningful.
And a page can rank without generating useful traffic or conversions.
This distinction becomes important when diagnosing performance.
Businesses often respond to declining traffic by publishing more content even when the actual problem is that search engines are struggling to process the pages already on the website.
Before creating another 100 articles, make sure your existing URL inventory makes sense.
Control unnecessary URLs
A website should not create hundreds of indexable URLs when only one useful page exists.
Ecommerce sites are particularly vulnerable to this.
A single category might generate URLs such as:
/products/shoes
/products/shoes?colour=black
/products/shoes?sort=price
/products/shoes?utm_source=email
/products/shoes/
Some of those variations may be useful to users, but that does not automatically mean each variation deserves to exist as a separate search result.
Similar problems can come from filters, sorting parameters, search pages, tags, tracking parameters, session IDs, duplicate product variations and inconsistent URL formats.
The objective is not to make your website artificially small.
It is to ensure that every page you ask a search engine to index has a clear purpose.
A useful rule is:
Every indexable URL should have a reason to exist.
If two URLs provide essentially the same information to the same audience and satisfy the same search intent, you should probably question why both need to be indexed.
Keep canonical signals consistent
Each important indexable page should have a preferred URL.
For example:
<link rel="canonical" href="https://example.com/services/seo">
That URL should ideally be consistent with the version you use throughout the rest of your website.
Your sitemap should contain the same URL.
Your internal links should point to it.
Your structured data should reference it.
Your social metadata should use it.
Problems arise when the sitemap contains one version, your navigation links to another and the canonical declares a third.
Do not make search engines resolve contradictions that you can eliminate yourself.
Keep your XML sitemap clean
A sitemap should primarily contain URLs you genuinely want search engines to discover and process.
That means removing URLs that redirect, return errors, contain noindex, duplicate another canonical page or belong to staging environments.
For larger sites, separate sitemaps can make diagnostics easier:
/sitemap-products.xml
/sitemap-services.xml
/sitemap-articles.xml
/sitemap-locations.xml
If article pages are indexing normally while product pages are struggling, separate sitemap reporting can help expose the pattern quickly.
A sitemap is not a command forcing Google to index something.
Think of it as a clean map of the URLs you believe are important.
3. Build useful internal links
Search engines discover and understand pages partly through links.
That means important pages should not exist only behind a search box, filter, JavaScript button or XML sitemap.
Wherever possible, use normal crawlable HTML links:
<a href="/services/technical-seo">
Technical SEO services
</a>
Internal links are not only a crawling tool.
They should also help users naturally progress through your website.
For example:
Beginner guide
→ Detailed process
→ Service page
→ Case study
→ Pricing or enquiry
That is much stronger than publishing 100 disconnected articles that never guide the reader towards another useful page.
Internal linking should reflect how information on the website relates to real customer decisions.
4. Make every important page easy to understand
Technical accessibility gets the page into the room.
It does not make the page useful.
Once a search engine or AI system can access a page, the next question is whether the meaning of that page is clear.
A visitor should not have to investigate half your website to understand:
- what the company does
- who the product is for
- which problem it solves
- how it works
- what it costs
- where it is available
- what makes it different
- what its limitations are
This sounds obvious, yet many websites hide useful information behind vague brand language.
Consider this heading:
Welcome to the Future
It may sound good in a pitch deck, but it communicates very little.
Compare it with:
Cloud Accounting Software for Small UK Businesses
The second heading tells a new visitor exactly what the page is about.
You can still write creative copy.
Just make sure the important meaning is stated somewhere clearly and prominently.
Answer important questions early
A surprising amount of SEO content takes too long to answer the question that brought someone to the page.
An article about how long SEO takes might begin with several paragraphs about the evolution of search marketing before giving an answer.
Instead, answer the question first.
For example:
A new website may begin showing growth in impressions and non-branded rankings within the first few months, but becoming meaningfully competitive can take six to twelve months or longer. The timeline depends on the market, the website's current condition, content quality and authority.
Now you can explain the variables.
This structure works because the reader gets immediate value without sacrificing depth.
It also makes the information easier for other systems to identify and summarise.
5. Give people the information they need to make decisions
Many business websites are technically optimised but commercially vague.
They talk about flexible solutions, industry-leading services and seamless experiences without providing the details buyers actually need.
Depending on the business, useful information might include pricing, timelines, compatibility, service locations, delivery time, product limitations, eligibility, cancellation terms, examples or implementation details.
Compare these two descriptions.
We offer flexible SEO solutions for businesses of all sizes.
And:
Our technical SEO audits cover crawling, indexing, JavaScript rendering, canonicalisation, internal linking, structured data and Search Console diagnostics. Most audits take two to four weeks depending on the size and complexity of the website.
The second version is easier to understand.
It is easier to verify.
It answers questions a potential customer might genuinely ask.
And it gives search and AI systems much more specific information to work with.
6. Stop publishing content simply to fill a calendar
Publishing more frequently does not automatically create more search visibility.
Before creating another page, ask what purpose it serves.
Who is searching for this information? What decision are they trying to make? What can you add that is not already repeated everywhere? Which existing pages should connect to this page? Where should the reader go next?
A content plan becomes much stronger when it is built around customer questions rather than keyword volume alone.
Those questions usually fall into a handful of useful groups.
People want to understand something.
They want to solve a problem.
They want to compare alternatives.
They want to know whether they should buy.
Or they want validation before making a decision.
Many companies produce an enormous amount of basic educational content while barely creating content around comparison, purchase and validation questions.
The result is predictable.
They get traffic.
But not enough of that traffic turns into revenue.
7. Create content worth citing
AI systems can summarise generic information extremely well.
That means generic information by itself is becoming increasingly replaceable.
Consider an article titled:
10 Benefits of SEO
You can probably predict most of the article before reading it.
It will mention visibility, organic traffic, credibility, user experience and long-term growth.
Thousands of websites already publish essentially the same information.
There is very little reason for another website or AI system to specifically rely on yours.
Now compare that with:
What We Found After Auditing 400,000 Crawled but Not Indexed URLs
That article could contain original data, examples, patterns, technical problems, implementation details and results.
It becomes harder to replace because it contains information that did not previously exist elsewhere.
That is an increasingly important content advantage.
Build content around evidence
Not every article needs an enormous proprietary research project.
Evidence can come from relatively simple sources:
- first-hand experience
- original data
- a real process
- a useful case study
- screenshots
- a specific experiment
- product data
- customer questions
- an expert explanation
- a strong comparison
The difference often comes down to specificity.
Instead of:
Internal linking is important.
You could write:
We found 126 indexable product pages receiving no internal links outside the XML sitemap. After adding category and related-product links, Google began recrawling the affected section more frequently.
The second version contains something concrete.
Specificity makes information useful.
Useful information is more likely to earn links, mentions and citations.
Separate evidence from opinion
Credibility also depends on how claims are presented.
Avoid writing:
Research proves that AI traffic converts better.
unless you genuinely have evidence supporting such a broad conclusion.
A more accurate statement might be:
In our analytics, visitors referred by ChatGPT spent more time on the website than several other referral sources during this period. This was one website, so I would not treat the result as a universal benchmark.
That qualification does not weaken the argument.
It makes it more credible.
Whenever possible, identify where the information came from, when it was collected and what limitations apply.
8. Keep commercially important information current
Search visibility is not useful if the information being discovered is outdated.
This becomes especially important for:
- pricing
- product documentation
- comparisons
- event pages
- policy pages
- statistics
- team information
- locations
- "best tools" articles
Updating the publication date alone does not make a page fresh.
The information itself needs to change when reality changes.
This becomes even more important in AI-led search because an old price or discontinued product can continue appearing in generated answers long after the business has changed.
For commercially important pages, content maintenance should be part of the SEO strategy rather than an occasional cleanup project.
9. Use structured data to clarify, not compensate
Structured data can help search systems understand what type of information exists on a page.
Depending on the page, that might include:
- Organisation
- Person
- Product
- Service
- Article
- BreadcrumbList
- LocalBusiness
- Event
- JobPosting
- VideoObject
But more structured data is not automatically better.
The markup should match what users can actually see on the page and accurately describe its primary purpose.
For a blog article, the fundamentals are often enough:
- headline
- author
- published date
- modified date
- main image
- publisher
- canonical URL
Structured data should make clear information easier to interpret.
It should not be used to compensate for weak or missing content.
10. Build a consistent identity across the web
Search systems need to understand that different references across the internet belong to the same company, product or person.
That becomes harder when a business constantly changes how it describes itself.
For example:
MarketingPal
Marketing Pal
MarketingPal Agency
MP Marketing
Marketing Pal SEO
These may all refer to the same company.
But why create unnecessary ambiguity?
Your official company name, website, products, services, founders, locations and public contact information should be reasonably consistent across your website, social profiles, directories, review platforms and other relevant sources.
This is not about repeating a brand name unnaturally.
It is about reducing uncertainty around who the company is and what it does.
11. Earn relevant links and mentions
Backlinks still matter, but link building becomes more useful when the objective is not simply to increase a number in an SEO tool.
Ask whether the source is relevant.
Does the mention make sense in context?
Is the link pointing to the right page?
Would an actual reader find it useful?
A mention of your company may reasonably point to the homepage.
A discussion of a specific service should usually point to the service page.
A product review should point to the product.
A citation of original research should point to the research itself.
These deeper links help external systems discover the pages that actually contain the useful information.
And the easiest way to earn good links remains remarkably simple:
Create something worth referencing.
Original research, useful tools, case studies, datasets, strong guides and genuinely useful expert explanations give other people a reason to link to you.
12. Measure Google and AI visibility together
Businesses do not need four completely separate marketing disciplines for SEO, AEO, GEO and LLMO.
They need a clear view of whether their company is being found, understood and chosen.
Traditional search reporting should still include metrics such as organic clicks, impressions, non-branded queries, indexing issues, conversions, leads and revenue.
But AI search adds another useful layer.
You may also want to understand:
- how much referral traffic comes from AI assistants
- which pages receive that traffic
- whether those visits convert
- whether your brand appears for important questions
- which competitors appear alongside you
- whether your company is cited
- which pages are cited
- whether the information presented about your company is accurate
The purpose is not to create another vanity dashboard.
The purpose is to understand whether visibility is contributing to commercial outcomes.
Track meaningful questions, not random prompts
AI answers can vary.
The same question may produce different outputs depending on the platform, model, wording, location, sources available and date.
That means checking a random set of prompts once and declaring your "AI ranking" is not particularly useful.
Create a fixed set of questions based on real customer searches.
For example, an accounting software company might monitor questions such as:
Best accounting software for a small UK business
Affordable accounting software for freelancers
Xero alternatives for small businesses
Accounting software with VAT support
Best bookkeeping tool for a five-person company
Then track whether the brand appears, whether it is cited, how it is described, which competitors appear and which sources influence the answer.
Run the same group over time.
Treat the results as a trend rather than a permanent ranking.
Most importantly, connect visibility back to visits, enquiries, sign-ups and revenue.
Visibility is useful.
Business growth is the objective.
What does not work
The growth of AI search has also created an entire category of shortcuts that sound more important than they are.
Publishing hundreds of generic articles
More URLs do not automatically create more authority.
If every article repeats information already available on hundreds of other websites, you are increasing the size of the site without necessarily increasing its usefulness.
Adding schema everywhere
Structured data cannot rescue an unhelpful page.
Markup describes content. It does not replace it.
Treating llms.txt as an AI SEO strategy
An llms.txt file may be a useful supporting resource for systems that choose to use it.
It does not replace crawlability, indexing, useful content, internal links, authority or good site architecture.
Treat it as an optional supporting file rather than the foundation of your strategy.
Creating content only for machines
Do not create awkward pages that no real customer would want to read simply because you believe an AI crawler might prefer them.
Clear answers, complete information, meaningful headings and real facts help both humans and machines.
Tracking mentions without checking accuracy
Being mentioned is not automatically positive.
An AI system can show an outdated price, recommend the wrong product, confuse two companies or misrepresent your positioning.
Measure visibility and accuracy together.
Chasing traffic without understanding intent
Ten thousand people looking for definitions can be less commercially valuable than 500 people actively comparing solutions.
Traffic is an input.
Revenue is the outcome.
A practical 90-day SEO and AI search plan
You do not need to rebuild your entire marketing strategy overnight.
A focused 90-day plan is usually more useful.
Weeks 1–2: Fix access and indexation
Start with the technical foundation.
Audit robots.txt, XML sitemaps, status codes, canonicals, indexation, rendering, duplicate URLs, internal links, orphan pages and Search Console exclusions.
Prioritise pages connected to revenue rather than trying to fix every small warning simultaneously.
Weeks 3–4: Improve your core commercial pages
Review the homepage, product pages, services, pricing, comparisons, locations, important FAQs and contact pages.
Each page should clearly explain what is being offered, who it is for, which problem it solves, how it works, what it costs where relevant and what the visitor should do next.
Weeks 5–8: Build around customer questions
Collect real questions from Search Console, sales conversations, customer support, reviews, internal search, Reddit, competitors and AI assistants.
Then organise those questions by intent.
Prioritise problem-solving, comparison and buying questions rather than producing endless introductory articles.
Weeks 9–10: Add proof
Look for places where the website makes broad claims without evidence.
Add case studies, examples, screenshots, original data, product information, expert explanations and clear methodology.
Replace generic statements with specific evidence wherever possible.
Weeks 11–12: Measure what changed
Create one reporting view covering traditional search visibility, AI referrals, important landing pages, leads, sales, conversions, brand mentions and the accuracy of those mentions.
Then use those results to decide what deserves attention next.
SEO should become an improvement loop rather than a publishing treadmill.
What I have seen in practice
I have seen this shift directly in client work.
On one project, first-touch users from ChatGPT increased by 569%.
It would be easy to package that result as proof of some secret AI SEO framework.
That is not what happened.
Most of the work was extremely familiar.
We removed thin and low-value pages, strengthened important product pages, added clearer specifications and use cases, improved titles and headings, fixed crawlability, improved internal linking and made the site's product information easier to understand.
ChatGPT later sent around 1,800 active users to the website.
The result looked like "AI growth."
Most of the work looked like good SEO.
That is the important lesson.
The channels are changing.
The interfaces are changing.
The way people ask questions is changing.
But businesses still need to make their information accessible, useful, specific, accurate and trustworthy.
The future of SEO is bigger than rankings
SEO is no longer only about winning one of ten blue links.
People discover businesses through Google Search, AI Overviews, ChatGPT, Perplexity, Gemini, Reddit, YouTube, review platforms, industry websites and communities.
That does not mean you need a completely different optimisation strategy for every platform.
It means you need to become one of the clearest and most useful sources in your category.
Start with the fundamentals.
Make the website accessible.
Make important pages indexable.
Keep your information complete and current.
Answer real customer questions.
Create evidence worth referencing.
Build genuine authority outside your own website.
And measure visibility alongside commercial outcomes.
That is still SEO.
It is also the foundation of AI SEO.
Want to know where your search visibility is leaking?
I work with companies on technical SEO, content, website structure and AI search visibility. If you're getting traffic without enough growth, struggling with indexation, or want to understand how your brand appears across Google and AI search, I can help you identify what needs to be fixed first.
Book a callAbout the author
Tanuj Sharma
I'm a growth marketer and SEO consultant working across technical SEO, AI search visibility, content strategy, website architecture and product growth.
More about me →Published July 30, 2026, last updated August 13, 2026.