Brayan Iliev
How to read your results: what drops mean and when not to take action
When ad results drop, panic leads to chaotic changes and even worse data. In this art...
Contact
100% FREE
A few years ago, “classic” SEO seemed predictable: you optimised your on-page elements, built quality backlinks and waited for Google to reward you with a place in the top 10. Today the landscape has changed completely because of ChatGPT-style answers, Perplexity, Google AI Overviews and other new formats. When users ask a question, they first […]
A few years ago, “classic” SEO seemed predictable: you optimised your on-page elements, built quality backlinks and waited for Google to reward you with a place in the top 10.
Today the landscape has changed completely because of ChatGPT-style answers, Perplexity, Google AI Overviews and other new formats. When users ask a question, they first see a summary generated by a large language model (LLM), and only then the classic links. This means your content can be used in the answer even if the user never clicks on it. That is why AIO optimisation (AI-optimised content) – or, more broadly, AI search optimisation – is becoming a key factor.
If your content is well structured, useful and “understandable” to LLMs, you can appear as a source in AI-generated answers – which brings exposure, authority and potential traffic, even without direct clicks. In other words, optimising for AI answers is no longer a “bonus” but an essential part of any modern strategy.
What does this mean for WordPress sites? More and more often, traffic starts its journey not in the standard SERP but in an LLM summary that pulls fragments from your text. If you do not signal the topic and the quality of your information properly, you risk staying out of the model’s view – and therefore far from the user.
The new battle is for “trusted source” status. Google already talks about E-E-A-T (Experience, Expertise, Authoritativeness, Trust), and AI systems add a layer of transparency: they want to know who wrote the content, when it was updated and why it should be trusted. Reaching that status is not just a matter of meta tags; it takes an overall strategy in which classic SEO and the new AI optimisation (AIO) intertwine.
In the traditional SERP, Google assessed links, anchor text and PageRank to rank results. In an AI scenario, the LLM first “reads” the web and then synthesises a direct answer. The difference? Users get a quick answer without clicking through five pages. For you, this means fewer organic clicks, but also a new opportunity: if your URL is cited below the answer, your CTR can jump, because you are positioned as a source of “trusted” information.
It is important to know that AI systems cite more selectively. They usually show only 3–4 links with a short summary. That is why AIO optimisation (AI-optimised content) and the broader AI search optimisation strategy are no longer exotic but a necessity. If your content is not structured and optimised for language models, your chances of being included among those links drop significantly.
In other words, optimising for AI systems does not replace SEO – it builds on it. It involves clear headings, direct answers to questions, factual accuracy and a well-formatted structure: everything that helps LLMs “understand” and choose your page.
LLMs work with semantic vectors, not just keywords. That is why contextual matches – synonyms, co-occurrences and topical connections – become critical. If you are writing an article about “WordPress site speed”, also add concepts such as “Core Web Vitals”, “caching” and “CDN” to build rich context.
Another factor is how current the data is. Models prefer pages that are updated often and carry clear time stamps. An author profile with a short bio and a link to LinkedIn strengthens the E-E-A-T signal and increases credibility.
A quick reminder about classic SEO, because without it AIO has nowhere to start.
First, technical optimisation still comes first: speed, mobile-friendliness and security. Google still penalises sites that take more than 2–3 seconds to load and fail the Core Web Vitals tests.
Second, on-page factors – titles, descriptions, internal links – provide the initial context that AI then “enriches”.
Third, external authoritative links remain the foundation of trust signals.
Piling on plugins kills Core Web Vitals. Combine caching (WP Rocket or LiteSpeed Cache), a CDN (Cloudflare) and optimised images (WebP) for a TTFB under 200 ms. Mobile-friendliness comes from a responsive theme plus critical CSS. Do not forget automatic SSL – HTTP/2+ speeds things up for people and bots alike.
Use semantic variations – “WordPress optimisation”, “speeding up a WP site” – instead of repeating the same phrase. Internal links lay out the topical map for the crawler; anchor text should be descriptive, not “here”.
A guest post on a niche blog, an interview with an expert or a mention on a podcast – these still work. Add short author bios and a last-updated date to show experience and freshness.
AIO is the natural evolution of SEO: optimisation not only for algorithms that rank links, but also for language models that generate answers directly. While classic SEO aims for “rank me in the top 3”, with AIO optimisation the goal is “cite me as a reliable source”. This is the new context in which your content can be rewarded – even without a click.
AIO does not replace SEO; it complements it. AI search optimisation adds a new layer to familiar tactics: semantic clarity, verifiable authorship, UX signals and structured data, which LLMs read as indicators of quality.
In practice, this means:
Organising your content into well-structured topic clusters
Using schema.org and other structured data that make machine understanding easier
Monitoring behavioural signals such as bounce rate, time on page and engagement, which can act as proxy metrics for trust in the eyes of AI systems
All of these approaches work in WordPress: with the right architecture, suitable plugins and UX improvements, you can turn your site into a source that LLMs prefer.
In other words: if SEO gets you close to the first page, AIO optimisation puts you in the answer itself.
The pillar–cluster model organises a main “pillar” page (e.g. “WordPress optimisation”) and a series of in-depth articles linked to it (“Image optimisation”, “Caching”, “CDN”). This helps the model understand that you cover the topic comprehensively – and increases the chance that it is your information the bots show.
JSON-LD markup (Schema.org) makes facts machine-readable. Add FAQ or HowTo schema and you will see ChatGPT extract exact paragraphs from your page more often. Plugins such as Schema Pro or WordLift make the process easier.
Bounce rate is now interpreted as a lack of trust. Session duration shows engagement: the longer readers stay, the more likely the AI is to consider your content high quality. A readable font, broken-up paragraphs and tables with quick facts all help.
The bots that crawl your website are no longer just search engine crawlers; LLM systems such as OpenAI, Anthropic and Perplexity send bots too. The llm.txt file tells them what they may and may not access – an equivalent of robots.txt, but specifically for language models.
The first advantage is control: you can block sensitive folders (e.g. members’ areas) from AI crawling without getting in Googlebot’s way. The second is traceability: your logs will clearly show when an LLM crawler visited the site.
The file goes in the root directory and is served with the MIME type text/plain. The syntax is simple: Allow: /blog/ or Disallow: /private/. Use it if you have content you want to keep away from LLMs (e.g. PDF courses) or if you want to point the bot to specific resources.
The easiest way to create a basic llm.txt file in WordPress is to use one of the many plugins already available in the plugin directory. “Website LLMs.txt” is the plugin we settled on. It lets you choose which post types to include in the llm.txt file, with further settings for adding the file to the sitemap and for how often it is updated (daily, weekly or on every change).
AI search engines still read the classic <title> and <meta name="description">, but new fields are appearing too. Metadata is becoming the bridge between your content and the summary the model generates.
<title> – Include the main keyword within the first 60 characters.
<meta name="description"> – A 150–160 character synopsis that mentions the benefit to the user.
robots – The attributes max-snippet:-1 and max-video-preview:-1 signal that you allow full extraction.
We suggest two experimental tags that some LLMs are already testing:
ai-topic states the topic of the page – copy the H1 word for word to avoid ambiguity.
ai-content declares whether the text is fully AI-generated (true), partly (partly) or human-written (false).
In WordPress, the easiest way to add these tags is with a plugin such as “Header and Footer Scripts”.
LLMs love structured answers. Mark up your FAQ sections with JSON-LD to allow direct extraction:
Mark up price, availability and rating. In AI snippets, this lets the model present a current offer directly in the answer. WordPress plugins such as WooCommerce SEO framework already support Product schema out of the box.
The plugin ecosystem is reacting quickly. Popular SEO extensions are building in AI features, and new tools promise automatic llm.txt management.
Yoast added Content AI, which suggests semantic keywords and FAQ ideas.
RankMath has an AI writer for meta descriptions that avoids repetition.
All-in-One SEO is testing an AI generator for schema markup, keeping errors to a minimum.
“Website LLMs.txt”, the plugin we mentioned above, automatically creates and updates the file with every new post or at a set interval.
AIO success shows up not only in organic sessions but in new metrics that are still taking shape.
Server logs will start flagging AI crawlers (e.g. OpenAI-UserAgent). You can count these visits as “AI impressions”.
Second, the number of citations in generated answers is an indirect metric – the more there are, the more visible you are.
Google Search Console – AI Overview (beta) shows how many times your URL has been used in overviews.
Perplexity Dash reports citations by domain.
Custom GA4 events can log referer with the value llm, if the crawler follows a policy.
Regularly ask ChatGPT or Perplexity questions from your niche and see whether they cite you. Keeping a table of results over several weeks will show you progress or decline.
AI search engines do not cancel out SEO – they extend it. By combining classic good practice with AIO techniques – structured data, llm.txt and meta tags – you build “human” and “machine” value at the same time. Keep an eye on the new metrics, experiment with plugins and don’t be afraid to let your content speak for itself.
Need guidance? Talk to the Evol.bg team!
