SEO and GEO guide
Answers to questions about SEO, GEO and AI visibility. Every article answers in its first paragraph, then shows how to check it on your own site.
SEO
Indexing, titles, structure, links and what all of it costs. Each article answers one question and rests on a primary source.
Read more →
GEO and AI
GEO, AEO and AI SEO are one body of work measured by three questions. Here is what each means and what to actually change on the page.
Read more →
Speed
Core Web Vitals, the PageSpeed score and what they actually measure. A number without an address changes nothing, so we talk about specific files.
Read more →
Security
Certificates, security headers and what a site says about itself for no reason. We demand only what matters to an ordinary website.
Read more →
Benchmarks
A score means nothing without a comparison. Here is what ordinary looks like, what good looks like, and what separates them.
Read more →
What the terms you keep hearing mean
Thirty-four terms from SEO, GEO, speed and security. Each defined exactly as far as a decision needs — with the caveat where one is real.
Read more →
-
Which free SEO tools exist and what does each one actually show?
Several free SEO tools exist and they answer different questions. Search Console shows what Google knows about you. PageSpeed measures speed. Lighthouse audits one page. Screaming Frog crawls up to 500 URLs. Rich Results Test validates schema. None of them answers every question.
-
How do I check whether AI sees and cites my website?
A GEO audit checks three things in order. First, whether an AI crawler can reach the site and whether any content exists without JavaScript. Second, whether the page declares facts through structured data, authorship and dates. Third, whether the text is shaped for citation. Each step has its own method.
-
What is AI SEO and how does it differ from GEO and AEO?
AI SEO is the umbrella name for the whole field — adapting a website for AI-driven search. GEO is about whether a model knows and recommends you, and AEO is about whether a direct answer gets lifted from your page. Technically they overlap almost entirely, so separate task lists do not exist.
-
What do the sites scoring above 80 actually do?
Sites above 80 points almost never use anything special. They completed the ordinary list: indexing with no leftovers, mobile speed, full structured data with author and dates, unique titles and descriptions, a deliberate robots.txt, and answers on the pages. What makes the difference is completeness, not budget.
-
How do internal links work and how do you lay them out properly?
Internal links are links between your own pages. They do three things: show a crawler the route to a page, pass along a share of the site's authority, and describe what the target page is about through the link text. A page with no incoming internal links often goes unindexed entirely.
-
Does the SEO score differ by industry?
Scores differ less in size than in where they are lost. Shops most often lose points to duplicates, filter URLs and speed; service sites to thin content and missing structured data; publishers to indexing and third-party script weight. An identical number in these three means a different list of jobs.
-
What is Markdown content negotiation and why should a site support it?
Markdown negotiation serves two versions of a page from one URL: HTML to a browser, and plain text without layout to any agent sending an Accept: text/markdown header. The agent receives only what it actually reads and spends several times fewer tokens. No new URLs are created for it.
-
Which SEO and GEO numbers are worth citing?
This page holds two kinds of numbers, deliberately kept apart. The first are thresholds from primary sources: Core Web Vitals limits, title and description lengths, our score scale. The second are our assessments from working on websites, labelled as assessments rather than as the findings of a study.
-
How is the SEO score calculated and what does the letter mean?
A scan returns five separate scores from 0 to 100: SEO, GEO, AEO, speed and security. Each is reduced only by confirmed findings, weighted by impact; passed, not-applicable and unverified checks always have zero impact. A letter accompanies the number: A+ from 97, A from 92, B+ from 86, B from 78, C+ from 70, C from 60, D from 45, below that F.
-
What is E-E-A-T and how do you show experience and authorship on a page?
E-E-A-T stands for experience, expertise, authoritativeness and trust, taken from Google's quality rater guidelines. It is not an algorithm signal but a set of criteria the algorithm tries to approximate. In practice it comes down to four visible things: an author, a date, sources, and facts nobody could write without having done the work.
-
How ready are websites for AI search?
On the AI visibility side an ordinary site stands lower than on the SEO side — usually between 30 and 50 points. What is missing is not content but four things: a deliberate decision about AI bots, text without JavaScript, questions with answers, and verifiable facts. All four are fixed once and cost no monthly invoice.
-
What kind of pages does AI cite most often?
Four kinds of pages get cited most: a question with its answer directly underneath, pricing and terms with concrete numbers, definitions and glossaries, and primary documents. They share one property — you can take two or three lines out of them and those lines stay true without the rest of the page.
-
How much do SEO services cost and what are you actually paying for?
Ongoing SEO work typically costs between 300 and 2000 euros a month, and a one-off technical audit between 300 and 1500. The difference is driven by scope and competition in your niche, not by the quality of the provider. A fixed price quoted without looking at your site is the first sign scope was never discussed.
-
Is llms.txt actually needed, and what does a good one look like?
llms.txt is still a proposal that no vendor has publicly committed to reading, so its benefit today is small and unproven. But it is cheap. A good file is short, maintained by hand, and lists live URLs with one line of context each — not a sitemap copy, and not a list that went stale after the first restructure.
-
Which structured data is actually used, and where does it go wrong?
In practice we see two extremes: either no structured data at all, or a template Organization node and nothing else. The mistakes are consistent too — data that does not match what the page shows, Article without an author or dates, Product without a price or availability, and FAQPage on a page that has no FAQ.
-
Do websites actually block AI crawlers?
Most of the time AI bots are neither blocked nor admitted deliberately — robots.txt simply does not name them. What happens then depends on the general rules: sometimes everything passes, sometimes everything is blocked along with the rest. Either way it is a decision nobody actually made.
-
What are sitemap.xml and robots.txt for and how do you set them up correctly?
robots.txt tells crawlers which parts of a site not to fetch, while sitemap.xml lists the URLs worth reading along with the date each last changed. Neither forces Google to index anything, but a mistake in either one can shut an entire website out of search.
-
What SEO score is normal for an ordinary website?
The usual score we see is between 30 and 60 out of 100. About 50 of that is already there on launch day — that is what a modern template gives you. Above 80 is rare, and it comes not from a bigger budget but from someone fixing indexing, headings, structured data and mobile speed once.
-
How many websites are ready for AI search?
Usually not. Across roughly 500 audits the AI visibility side almost always stands lower than SEO. Four things are missing: a deliberate decision about AI bots, text without JavaScript, questions with answers, and verifiable facts. None of them can be bought — they are assembled inside the site.
-
How do you get into Google AI Overviews and how is it different from ranking?
AI Overviews is a Google-generated summary at the top of the results that cites a handful of sources. What gets cited is not the page ranking first but the page holding a precise, self-contained paragraph answering one specific part of the question. A tenth-place page can be cited while the first place is not.
-
What state is SEO in across ordinary websites?
Across roughly 500 audits the same pattern repeats: a site with a strong link profile sits at about 60 out of 100 technically — roughly what it had on its first day. Links and articles raise authority but never touch indexing, speed, structured data or internal links.
-
Which side is losing you more right now?
There is nothing to choose — only the question of which side loses you more. Across roughly 500 audits SEO and GEO stand on the same foundation: indexing, speed, structure. What differs is the top layer, and bought links raise only the SEO half.
-
Why did my organic traffic from Google drop and how do I find the cause?
Organic traffic drops have four distinct causes, and the shape of the fall separates them. An overnight collapse means a technical fault or a manual action. A gradual slide over weeks means an algorithm update or a competitor. One page falling alone means a problem with that page. Rankings holding while clicks fall means zero-click search.
-
Which Schema.org structured data types does a website actually need?
Structured data is a JSON-LD block that repeats your page content in a form machines can read. In practice you need four types: Organization, WebSite, BreadcrumbList and a page type — Article, Product, Service or FAQPage. The remaining Schema.org types do nothing for most sites.
-
Why are two of my pages competing with each other?
Cannibalization is when several of your pages answer the same question. Google cannot show both, so it picks one — often not the one you would choose — and the position starts swinging between them. The fix is almost always merging, not optimising.
-
What does "not enough data" mean in the Core Web Vitals report?
It means your site has too few real Chrome visitors. The Core Web Vitals assessment is based on CrUX — measurements from real people, not a lab test. Until there is enough traffic there will be no assessment, and no amount of page speed work changes that.
-
How do you build a page so that AI cites it?
A model does not read a page in order. It splits the text into passages and picks the one that answers the question, which makes headings the boundaries of those passages. That produces four jobs: a question as the heading with the answer directly beneath it, 40 to 120 words per passage, JSON-LD with Organization and Article nodes, and access for answer crawlers in robots.txt. Internal links tie pages on one topic together, so a model finds a topic hub rather than a single answer.
-
Which GEO and AI SEO claims are myths?
The ten most common GEO myths are these: llms.txt is supposedly required, GEO will supposedly replace SEO, a model can supposedly be fooled with keywords, Google search will supposedly drop 25 percent, a page supposedly needs 1000 to 1500 words, content written with AI is supposedly already GEO, blocking every AI crawler is supposedly safest, FAQPage schema will supposedly earn a rich result, the result can supposedly be guaranteed, and ranking is supposedly the same as being cited. None of them survive a check against primary sources.
-
How do you measure whether GEO work is actually working?
GEO results can be measured in four ways: traffic from AI sources tracked separately from other traffic, Search Console impressions and positions, a monthly citation check against the same list of questions, and technical check results before and after the work. Three things cannot be measured: how many people read the answer without clicking, when a model will refresh its data, and why in one specific case it cited one source rather than another.
-
Why is my website slow, and where do I start?
Almost always three things, and almost always in this order: oversized images, render-blocking CSS and JavaScript, and a slow server response. Everything else is detail that only becomes visible once those three are handled.
-
How do you check a GEO expert's claims before paying?
Ask for a concrete deliverable, verifiable example and measurement method. Unsupported citation guarantees or unclear scope warrant further questions. Missing robots.txt, schema or an AI mention alone does not prove that a specialist is fake.
-
Why does GEO need monitoring rather than a single audit?
An audit shows what is on the site today. But a site changes without its owner knowing: after a CMS update, after an editor's fix, after a plugin adds noindex, or when server settings change. None of these announce themselves. So what is needed is not a second identical report but monitoring: the change against the previous scan, an email only when something changed, and a history that shows when it happened.
-
How do you choose a GEO specialist in Lithuania?
Choose a GEO specialist by a clear scope, verifiable examples and a measurement method. Ask them to separate the technical audit, content work and AI answer monitoring. This guide helps compare proposals; it is not an agency ranking or an endorsement of a particular provider.
-
Why doesn't Google index my pages?
The usual reason is neither a penalty nor an algorithm secret. The page is either technically unreachable for Googlebot, or it asks not to be indexed (`noindex`, a `canonical` pointing elsewhere), or Google found it, read it and decided it adds nothing new to what the index already holds.
-
How do I write a clear title and meta description?
Write a unique title that identifies the page's purpose and a meta description explaining what the visitor will get. Keep both consistent with the H1 and visible content. Search engines choose their own result titles and snippets, so displaying your exact wording is not guaranteed.
-
What is llms.txt, and do I actually need it?
llms.txt is a plain text file at the root of a site that tells AI systems what the site is and which pages to read first. It is not a standard and no vendor has publicly committed to reading it — useful, but not the first job on the list.
-
Can ChatGPT see my website, and how do I check?
Check technical access and citations in specific answers separately. Access needs rule and server evidence; a citation needs a prompt, date and real source link. One answer or an audit score cannot establish your overall visibility in ChatGPT.
-
What is AEO, and how do I get into the direct answer?
AEO (Answer Engine Optimization) is shaping a page so a direct answer can be lifted from it — a Google featured snippet or a paragraph inside an AI answer. What decides it is not length but structure: the question as a heading and the answer immediately below.
-
What is GEO optimization, and how is it different from SEO?
GEO (Generative Engine Optimization) is preparing a website so that AI assistants — ChatGPT, Gemini, Perplexity, Claude — understand it, cite it and recommend it. It differs from SEO in its goal: SEO wants the click on your link, GEO wants your answer inside the text the person reads without clicking anything.
-
What do I do about "Discovered – currently not indexed"?
It means Google knows your URL but has not once come to read it. This is not an error or a penalty but a decision about importance: the site has many URLs and this one does not look worth the queue. Internal links and fewer junk URLs are what help.
-
Should I block AI crawlers?
Choose bot rules by purpose. You can allow AI search while restricting training use. OpenAI's search and training permissions, for example, are independent. Check each provider's documentation and server access before changing rules.
-
How do I set up two languages and hreflang correctly?
Each language needs its own stable URL, and both versions must point at each other. Google only accepts an `hreflang` group when it is reciprocal: if A points to B but B does not point back, the whole declaration is ignored and the two languages start competing with each other.
-
Why is an SEO audit in the local language more useful?
The difference is measurement, not translation. Lithuanian text written without diacritics, hreflang pairs between the Lithuanian and English versions, and local search competition are things an English-first tool either does not check or scores incorrectly.
-
What does the PageSpeed score actually mean?
The PageSpeed score is not your site's speed. It is a weighted average of five lab metrics from one run on a simulated mid-range phone with a throttled network. Real visitor experience shows up in field data, and the score varying between runs is normal.
-
Which security headers do you need and what do they do?
An ordinary site needs six: HSTS, Content-Security-Policy, X-Content-Type-Options, Referrer-Policy, X-Frame-Options and Permissions-Policy. Five of them are a single line in the server config, while CSP needs tuning — which is exactly why it gives the most.
-
What do you check to judge a website's security?
Four things are checkable from outside: whether the TLS certificate is valid and HTTP redirects to HTTPS, whether security headers are sent, whether the page has mixed content, and whether admin or configuration paths are publicly reachable. The fourth layer — patching — is only visible from inside.
-
How do you write an FAQ so an answer can be taken from it?
Every question must be a real heading (h3), with the answer immediately below it. Headings are passage boundaries: without them a twenty-question block is one large text, and a model cannot cut out the answer to a single question.
-
How does a page get into Google's featured snippet?
Google takes the featured snippet from results that already rank on page one, then picks the one whose passage answers most precisely. So the work is twofold: first reach the top ten, then format the answer so it can be cut out cleanly.
-
What makes an AI assistant cite one source over another?
A model picks a passage, not a page. The winner is the one that answers on its own, is reachable without JavaScript, rests on a verifiable fact and carries a visible date. Position in Google results is neither a condition here nor a guarantee.
-
What do you do when AI answers take traffic from the site?
Short factual questions no longer lead to a website — the assistant answers them itself. What remains are the questions that need a decision, a price or an action. So the work moves from volume to precision: fewer visits, but closer to a purchase.
-
How does a website get into AI assistant answers?
You need three things at once: do not block AI crawlers in robots.txt, serve the content from the server without JavaScript, and structure the text so the answer to one question is one passage. A blocked crawler sees nothing, and empty HTML means an empty page to a model.
-
When is an automated SEO audit better than a manual one?
Automation wins where a check is a rule and repeats: indexing, titles, links, speed, schema. A person wins where judgement is needed — about topics, copy and priorities. The worst option is doing by hand what a machine does more accurately.
-
What does a technical SEO audit actually cover?
A technical audit checks whether a search engine can reach the page, read it, understand it and show it. Four layers run in strict order: server response, indexing rules, rendering with JavaScript, and only then titles, schema and internal link structure.
-
What does a free SEO scan show, and what does it miss?
A free scan answers the technical question: is the page reachable, is it understandable to search and to AI, and is anything stopping it from being shown. What it does not answer is whether the topic is worth writing about at all, and whether your offer beats the one next door.
-
Where do you start when fixing a site's SEO?
Start with indexing, not keywords. If Google does not have the page, nothing else has anywhere to work. The order is: are the pages indexed, do the titles tell pages apart, does the content answer the question, is speed getting in the way, and only then — are the numbers moving.
Updated