Blog
Notes on getting cited by AI.
How answer engines decide who to name, and what earns a brand a place in the reply.
- Where to start with GEO The first month is about learning what the machines already say, not fixing everything at once.
- How we became our own first case study You most likely found us through an AI. That was the plan.
- If AI gets far smarter, does GEO still matter? The smarter the system, the more it rewards being a source worth trusting.
- Will smarter models still bother citing sources? Grounding is not a limitation of today's AI. It is how trustworthy answers get built.
- When machines do the thinking, trust is the bottleneck Capability is getting cheap. Being believed is not.
- The business case for being the answer Why a mention with no click can still be the most valuable thing you own.
- A few GEO myths worth dropping The field is young enough that confident nonsense spreads easily.
- Why proprietary data earns the citation If the fact can only come from you, you become the source.
- The state of AI search in 2026 A rough map of where the answers are coming from and why it matters.
- When the honest answer about you is unflattering You cannot scrub a reputation, but you can work on deserving a better one.
- AI and the local recommendation Ask an assistant for somewhere nearby and it leans on the same old signals, sharpened.
- AEO, GEO, AI SEO: untangling the words The labels multiply faster than the ideas behind them.
- Retrieval, RAG, and what the model actually fetches Understanding the plumbing explains why clear pages win.
- Cutting weak pages to lift the strong ones A smaller, sharper site often reads better to a machine.
- Question-shaped content, answered straight When your page asks the buyer's question and answers it cleanly, machines notice.
- Third-party reviews quietly steer AI suggestions What the review sites say about you, the assistants tend to repeat.
- Comparison content and how AI recommends So much buying is a choice between named options, and models know it.
- How recency shapes what AI tells people Engines favour current sources, so stale pages fade from the answers.
- Markup that machines can parse without guessing Clean structure is quietly one of the kindest things you can do for a crawler.
- When agents start doing the buying Assistants are edging from advice toward action, and that changes the stakes.
- Depth beats a scattering of one-off pages Models trust sources that clearly own a subject.
- When your JavaScript hides you from the machines If the content only appears after scripts run, assume it is unseen.
- Let the AI crawlers actually read you Plenty of brands block the very systems they want to be seen in.
- Do backlinks still matter for AI search? Yes, but their job has shifted from votes to visibility.
- Earning citations the way you earn press The old craft of getting written about has a new payoff.
- The numbers a GEO programme should live by A short set of honest metrics beats a dashboard full of vanity.
- GEO for agencies, consultancies and service firms Reputation is your product, and models read reputation closely.
- GEO for ecommerce brands Shoppers are asking assistants what to buy, by category and by need.
- GEO for SaaS companies Buyers ask assistants to shortlist tools. You want to be on the list.
- How a model forms a view of your company It is less a database entry than an accumulated impression.
- Map the questions your buyers actually ask Visibility is meaningless until you tie it to real prompts.
- You cannot improve what you never check Watching how engines describe you is the unglamorous core of the work.
- When the answer about you is simply wrong Fabricated facts about your brand are a problem you can work on.
- Original data is citation bait, in the good way Models love a number they cannot find anywhere else.
- Why assistants keep quoting Reddit Models trust lived experience, and forums are full of it.
- Why Wikipedia and Wikidata move AI answers Models treat them as a backbone of fact, for better and worse.
- Authority signals the models lean on Experience, expertise and trust still decide who gets believed.
- Getting cited by Chinese AI models DeepSeek, Qwen, Doubao, Ernie and Kimi read a different web, and it changes the playbook.
- Gemini, Google's own assistant, and your brand It leans on Google's understanding of the web, which you already influence.
- Being found inside ChatGPT When it searches, ordinary web signals decide what it reads.
- What a model knows versus what it looks up The difference decides whether fresh facts about you can show up at all.
- Showing up in Google's AI Overviews The overview sits above the links, and it plays by familiar rules.
- Getting cited in Perplexity Perplexity shows its sources, which tells you exactly what to aim at.
- Mentions are starting to matter more than links A model can recommend you without ever linking to you.
- The zero-click answer is the normal case now A lot of your audience gets what they need without ever arriving.
- Why we built QWD We kept watching good companies get described badly by machines.
- Write pages a model can actually quote Answer engines lift sentences. Give them good ones.
- Structured data, read by machines that are guessing Schema will not rank you, but it stops the model from guessing wrong.
- Why your facts have to match everywhere Models cross-check. Contradict yourself and you get left out.
- llms.txt, and whether it is worth your time A small file with a sensible idea and a modest payoff.
- Share of answer, the metric that replaces rank If you can only track one GEO number, track this one.
- The quiet move from search boxes to answers The interface barely changed. The behaviour behind it did.
- How answer engines decide who to cite What happens between a question and the names that show up in the reply.
- How GEO differs from the SEO you already know The habits carry over. The scoreboard does not.
- What generative engine optimization actually means A plain definition of GEO, and why it is a different job from classic SEO.