The problem. A growing share of searches now end without a click. The person asks a question, receives a composed answer, and never visits a website. If your SEO strategy assumes that ranking produces traffic, part of that assumption has quietly stopped holding.
The response is not to abandon SEO. It is to recognise that being ranked and being cited are now two different outcomes requiring partly different work.
What it costs the business
Reporting stops describing reality first. Impressions hold, clicks fall, and the obvious conclusion is that rankings slipped. Often they have not. Your page is being read, summarised and used, and the visit is simply not happening.
The strategic cost is larger. If a buyer forms their shortlist inside an AI answer and your business is not named in it, you are excluded before the stage where your website could have persuaded anyone. You do not get to compete on the strength of your site if you are absent from the summary.
What is different about being cited
Classic ranking rewards a page that comprehensively covers a topic. Citation rewards something narrower: a passage that answers one specific question cleanly enough to be lifted.
The practical consequences are unintuitive.
- Structure matters more than length. A clear question as a heading with a direct answer immediately beneath it is far more liftable than the same information distributed across four elegant paragraphs.
- Specificity beats comprehensiveness. A concrete figure, a defined range, a named process. Vague coverage gets summarised into somebody else words.
- Being quotable requires being decisive. Hedged writing that avoids committing to anything gives a model nothing to attribute to you.
- Consistency across the web matters. Models weigh corroboration. If your claims appear consistently across your site, your listings and third-party mentions, they are more likely to be repeated.
What to actually change
Write the question as a heading, and answer it in the first two sentences. Then elaborate. This is the single highest-leverage structural change, and it also improves the page for human readers in a hurry.
Publish the specifics you have been withholding. Ranges, timelines, what drives cost, what the process actually involves. Models cannot cite what you decline to state, and neither can buyers.
Use plain markup properly. Real headings rather than styled text, ordinary lists, sensible tables, and structured data where it fits. Machine readability is not a trick, it is just not making the parsing hard.
Keep your entity details identical everywhere. Name, location, services, contact. Inconsistency between your site and your listings weakens confidence in all of it.
Have something to be cited for. Original data, a documented method, a first-hand account. Pages that restate the consensus get summarised into the consensus, without attribution, because there is nothing there that needed you.
What to measure now
Click-through rate on informational queries becomes a diagnostic rather than a performance metric. Falling clicks against stable impressions is the signature of answer extraction, not of a ranking problem, and treating it as the latter leads to pointless rewrites.
Beyond that: brand search volume, direct traffic, and simply asking new enquirers how they found you. When discovery happens inside an answer, attribution moves out of your analytics and into the conversation.
What has not changed
Most of it. Sites that are technically sound, load quickly, cover their subject properly and earn genuine references still do well in both systems. Commercial queries still produce clicks, because someone ready to buy wants to see who they are buying from.
The shift is at the informational end, and the correct reading is that informational content now has to earn its citation rather than merely occupy a position.
Our SEO and GEO services page sets out how we approach both, or send us a query you want to be cited for and we will tell you what is missing.
