How AI Assistants Decide Which Brands To Recommend

We also know the picture is unstable. Retrieval strategies are revised without announcement, and a method that explained answers well six months ago may explain them poorly today. Anyone selling certainty here is selling something they do not have.

There is almost always a specific, findable reason for this, and it is rarely that the model dislikes you. Here are the causes worth checking, roughly in the order that they tend to be responsible. llm seo

Which Publications Actually Get Cited Do not assume, and do not use prestige as a proxy. Run ten prompts in your category and count the cited domains, because the pattern varies enormously by industry.

The distinction to draw is between flat results with the inputs completed, and flat results with the inputs missing. The first is a category or timing problem and may be worth persisting with. The second is a delivery problem.

Coverage in a publication nobody cites is still worth having for reach and credibility. It is simply not visibility work, and counting it as such is how a public relations budget gets defended for years without moving anything measurable in this channel. llm seo

Entity Coherence Before a model can recommend you it has to be confident that the scattered mentions of your name refer to one company. That confidence comes from consistency across the details that identify you.

Where the Distinction Does Matter One place, and it is worth being alert to. Read broadly, answer engine optimization includes surfaces that are not generative at all, such as featured snippets and structured result features.

Ask what was done, not what happened. If listings were corrected, pages rewritten and outreach attempted, and the numbers are still flat, that is information about the market. If none of it happened, the numbers were never going to move.

The emphasis is on being included in a generated response, whether or not you are cited by name and whether or not it produces a click. The term appeared in academic work before agencies adopted it, which gives it slightly firmer footing than the alternatives.

One thing worth deciding before you start is who inside the business will answer factual questions. This work generates a steady trickle of small queries about lead times, price ranges and what you will and will not take on, and an agency that cannot get answers will either stall or guess. Naming one person and giving them twenty minutes a week removes the most common cause of these projects drifting.

Expect the timeline to be uneven. Crawler access can change what an assistant sees within days, because retrieval happens at answer time. Identity consistency takes longer, since scattered mentions have to be re-crawled before they join up. Third party coverage is slowest of all and is the part you control least directly, which is exactly why it is worth starting on it before you need the result.

It is also worth checking whether you are being confused with somebody else rather than ignored. Short names, generic names and names that begin with a number collide with other organisations more often than distinctive ones. Where that is happening, the answer will contain facts that are true about a different company, which reads as a hallucination and is usually an identity collision with a specific fixable cause.

That emphasis is worth watching, since retrieval is where most current influence actually lies. A proposal built primarily on getting into training data is describing a slower and far less controllable mechanism than one built on being retrievable now.

You Have No Stable Identity Models need to connect scattered mentions to a single entity. If your company appears under three different spellings, lists two different founding years, and gives an address on your site that does not match your directory listings, those mentions may never be joined up.

Track coverage for accuracy rather than only for volume. A monthly search for mentions of your company, read with an eye to whether the details are correct, produces a steady stream of small correction requests with a high acceptance rate. It is unglamorous work, it costs an hour, and it repairs sources that may otherwise feed answers about you for years.

Three acronyms, considerable overlap, and no governing body to settle the definitions. Different agencies use them differently, some interchangeably, and a few have invented a fourth to differentiate a proposal.

This is usually a few days of work, it frequently explains a poor baseline entirely, and it improves conventional search as a side effect. It is the cheapest part of the whole exercise and the most commonly skipped.

Days Fifteen to Thirty: Fix the Plumbing Someone technical checks that the crawlers feeding AI systems can reach your site, that your bot protection is not silently blocking them, and that your important pages contain real content without JavaScript running.

Generative Engine Optimization The broadest of the three in common use. It refers to being visible in systems that generate an answer rather than returning a list, which covers assistants, AI summaries on results pages and any interface that synthesises rather than links.