How Perplexity, ChatGPT And Gemini Pick Their Sources
A Reasonable Sequence Fix rendering first, since content a machine cannot see is the only total failure in the list. Then work through your commercially important pages one at a time, moving the direct answer to the top and replacing the vaguest paragraph with concrete figures.
What the Change Actually Is For a qualifying query, Google composes a short answer from several sources and displays it above the conventional results, with links to the pages it drew on. The user can read the answer, follow a source, or scroll past.
The important detail is that this does not replace the results page, it displaces it. Your listing is still there. It is simply lower down the screen and competing with an answer that has already satisfied a portion of the audience.
The honest position is that attribution in this channel is harder than in any other you are currently running, and the field has responded to that difficulty mostly by inventing numbers. Confident figures circulate widely, and a surprising share of them trace back to a vendor's own sample or to a study far smaller than the claim implies.
Each individual inconsistency looks trivial. Collectively they prevent a set of mentions from resolving to one confident record, and the symptom is a brand that gets described vaguely or hedged around rather than recommended.
How to Test Rather Than Trust Everything above is a starting hypothesis. Run twenty prompts in your own category across all three, from signed out sessions, recording the mode and the date, and count the cited domains for each.
Equally, do not restructure an entire site on the assumption that all informational content is now worthless. The impact is concentrated in a specific type of query. Measure which of your pages carry the signature before reacting, because the temptation to attribute every traffic decline to this is strong and often wrong.
The test that keeps this honest is simple. Show the rewritten page to somebody who buys from you and ask whether it is clearer. If the answer is no, no amount of extraction friendliness makes it a good page. ai seo agency
The condition is that the output has to be yours to keep and act on elsewhere, including the prompt set. An audit that only makes sense inside that agency's retainer is a sales document with a price attached.
Test it rather than assuming. Load your key pages with JavaScript disabled and see what survives. If the product specifications, pricing, service areas and contact details vanish, that is what a machine reads.
You also cannot cleanly attribute a purchase to a recommendation the buyer received three weeks earlier in a conversation you never saw. That influence is real, it is often the main value of the channel, and it will not appear in any report you own.
What Kind of Content Lost the Most The pages that suffered most are the ones whose entire value was a fact a summary can state. Definition posts, unit conversions, simple how-to answers, opening hours, basic specifications and the introductory paragraph content that many sites published purely to capture a query.
This is where the two disciplines meet. Work done to make pages quotable for assistants tends to help here as well, because the underlying problem is the same. A model is looking for a passage it can lift and attribute.
The second is content behind interaction. Accordions, tabs and modals are good interface patterns and their content is sometimes absent from the initial response. Check whether yours is present in the HTML even when collapsed, which is usually a configuration question rather than a design one.
The honest framing first: nobody outside these organisations knows the selection logic, and the systems change without announcement. What follows is drawn from observable behaviour, visible citations and published research, which supports useful generalisations and does not support precision.
What an Entity Is Strip the terminology away and an entity is just a thing the system believes exists: a company, a person, a product, a place. The system accumulates facts about it and attaches them to a single record.
One overlooked cost is your own time. Every engagement in this field needs somebody inside the business to confirm figures, approve crawler changes and answer factual questions, and a plan that assumes this is free will stall. Budget a few hours a month explicitly and name the person, because the alternative is an agency waiting on answers and billing for a month in which little shipped.
Set up a simple internal rule to stop the problem returning. One document holding the canonical name, address, founding year, leadership and product names, referenced by anyone creating a new profile, listing or account. Fragmentation is almost never a single decision, it is dozens of small ones made by people who had no way of knowing what the canonical version was.
How Identity Fragments Fragmentation is rarely deliberate. It accumulates through ordinary business activity: a rebrand that was applied to the website but not to old directory listings, a legal name that differs from the trading name, an office move recorded in some places and not others, a founder's profile that lists a different company spelling.