There are two ways an AI can use your content. In the first, it shows where the information came from — a link, a site name, a clickable citation. In the second, it absorbs what you wrote, blends it with ten other sources, and returns a smooth answer where you appear nowhere, even though you contributed.
Both scenarios shape the asker's decision. But only one gives you credit. And the major models behave very differently here.
Source Transparency Isn't a Technical Detail
For marketers, source attribution is what separates "being influential" from "being influential and getting recognized for it." When an AI cites your brand with a link, three things happen: the user can click, you get a reinforced authority signal, and there's an auditable trail that you took part in that answer.
When it uses your content without attribution, you still shape the response — your information is there, molding what the model says. But nobody knows. No click, no credit, no way to measure it the traditional way.
That's why each model's citation behavior changes your strategy, not just your ego.
The Three Postures
There are roughly three stances in play.
The ones that show off sources. AI search interfaces — like Perplexity and the search modes of ChatGPT and Gemini — are designed to display where information came from. Numbered links, side citations, explicit attribution. Here, getting cited tends to be visible and, in some cases, clickable.
The ones that synthesize silently. Models in pure conversation mode, without active search, rarely say where what they're saying came from. They answer from training, and training doesn't come with footnotes. Your brand can be shaping the answer with no visible mention at all.
The hybrids. Most assistants today switch between the two depending on the question, the settings, and whether search was triggered. The same tool can give you a link on one question and ignore you as a source on the next.
Why Models Differ So Much
The difference isn't a whim. Models that retrieve in real time have links to give — they're looking at real pages in that moment, so they can point to them. Models that answer from memory have no specific URL to cite, because the knowledge was distilled into patterns during training. There's no origin page to link.
Add product choices on top. Some companies bet that showing sources increases user trust. Others prioritize a clean, direct answer and treat the link as noise. Neither is "wrong" — but each creates a different world for your brand.
The Trap of Measuring Only What Links
Here's the most common strategic mistake: looking only at answers that cite sources with links and concluding that's the only place your presence exists.
Unattributed influence is real and, in many contexts, more frequent than linked influence. If you only measure what shows up as clickable, you're seeing the visible tip and ignoring most of the iceberg — all the answers where your content is shaping what the AI says without ever naming you.
That's why measuring AI presence can't boil down to "how many clicks came from AI." Clicks only capture the attributed scenario. The fuller question is: when someone asks about my industry, does my brand appear in the answer — with or without a link — and what does it say about me?
What to Do With This Difference
The good news is that both paths respond to the same fundamentals. Clear, authoritative content is more likely to be linked by search models and absorbed by memory models. You don't need two strategies; you need one strong strategy and measurement that sees both sides.
What changes is the expectation. In interfaces that show off sources, you can chase the clickable citation and some traffic with it. In interfaces that synthesize silently, the goal is subtler: making sure that when the AI talks about your topic, it speaks from what you published — and gets the facts right.
Measuring both — who links you and who uses you without saying so — is what gives you the real picture of your influence. Genoma monitors your brand's mention inside the answers, not just outbound links, precisely so you don't confuse "invisible" with "absent." Often you're there. You're just not showing up.