Concepts

Read this once. Every number in the report leans on the ideas here, and without them the charts are just colored rings. It’s a short read.

We measure three things, not one score

A single “visibility score” is easy to put on a dashboard and useless for deciding what to do. Three brands can share the same score for opposite reasons. So Palmata measures three separate things, and Brand Perception shows each as its own ring.

Are you mentioned? Did the answer engine name your brand in its answer at all. This is the floor. If you’re not mentioned, nothing else happens.

Are you cited as a source? Did one of your own pages get pulled in as evidence for the answer. Being mentioned means the engine knows you exist. Being cited means your own words shaped what it said.

Is the take positive? When the engine does talk about you, does it help your case or hurt it. A mention with a lukewarm verdict is a soft no, and it counts against you here.

A brand can be mentioned constantly and cited nowhere, which means the story about you is being told entirely by other people. Or it can be cited often but described flatly, which means your pages are getting read and still not making the case. Keeping the three apart is what tells you which problem you actually have.

Cited, considered, and not considered

When the engine answers a question, it reads a set of pages and then writes from some of them. Three things can happen to any page, including yours.

Cited — the engine read the page and used it in the answer. This is the goal. Your page is in the room and doing work.

Considered — the engine read the page but left it out of the answer. It was in the room and got passed over. This is the most actionable state in the whole report. The engine already found your page relevant enough to pull; something about the page lost the spot. That’s a fixable problem with a known page.

Not considered — the engine never pulled the page at all. For a question where you’d expect to show up, you weren’t in the running.

The gap between considered and cited is where most of the quick wins live. You don’t have to earn the engine’s attention from scratch — you already have it. You have to give it a reason to use you.

The four signal states

Every buyer question gets one of four labels, based on whether your brand showed up in the engine’s sources and whether it showed up in the answer. These are the exact words you’ll see in the app.

In Sources And Answer — your page was retrieved and your brand was clearly used in the answer. This is the win. The engine had your evidence and leaned on it. Defend it: keep the page current so a competitor doesn’t take the spot.

In Answer, Not In Sources — your brand was mentioned, but none of your pages were retrieved as evidence. The engine is talking about you from what other people wrote. You’re at the mercy of how they frame you. The work here is to publish the page the engine should have reached for, so your own account is on the table.

In Sources, Weak In Answer — one of your pages was retrieved, but the answer barely used it or didn’t mention you. This is the considered-not-cited case. The page is close. It got read and skipped, usually because it doesn’t answer the question directly enough or buries the part that matters. Restructuring an existing page is faster than writing a new one.

Not In Sources Or Answer — you weren’t in the engine’s sources and weren’t in its answer. For this question, you’re invisible. If it’s a question your buyers ask, this is category entry: there’s nothing to fix, only something to build.

The four states map cleanly onto four kinds of work — defend, publish, restructure, build — which is why the Actions view is organized the way it is.

Why we research your market first

A naive audit asks an answer engine a handful of questions a marketer would write and counts the mentions. The problem is that buyers don’t talk like marketers. They use different words, ask narrower questions, and care about things your positioning never mentions.

So before we ask anything, we research. We learn how and where you compete, who your buyers are, and how your customers and prospects actually describe your products and the problems they’re solving. We use that to prompt the answer engine the way those buyers would — to simulate the real interactions a potential customer might have when they’re trying to decide. We’re looking for the situations that expose your opportunities and that show how you and your competitors tend to come up.

That’s why the questions in your report read like things a real buyer would type, not a keyword list. The research is the difference between “how often does our name appear” and “here is exactly where a buyer would meet us and what they’d hear.”

A mention is not a recommendation

It’s tempting to count mentions and call it visibility. But being in the answer with a flat or skeptical verdict can be worse than not being there. The report tracks whether each mention helps or hurts your case, not just whether it happened. When you read Brand Perception, the AI Perception ring is the one that tells you whether the mentions you’re getting are working for you.