If you’re still measuring organic search the same way you did three years ago, there’s a good chance you’re missing a big part of what actually impacts your customers.
Traditional SEO has given us something fairly predictable: a ranked list of links, click data, and a measurement model centered almost entirely on Google. AI research is a different beast. ChatGPT, Perplexity, Gemini, Copilot, Google AI Overview; These platforms aggregate answers, change output between sessions, and can influence a purchase decision without ever generating a click.
A potential client can ask an AI assistant which PR tool is best for their agency, read the answer, form a clear preference, and then type your competitor’s name directly into Google. The AI platform led this decision. Your analytics don’t capture this.
So the old model of measuring what can be measured and calling it accomplished is no longer sufficient in itself.
The good news: There’s a smarter way to deal with this.
Three layers actually tell you what’s going on
SEO consultant Aleida Solis has developed a practical framework for measuring AI search that is really useful. It divides performance into three connected layers, each answering a different question.
Layer 1: Attendance – Are you actually present?
This is the starting point. Before you can fix anything, you need to know if your brand appears in important AI answers, and how it is represented when it does.
Measuring attendance isn’t just about whether or not your name appears. It covers:
- The platforms they appear on
- Whether you are recommended or just listed
- Whether those signals include a clickable link
- Whether you win when AI compares you to competitors
- Whether you are described accurately
If an AI assistant incorrectly describes your business to thousands of people every week, that’s a business problem, even if your rating looks good.
Layer 2: Readiness – Are you structurally ready to be found?
Preparedness is the diagnostic layer. It explains why your vision looks this way.
If you’re showing up but not getting links, this may indicate how your content is structured. If your recommendation rate is low, it may be due to differentiation or a trust issue. If you are being described incorrectly, entity consistency across the web is likely to blame.
The goal of preparedness work is to stop making overall improvement on a problem that has not been properly diagnosed. Different vision gaps have different root causes. This layer helps you find the right lever.
Layer 3: Business Impact – Does any of this actually work?
This is where the analogy becomes honest. AI referral traffic is a useful signal, but it represents the floor, not the ceiling, of AI’s contribution to your business.
A large percentage of AI-influenced conversions never show up as AI referral sessions. Users see your brand in the AI answer, don’t click on it, then search for you directly or type in your URL. This conversion is attributed to organic or direct branded traffic. The impact of AI is invisible unless you measure it.
Business impact measurement combines observed data (actual AI referral sessions), agent signals (branded search trends, direct traffic increases, survey data), and model estimates to give a more complete picture. None of these individually tell the whole story. Together, it brings you closer to the truth.
Why is this more important than another control panel?
The value of this three-layered approach lies not in the metrics themselves, but in how they are communicated.
Attendance tells you what’s happening. Preparedness tells you why. Business Impact tells you if something is commercially important. When you play them in sequence, each one delivers a premise to the next. This turns a reporting exercise into something you can actually act on.
We’ll go deeper into each layer over the next few posts in this series. But if you want to start now, the first step is simple: stop seeing AI as binary. Whether your brand shows up or not is not the question. How it appears, where, and with what effect, that is where the real work is.
In the next article, we’ll look at how to build a quick library that actually represents how your customers use AI and why most companies get it wrong.
