Why the baseline matters so much in AI search.
In traditional SEO, rankings are visible and tools track them well. In AI search, visibility is far harder to see: answers vary by phrasing, by user, and over time, and there is no ranking report. That opacity is exactly why a baseline matters more. Without one, any provider can claim improvement, and you have no way to check. Setting the baseline before the work starts is what makes AI SEO accountable rather than a matter of faith.
The four things to measure before you start.
Citation presence. Whether your brand appears in answers from ChatGPT, Perplexity, Gemini, and Google AI Overviews, across a defined set of prompts that reflect real buyer questions. This is the core metric. Competitive citation. Which competitors are cited on those same prompts. AI citation is comparative, so knowing who currently owns the answer tells you what you are displacing. AI referral traffic. What traffic you already receive from AI engines, which is visible in analytics and often overlooked. It establishes the commercial baseline. Entity and schema coverage. How clearly your brand, services, and expertise are defined and structured today, since this is what the work will change.
The metrics that will mislead you.
A single lucky citation. AI answers vary, so one test showing your brand cited proves nothing. Presence must be measured across a prompt set, repeatedly. Traditional rankings as a proxy. Ranking well does not mean being cited. Using rankings to judge AI visibility hides the gap you are trying to close. Vanity prompts. Testing prompts nobody actually asks, especially your own brand name, produces flattering results and no insight. Baseline on the prompts buyers really use. Raw traffic. AI referral traffic is useful but small for most brands today. Judging AI SEO purely on traffic volume, this early, will undervalue work that is genuinely building citation presence.