There's a lot of snake oil in AI search measurement. In this episode of Phasewheel In Brief, host Kiley Hylton and Phasewheel Founder and CTO Caitlin Morin cover which AI metrics are worth tracking, which ones to ignore and how to report AI search performance in a way your CEO will actually understand. What you'll learn: • The three AI metrics Phasewheel tracks for clients: Share of Voice, Citation Share and Topic Visibility • Why a small Share of Voice number can still mean you're winning • Why asking ChatGPT the same question every week isn't real measurement • Why raw mention counts can be misleading • Why clicks are a directional signal, not your north star • What crawlability is and why it's now its own reporting pillar • How fast you should expect to see citation movement after shipping changes •The one metric to bring to your CEO every month Resources Mentioned • Phasewheel crawlability checker: see which AI bots are blocked from your site, beyond what your robots.txt shows: https://crawlcheck.phasewheel.com/ • Third-party AI visibility tracking tool: DM Caitlin Morin on LinkedIn for the name • Google Analytics (GA4): for AI referral traffic • Google Search Console Q&A Q: What's the AI-search equivalent of ranking number one in Google? There isn't a direct one. The closest thing is being cited consistently, for the right prompts, in the way you want to show up. Some tracking tools assign a ranking position, but it's a vanity metric and we don't make decisions based on it. Share of Voice isn't a perfect stand-in either, since you could rank first in Google without a huge share of the conversation. Look at the whole ecosystem instead of chasing one position. Chasing rank alone led plenty of brands down a bad path in the old SEO days. Q: How long after we ship changes before we see citation movement? Within weeks, but anyone who gives you an exact number, or guarantees the top spot in two weeks, isn't being honest. Speed depends on the model and on your baseline. Perplexity moves fastest because it's the least dependent on training data. ChatGPT, Claude and Gemini decide prompt by prompt whether to search, so they tend to pick things up faster than models that rely entirely on training. Your first check is crawlability: are AI bots actually hitting the new page? If not, look for problems like an accidental noindex tag. If you see no meaningful movement in Share of Voice, mentions or crawl activity within three to four weeks, it's time to review and adjust. Over about three months, you should see significant change in your citation rate. A steady publishing "heartbeat" helps AI engines treat your site as a reliable source. Q: What is the one metric we should be reporting to the CEO every month? Honestly, it's never just one metric. Every number needs context. But if you have to pick one, report Citation Share on revenue-critical prompts. It answers the question leadership cares about: are we in the room when a buyer is about to make a purchase decision? It won't tie to revenue one-to-one the way ROAS does in paid search, since that tracking doesn't exist yet. You should still be able to connect it to revenue, and if a metric can't tie back to revenue, it shouldn't lead your report.