How tracking & scoring works

When a shopper researches a product these days, they often skip the search bar and just ask an AI assistant. superAEO measures whether your store shows up in those answers — and it does it carefully enough that the numbers hold up. Here's exactly how the tracking and scoring work under the hood.

The five engines

superAEO queries the AI assistants buyers actually use for product research. Every tracked prompt is run through all five engines, so you see how each one answers and where you stand across the board.

• ChatGPT — the most widely used assistant, and the one most shoppers reach for first.
• Gemini — Google's assistant, increasingly woven into how people search the web.
• Perplexity — an answer engine built around citing its sources, popular for research.
• Claude — Anthropic's assistant, known for thorough, well-reasoned responses.
• DeepSeek — a fast-growing assistant expanding the set of answers buyers see.

Because every prompt hits all five, a strong showing on one engine never hides a blind spot on another. You always know the full picture.

Daily, automatic runs

Tracking runs on its own every day. You don't have to press a button, remember a schedule, or keep a tab open — superAEO wakes up, runs your prompts across all five engines, and updates your dashboard before you check it.

When you want a fresh reading right now — say, right after publishing a change — you can also trigger an on-demand run from the dashboard. Newer stores are sampled more often while they build up history, then settle into a steady daily cadence once there's enough data to stand on.

Why we sample repeatedly

AI answers are not deterministic. Ask the same assistant the same question twice and you can get two different answers — one that mentions your brand, one that doesn't. If superAEO judged you on a single response, your score would bounce around for reasons that have nothing to do with your store.

So instead of asking once, superAEO asks each prompt multiple times and averages the results. A single lucky answer can't inflate your score, and a single unlucky one can't sink it. Early on, when there's little history to lean on, it gathers extra samples to build a solid baseline you can trust before it starts reporting trends.

Confidence intervals

Every rate superAEO reports — like how often you're mentioned — comes with a 95% confidence interval. Instead of a bare number, you get a range: the band within which the true rate most likely sits. We calculate these using the Wilson method, which behaves well even when a rate is near 0% or 100% or when sample counts are still small.

The width of the interval tells you how settled a number is. A wide range means there's still uncertainty — superAEO is telling you it hasn't seen enough yet to be sure. As more samples accumulate, the range tightens around the true value, and your numbers become more decisive.

How your Visibility Score is built

Your Visibility Score rolls everything up into a single number from 0 to 100. It combines how often you're mentioned and how often you're cited across all of your prompts and all five engines, weighting the signals that matter most for being recommended by an AI assistant.

The point of one number is glanceability — you can tell at a glance whether you're moving up or down week to week, without parsing a wall of stats. But the score never hides the detail underneath. You can drill into any single engine or any individual prompt to see exactly what's driving the headline figure, so the number stays honest and explainable.

Verifying that a change worked

When you complete an action from your improvement plan, superAEO doesn't just take your word that it helped. It re-measures the specific prompts the action was meant to affect, then runs a statistical test — a Z-test — to compare your citation rate before and after the change.

The action is only marked Verified when that test shows the improvement is statistically significant, not just normal run-to-run wobble. This is the same discipline behind the rest of superAEO: a win has to clear the bar of being real before it counts. You spend your effort on changes that genuinely move the needle, and you can prove the lift when you're done.