We do not say a fix caused your AI-visibility score to rise — here is the honest math we show instead

You mark a fix "done" in your dashboard, come back a week later, and your score is up 6 points. Every instinct says the fix worked — and every tool in this category would happily tell you that, because "your fix worked!" is a satisfying thing to show a paying subscriber. We won't say that, and this post is the honest explanation of why, plus exactly what we show you instead: the real, unvarnished math, including the times it comes back at zero or negative.

What actually happens when you click "done"

Every fix on your dashboard is tied to one real thing you're losing today — a specific buyer question where no AI engine named your brand, pulled directly from your latest scan, never a generic "publish more content" suggestion disconnected from evidence. When you move one from "to do" to "done," we do exactly one thing at that instant: we record your visibility score as it stands right then, and we stamp the time. Nothing else happens. No score is edited. No competitor's answer is deleted and rerun. We simply write down where you stood the moment you told us this was finished, so we have an honest before-and-after pair to compare against your next scan.

That capture only happens once. If you re-open a fix and mark it done again later, we keep the original number — re-confirming "done" never quietly resets your baseline to a more flattering starting point. And if you move a fix back to "doing" or "to do," we clear the capture entirely, because a fix that isn't finished has no lift to measure yet.

The number we show you, and the one we won't

Once a fix is done and you rescan, we show you the plain arithmetic: your latest score, minus the score you had the moment you marked it done. If that number is positive, we show "+6 points since you marked this done." If it's flat, we say so: "no change yet." If it's negative, we show the negative number, exactly as measured.

What we don't do is attach the word "because." We'll never render "you gained 6 points because of this fix." The honest label is "since you marked this done" — a statement about timing, not a claim about cause — and that distinction is deliberate, not a legal hedge.

Here's why it has to work that way. A visibility score is a live read on four separate, non-deterministic language models, each one probabilistically generating an answer to the same buyer question, re-run on a schedule. Between the moment you mark a fix done and your next scan, dozens of things can move that number that have nothing to do with your fix: the underlying model gets a quiet update, a competitor publishes their own comparison page, one of your other in-flight fixes lands in the same window, or an engine simply answers the same question differently this week than it did last week, because that's what generative answers do. With a single before-and-after pair and no control group, there is no honest way to isolate "this specific fix moved the number by this specific amount." Any tool that shows you that precise a claim is showing you a number it can't actually back up.

Why zero and negative deltas matter as much as the positive ones

This is the part that actually separates an honest tracker from a marketing dashboard: we render the negative case exactly as plainly as the positive one. If your score has dropped since you marked a fix done, you'll see the real negative number, not a hidden metric or a quiet reset. If nothing's moved, you'll see "no change yet" instead of the row just disappearing.

That matters because a tool that only ever shows gains isn't measuring anything — it's just reassuring you, and reassurance isn't the job. The honest version is slower and less flattering, but it's the only version where a positive number actually means something when you finally see one, because you know the same tracker would have told you if it hadn't worked.

Grounded in the real evidence you lost, not generic advice

Each fix also carries the actual evidence behind it: the literal buyer question your brand lost, the rival AI named instead, and — where the scan surfaced them — the real domains AI engines leaned on for that question. From that evidence we generate something you can act on immediately: a drafted FAQ answer, a comparison-page outline seeded with the actual competitor who beat you, or a target list of the exact sites to try to get listed on. We only ever build that artifact from data your scan genuinely captured. If a fix has no groundable evidence behind it yet, we don't invent one to fill the space — you simply won't see an artifact for that row.

What to actually do with the number

Treat a single "since you marked this done" delta as one directional data point, not a verdict. The way to use it honestly is the same way you'd read one week of traffic data: don't decide anything off a single reading. Keep marking fixes done as you finish them, keep rescanning on a normal cadence, and watch the pattern across several fixes over several weeks. A fix that consistently precedes a positive delta, repeated across multiple brands or multiple questions, is a much stronger signal than any single instance — and that's a pattern you build by tracking honestly over time, not one a single inflated number can give you upfront.

If you want the full picture of how the underlying score is built in the first place, the score formula itself is fully documented — the same "show the real math, including the parts that don't look good" rule that shapes the fix tracker also shapes the score it's tracking. It's the same commitment behind why we show "Not measured" instead of a fabricated zero when an engine wasn't actually reached.

See your own real fixes

The only way to know what's actually moving your score is to have real fixes tied to real evidence in the first place. Run the free scan, see exactly which buyer questions you're losing today, and — once you're tracking a brand — watch the honest, correlational math build up over your next few rescans. No invented lift, no hidden losses, and no number we can't show you the arithmetic behind.