Skip to main content
Your AIS Score will move between scans — sometimes by a little, sometimes by a lot. This page explains the mechanics behind that movement, so you can tell a meaningful signal from expected noise.

​
Score bands

Every AIS Score maps to one of five bands:

Excellent — 80 to 100

Consistently surfaced across AI answers.

Good — 60 to 79

Frequently mentioned across AI platforms.

Fair — 40 to 59

Visible for some prompts — clear room to grow.

Poor — 20 to 39

Largely invisible — most prompts don’t surface you yet.
Below 20 is Critical — effectively absent from AI answers. The band edges (80/60/40/20) are fixed thresholds on the composite score; a band change always reflects a real move across one of those lines, not a display artefact.

​
The credibility ramp for new brands

Two of the four signals behind your AIS Score — Position and Sentiment — can only be measured on runs where your brand is actually mentioned. Early on, that’s a small sample, and a small sample is unreliable: one mention at rank 1 says almost nothing about how you’ll rank on the next hundred runs. To keep the score honest, both signals are credibility-shrunk toward a neutral midpoint of 50 in proportion to how much evidence exists, and ramped in over your first three mentions:
The practical effect: a brand with very few mentions will see its Position and Sentiment scores swing more with each new scan than a mature brand will. A single new mention at rank 1 might move a 2-mention average a long way; the same mention barely nudges a 60-mention average. This is expected and it’s the honest behaviour — the alternative would be letting one lucky early mention lock in a misleadingly high score before there’s any real evidence behind it. As your mention count grows past roughly 8, the shrink pulls less and the score starts tracking your raw when-mentioned performance more directly. The breakdown drawer always shows both the raw value and the credited value, along with your current mention count, so you can see exactly where in the ramp you are.
If your score looks volatile in the first few weeks of tracking a brand, check the mention count in the breakdown drawer before assuming something’s wrong. A handful of mentions moving the score by several points is the ramp working as designed, not instability in the underlying methodology.

​
Awaiting data vs a real zero

Surfais distinguishes carefully between two states that can look similar but mean opposite things:
  • Awaiting data — you haven’t run a scan yet. There’s no AIS Score to show, so the dashboard displays an empty gauge labelled “Awaiting first scan” rather than a numeric 0. No band, no delta, no breakdown.
  • A real zero — you’ve been scanned, and your brand was not mentioned in any of the tracked runs. This is a genuine, measured Critical-range result, but Surfais still shows it gently as “No visibility yet” rather than an alarming red score, because a true zero and a score that’s merely low in the Critical band are different situations worth reading differently.
Never read “no score shown” as “your visibility is zero.” Run your first scan before drawing any conclusion — see Manual scans or Scan schedule to get your first data in.

​
Common causes of movement

Once you have an established score, a handful of routine actions can shift it in ways that look like a visibility change but are really a measurement change:
  • Adding prompts. Every signal in the AIS Score is a rate or a ratio over your total scanned runs. Adding prompts changes that denominator — your presence rate, platform breadth, and Share of Voice field can all shift even if nothing about your actual AI visibility changed, simply because you’re now measuring a wider (or different) set of questions. See Prompts.
  • Platform answer variance between runs. AI platforms don’t always answer the same prompt identically twice — model updates, live web results, and inherent non-determinism mean a re-run of the same prompt can surface a different set of brands or a different framing. A small amount of run-to-run variance is expected and isn’t a sign of a broken measurement.
  • Competitor set changes. Adding or removing a tracked competitor changes the Share of Voice field directly. Since the field also includes every other brand the AI names (not just tracked competitors), your Share of Voice can move even without any change to your own competitor list, if the unattributed field shifts.
  • New countries. Adding a country under Countries adds new prompt × platform × country runs to your total, which changes every rate-based signal’s denominator in the same way adding prompts does.
None of these are problems to fix — they’re the expected consequence of measuring a real, changing conversation. Before treating a score movement as a signal about your actual visibility, check whether your tracked scope (prompts, competitors, countries) changed in the same period.

AIS Score

The headline number these bands and this ramp apply to.

Position

The credibility shrink mechanics in full detail.

How scans work

What actually happens during a scan run.