AI Visibility — Causal Re-Verify Loop

anthropic.com — ship the fix, re-replay all five engines, prove the lift cleared significance
measured · Jun 2026
Eight tools — the full AI-Visibility suite
01

The empty quadrant — what no rival can show

02

The four-step causal loop

03

The honest pre-arm — Wilson intervals, not a point

Per-engine citation rate · 22-question replay

04

The same question, the opposite answer

05

Powered where it's weakest

AI-referral traffic weight vs your rate there

The 77% / 11% traffic weights are a labeled model-estimate — used only to prioritise, never scored.

replays per engine, budgeted by variance

06

The lift simulator — put the cost on screen

drag the two rates — replays/arm recompute live from the methodology formula
55%
70%
163
replays per arm to prove this lift at p<0.05, power 0.80

07

Trigger → deployable fix → latency

08

The honesty ledger — the anti-anecdote

CorrectionWhat it fixesTierSource
09

Volatility gates reportability

Drift bands · CoV → what we're allowed to claim

Current state

10

Citation-slot change feed — the baseline the loop diffs

11

How a fix is proven + the revenue inverse

The revenue inverse — same model, run backwards

DIRECTIONAL

What we did NOT measure

The competitor anecdote wall

named claims, contrasted
ToolTheir published claimType

Bliss Optimizer · AI Visibility · Causal Re-Verify Loop