All releases
v1.15.0

Methodology, Compare & Prompt Import

The dashboard now counts the same answers four ways with confidence intervals, a new Methodology page explains each formula and sets another tool’s export beside ours computed the same way, and My Prompts can import a prompt list from a CSV after a dry run.

Features

  • Methodology Page — A new Methodology page, linked from the dashboard and from Prompt Tracking, explains how each of the four formulas is counted, shows how many answers went into the numbers and why any were left out, and breaks every formula down per model with its own interval.
  • Compare With Another Tool's Export — Upload another tool's raw-answer export on the Methodology page, tell us which columns hold the prompt, engine, date and result, match its engine names to the models we track, and see its mention or citation rate next to ours, computed the same way over the same window. The gap between the two is split into the part a different mix of prompts and models explains and the residual that mix does not, with a per-prompt breakdown and a list of what the comparison left out. Exports with one row per citation rather than one per answer are turned away with an explanation of what to upload instead. A comparison is read-only — nothing is saved — and Compare is not available in the portfolio view (no client workspace selected) or to viewer members.
  • Import Prompts From a CSV — My Prompts has an Import CSV button. Pick a CSV, TSV or TXT file and the column that holds your prompts, and a dry run checks every row first — flagging prompts you already track, duplicates within the file, invalid rows, and rows beyond your plan's prompt limit — before anything is created.

Improvements

  • The Same Answers, Counted Four Ways — The Visibility dashboard now shows the same answers counted four ways (mention rate, citation rate, position-weighted visibility and share of mentions), each with its 95% confidence interval. A formula the recorded data cannot support shows a dash and the reason instead of a zero.
  • Current Gemini, GLM and Qwen Models — Prompts tracked on Gemini now run on Gemini 3.8 Flash in place of 3.7 Flash, GLM prompts run on GLM 5.3 Flash, and Qwen prompts moved to the September snapshot of Qwen 3.8 Max. My Prompts shows the new model names.

Fixes

  • Publishing Waits for a Yes — When the Omni Agent asks before publishing to your site, a reply that is only a refusal or a pause, such as "no", "not yet", "wait", "hold off", "stop" or "cancel", is never taken as approval.