Answer Receipts, Unbranded Questions & Team Controls

Features

• Read the Answer Behind Every Run — Expand a run in a prompt's Runs tab to read the AI answer its numbers were counted from, with your brand and tracked competitors highlighted and the sources it cited a click away. Gemini answers written with Google Search are not displayed for now; they still count in every number. The Omni Agent can now search the full text of these stored answers, not just their opening lines.
• Branded and Unbranded Questions — A prompt that names your brand, an alias or your domain is tagged Branded in My Prompts. Home and the Methodology page now lead with unbranded questions, where an AI has to bring you up on its own, with the questions that name you on a second line. Prompt Metrics has an All / Unbranded / Branded switch, and a new client workspace gets two unbranded discovery prompts once its market is set.
• Manage Your Team — Organization admins can promote a member to admin or remove a member from Members & Seats, each after one confirmation, and resend a pending invitation in one click. An organization always keeps at least one admin, a removed member loses access at once, and a resent invitation goes out with a fresh link.
• AI-Referred Traffic on Home — With Google Analytics 4 connected, Home compares the last 30 days of visits referred by AI assistants with the 30 before: sessions and their share of all traffic, key events, revenue when your property reports it, and the top assistant. Reports gain a matching AI-Referred Traffic (GA4) section. A figure GA4 has not reported shows a dash rather than zero, a failed or stale sync is flagged, and the card is not shown in agency client workspaces.
• Methodology Changes on Your Charts — The Visibility trend on Home and the Performance trend in Prompt Tracking mark each date we changed how a number is measured, with a tooltip saying what changed, which numbers it affects and who it applies to. The layer is on by default and can be switched off per chart.
• Client Contacts on Scheduled Reports — In a client workspace, organization admins can add up to five client contacts to a scheduled report, and each contact gets the PDF in their own email with a one-click unsubscribe link. Every schedule shows who it reaches.

Improvements

• More Ways to Read Methodology — The Methodology page adds a model-balanced rate that counts each engine with 10 or more answers once, mention rate split into how often a tracked brand is named and how often it is you when one is, and your share of cited URLs with your rank against tracked competitors. Citation rate is now called Citation coverage, Position when named (how early you appear in the answers that name you) replaces position-weighted visibility, and the page shows its window's dates and the runs that failed, are pending or were skipped.
• No AI Answer Is Not a Zero — When Google shows no AI Overview or AI Mode answer to a question, the check is counted as "no AI answer shown" and left out of mention, citation and position figures instead of being averaged in as zero. Methodology shows how often each Google surface answered.
• Overall Score Across Your Clients — Command Center cards, its average KPI and the portfolio report, on screen and in the PDF, now show each client's Overall score and BPI, the numbers that client sees on Home, in place of a raw analysis average. A client with nothing to score shows a dash and the reason.
• Every Scheduled Check Asks Again — A scheduled check is never answered from a stored copy of an earlier answer; each one asks the engine afresh, so the "Check live every run" option is gone.
• Skipped Engines Are Listed — When an engine is skipped or slowed for a prompt, because of Grok's weekly limit or monthly budget, an add-on that is not active, or your plan's limit, the Runs tab lists it with the reason, and Execute now says so in its confirmation.
• Honest Run Badges and Report Counts — A failed or pending run now reads "Failed, not counted" or "Pending, not counted", and the position chip says how far into the answer you are first named. Report PDFs count completed answers, show each cited source as "in k of n answers", list unbranded questions first and date the cover with the report window. A Measurement baseline preset is the new default, with the Executive Summary optional.
• Agent Journeys — Journeys now default to the ChatGPT model, and a Grok journey runs only within the workspace's Grok allowance (the Grok add-on under the new plans). A run that cannot get a browser is relaunched up to three times and refunded if it never starts, and a scheduled journey no longer starts twice.
• Failed Site Analyses Retry Themselves — A failed site analysis is retried after 15 minutes and again after 6 hours before returning to its regular schedule, with one alert if it still fails. A site that refuses our crawler through robots.txt, a bot challenge or a 401 or 403 is not retried, and you are told straight away.
• Full-Width Settings — Every settings page now uses the full width of the window, as Apps does. On wide screens related sections sit side by side, report branding shows its cover preview beside the form, and the client list is a table with slug, domain and date added.
• Faster Dashboard Loads — Your browser now remembers our API's cross-origin check for up to two hours instead of five seconds, so most dashboard requests skip an extra round trip.
• Privacy, Terms and Subprocessors — The Privacy Policy and Terms were rewritten to match how the product works, and a new Subprocessors page lists each provider that processes customer data, with its purpose, the data it handles and its location.
• Omni Sans — The app and the site are now set in Omni Sans, our own typeface.
• Turn On Newly Included Engines — When your plan gains an engine, such as Perplexity on moving from Starter to Pro, My Prompts offers to turn it on for your existing prompts in one click instead of leaving them on their old engines.
• Engine Names, Not Vendor Names — Answer views, the Models table, the run-now confirmation and the engine lists name the engine you know (ChatGPT, Gemini, Claude, Grok, Qwen, Kimi, GLM) instead of the company behind it.
• Agency Checkout Total — Agency checkout shows the monthly total for the number of client workspaces you choose, worked out with the same graduated pricing as the pricing page.

Fixes

• Overall Score Counts Only Your Site — The Website part of Overall averaged every site in the workspace, tracked competitors' included; since September 30 it counts only your own. Earlier days keep the old rule, and the change is marked on the trend chart.
• Your Site During a Re-Crawl — While your site was being re-crawled, answers could be measured against a tracked competitor's domain; they are now always measured against your own.
• No False Zeros for New Workspaces — First runs that finished before a new workspace's first crawl were recorded as zero visibility; they are now measured once the crawl completes.
• Card-Heavy Pages Read in Full — Pages built from many article cards, such as many WordPress homepages and blog listings, were read as a handful of words; the page's own text is now picked up.
• Completed Analyses Stay Complete — A slow brand-perception step could mark a finished site analysis as Failed; it now runs on its own, and a slow recommendations step is retried within the analysis's time limit instead of failing the run.
• Work Resumes After Our Updates — A Share of Voice analysis running while we deploy an update now finishes instead of failing, a site analysis interrupted the same way restarts instead of sitting on "indexing", and a nightly Google Analytics sync cut short is caught up within the hour. An open tab reloads cleanly after an update instead of going blank, and a dashboard panel that fails shows "This section failed to load" with Retry while the rest of the page stays.
• ChatGPT Counted Once — ChatGPT answers from its app and from its API now count as one engine in the Models tab, Source Influence, Prompt Metrics and reports instead of appearing twice.
• Joining an Organization — The invitation sign-up form now asks for the username sign-up requires, so new invitees can finish joining. Invited members land in their assigned workspace, or on a clear page saying none is assigned yet, instead of the owner's setup steps.
• Agency Seats Use the Agency Plan — A member without a plan of their own now gets the organization's plan for custom alert rules, Share of Voice, opportunity scans and the Omni Agent in client workspaces, and opens straight into the organization rather than a personal workspace. Settings > Clients > New Client now runs the full guided setup, so new clients get starter prompts and first runs.
• Scheduled Reports Reach the Right People — A client workspace's scheduled report now goes only to people who can open that client, members removed from the organization stop receiving reports, and a failed delivery is retried with the same PDF before admins are alerted. The cover names only the AI platforms in the report window, and the Citations tile shows its figure.
• Agent Work Stays in Its Workspace — Switching workspaces or accounts while the Omni Agent is working no longer carries an opportunity, document or pending chat save into the other workspace, and research in a client workspace with no domain set no longer draws on another workspace's pages.
• Report PDF Layout — A "+N more" citation note no longer overlaps the lines around it, a platform heading stays on the same page as its citations, and "What likely drove it" names engines such as Gemini and ChatGPT.
• Client Creation Is for Admins — Members without admin rights no longer see the new-client buttons in the Command Center and Settings > Clients, which only ever ended in an error for them.
• Clearer Labels — Agency and Enterprise plan badges show their own plan instead of "Premium Plan", the Models table names ChatGPT instead of a fixed model, run answers no longer show raw citation markers, the Command Center's "Active crawls" figure is now "Crawled today", and Improve's empty state explains that it tracks pages you add one by one.