Skip to main content
The Herald asks the engines on two schedules. A daily update collects and reads answers every day. A weekly report asks once more, pools the week’s answers and sets your figures.

​
Daily update

Once a day, at an hour set for each brand, Heralded asks every tracked prompt of the engines due that day: ChatGPT, Gemini, Google AI Overviews and Google AI Mode every day, and Perplexity and Microsoft Copilot three days after report day. It reads each answer the same way the weekly report does: whether you are named and on which rung, how it describes you, which brands it names and which sources it cites. A daily update doesn’t change your headline figures. It shows up as movement between reports, such as a prompt where you started or stopped being named, and it feeds the next weekly report. Daily updates don’t post a report message in the Herald’s conversation. A prompt you track on the Prompts page doesn’t wait for the next daily update: Heralded reads it within minutes on every engine its segment reads, and its answers show in the prompt’s pane the next time you open Prompts. That read records no movement, and the weekly report doesn’t pool it. The daily updates read the prompt as usual from then on.

​
Weekly report

On report day the weekly run also does the daily update. Heralded asks every engine every prompt, Claude included if you have it, records movement and opens work for losses that have held for three updates or reports in a row. Your daily series has a reading on all seven days from the engines that answer every day. It then pools the week’s daily answers with report day’s own into the same cells. You receive the weekly report email on report day, without a separate Daily changes email.
  • Every figure that reads an answer counts the whole week the same way: Mentioned, Recommended, the Heralded Score, Perception, share of voice and prompts won all use every answer of the week, so a prompt can read “Named 5 of 7 days”. Share of voice counts report day’s answers alone when the daily answers lack the reading it needs, and says so.
  • Each prompt’s state on the Prompts page and source counts come from report day’s answers alone.
New reports leave out daily answers collected before a change to how answers are read. Issued reports keep the answers and reading method they were issued with. Home’s status line summarizes the brand’s standing, its measured change and the actions first found in that report. It stays the same until the next weekly report. If the written line is unavailable, Home shows a plain summary of those facts. Daily updates and work you complete do not rewrite it. You need brand access to read it; shared report links do not include it. Each weekly report posts one message in the Herald’s conversation. It holds that report’s score, Mentioned, Recommended, the number of your pages cited, Home’s changes, new task suggestions, pages cited for the first time, top tasks with Start, and setup steps still open. The Start list shows the plan’s weekly task allowance and how many tasks had started when the report was built. Figures follow the same change rules as Home. Older messages keep their report’s figures. Everyone with brand access can read the message, even when nobody receives email and no Slack channel is connected. Rerunning the report job doesn’t add another message. When a report lands, Home updates and reads like “Weekly report 22 Sep · 3 daily updates since · Next: …”. Workspace members whose role can see brand data get it by email, with a link to the Herald’s report message. It arrives at 07:00 in the time zone set on their Account page, or when the report is ready if that is later. Without a time zone, it goes out as soon as the report is ready. A brand’s first report is emailed as soon as it is ready, whatever the time zone. Notifications turns the email off. A paused brand, a brand being deleted, or a workspace whose subscription has ended gets neither this email nor the Daily changes email. People with access to a single brand don’t get it. Slack gets the headline if you connected it. Home, the agent panel, the rail and the API name the next daily update and weekly report only while they will run. A paused brand, or one whose subscription has ended, shows none.

​
Your schedule

The schedule is under Reads on your brand’s Brand settings page:
  • Daily update shows the hour it runs. Heralded spreads brands across the day, so the hour differs from brand to brand.
  • Weekly report day starts as the weekday you confirmed your prompts, counted in UTC, or Monday if that was a weekend day. You can move it to any day from Monday to Friday. Home names the same UTC day.
  • Pause stops both until you resume. Figures keep their last values, and the brand’s prompts free their slots until you resume.

​
The first report

The first report starts as soon as you confirm a new brand’s prompts, and your first weekly report comes seven days later, or on the Monday after if that day is a weekend. Get started and Your first weekly report walk through each. Your first report is your baseline. The Herald posts four messages in order:
  1. Your baseline, with the score and band, Mentioned, Understood, Recommended and Praised as approximate values, your rank against tracked rivals, two strongest prompts and up to two gaps where a rival is recommended and you aren’t. Your top three startable tasks carry their owner, a short reason and Start, beside the plan’s weekly task count. The message also gives the next daily update, the first comparison date and setup steps still open.
  2. How to read this, with a line for each figure and links to its method. Ranges explain why the Herald only calls a move when it is real. This report is your starting line, so it has no change to call.
  3. How we work, introducing the Writer, Publicist and Engineer. You can ask questions in the thread, or use New chat for private questions. A change you ask for goes through confirmation.
  4. Close this chat to see your overview, linking to Home.
The baseline’s opening uses the same guarded line as Home. The rest comes from the report’s facts. Your first landing after it arrives opens the Herald’s chat at the baseline; each person sees it open once. The sequence keeps its original figures, tasks and dates. Replies keep the report’s prompts and gaps, including the rival named in each gap. Rerunning the report job adds no messages.