Support data has a short shelf life. A content gap you learn about this week is one you can close while the people who wanted it are still deciding. The same gap in a quarterly review is a record of business you already lost.
Which means the delivery mechanism matters as much as the data. A reporting tab gets opened twice.
What arrives
What we handled. Conversations, how many escalated to tickets, how many questions could not be answered, and tickets created and resolved. The shape of the week in five numbers.
The gaps. The clustered questions your site could not answer, each with how many times it came up, and a link straight to the full list. This is the section worth reading, and it is why the email exists.
AI spend. Estimated dollars on your own Anthropic key, with a plain note that your Anthropic invoice is the real number rather than presenting the estimate as authoritative.
Empty sections are omitted
If there were no unanswered questions this week, there is no gaps section. If AI spend was nothing, that section is absent too.
This is a small thing that decides whether the email survives. A digest padded with zeroes reads as noise and gets filtered within a month. One that only tells you things when there is something to tell keeps getting opened.
Who gets it
An address you nominate, falling back to the site administrator.
Nominating one properly is worth a moment’s thought. The person who should see failed questions is often not whoever installed the plugin: on a store it might be whoever writes the product pages, or whoever decides what you stock.
Off by default
One checkbox to enable it, plus the recipient. Switching it off unschedules cleanly rather than leaving a job running.
Why weekly
Daily is noise; most shops do not accumulate a week’s worth of pattern in a day. Monthly is too slow for the gaps list, which is the part with the shelf life.