# Brief format Two variants: **DM** and **channel**. The profile's `deliver_to` decides which. They share a structure and differ in what detail is safe to include. ## Design constraints The PM reads this on a phone, before coffee, between the alarm and the first meeting. **The first line must survive being the only line read.** Put the single most important thing there. If nothing is important, say that there — "nothing needs you today" is a complete and valuable brief. **Sections are ordered by what someone must do, not by data category.** Escalate first, because it's the only part with a deadline. Areas and customers are supporting context and come after. Never pad a section to look thorough. An empty Watch lane means a good day, and printing "no items" is more honest than promoting a Noted item to fill the space. Per-ticket detail, full cluster membership and the area-by-area table go in the **thread**. --- # DM variant Private, so it can carry named accounts and tier labels. ``` *Ticket Radar — * *Escalate today* • ** — tickets, accounts, since *Watch* • ** — tickets, up from typical. *Aging* • tickets open past days with no reply. Oldest: days, . *Your areas* : new (typical ) · : new (typical ) *Who we heard from* accounts. Notable: — . __ ``` ## Worked example — DM ``` *Ticket Radar — Tue 18 Aug* Invoice PDFs have been failing since Friday's release — 7 tickets, 4 accounts, two of them Enterprise. *Escalate today* • *Invoice PDF downloads return blank files* — 7 tickets, 4 accounts, since Fri 15 Aug Customers click download and get a 0-byte file. Started within a day of the 14 Aug billing release, so it reads as a regression. Clears the escalation rule three times over. Northwind [Enterprise], Contoso [Enterprise], Barton [Growth], Vale [Growth]. #48213 #48227 #48231 *Watch* • *Saved payment methods disappearing on mobile* — 3 tickets, up from ~0 typical. All Android, all this week. Under the bar on account count, but the trend is one direction. *Aging* • 4 tickets open past 7 days with no reply. Oldest: 12 days, Fabrikam [Enterprise], card decline loop they've now chased twice. #47905 *Your areas* Checkout: 4 new (typical 5) · Billing: 11 new (typical 3) · Payment Methods: 3 new (typical 2) *Who we heard from* 9 accounts. Notable: Northwind [Enterprise] — third billing ticket in ten days, worth a check-in independent of the PDF bug. _Covered 08:00 Mon – 08:00 Tue across Zendesk. 41 tickets read. Service Cloud not connected._ ``` Notice what this does. The headline names the problem and its blast radius in one sentence. Each escalation states *why it qualified*, so the PM can argue with the reasoning rather than taking a verdict on faith. Baselines appear inline as "typical N", which is what turns "11 new on Billing" from a number into an alarm. And the Northwind observation is a pattern across tickets that no single ticket would have revealed — that kind of connection is most of the value here. --- # Channel variant Same signal, wider room. Three things change, and the reasons matter more than the rules. **Address the PM by name.** In a channel, `@mention` them in the first line. A brief with no owner is a brief nobody acts on, and in a shared channel the diffusion of responsibility is real. **Drop commercial detail.** No tier labels, no "our biggest account", no revenue framing. A channel usually has a wider audience than the PM's mental model of it — contractors, new joiners, occasionally a guest account — and account value is the detail most likely to cause a problem if it travels. Say "4 accounts" rather than "4 accounts including two Enterprise". **Let ticket links do the identifying.** Engineering needs to reach the actual ticket to debug, and the link does that job without putting customer names on a channel wall. When `name_customers` is true (the PM explicitly overrode the default), names are fine — otherwise refer to volume and link out. The upside of this variant is real: engineering sees the escalation at the same moment the PM does, so the fix starts without a relay step. That's worth the coarser detail. ``` *Ticket Radar — * · <@PM> *Escalate today* • ** — tickets, accounts, since *Watch* • ** — tickets, up from typical. *Aging* • tickets open past days with no reply. Oldest days. *Areas* : new (typical ) · : new (typical ) _ Full detail in thread._ ``` ## Worked example — channel ``` *Ticket Radar — Tue 18 Aug* · @sam Invoice PDFs have been failing since Friday's release — 7 tickets across 4 accounts. *Escalate today* • *Invoice PDF downloads return blank files* — 7 tickets, 4 accounts, since Fri 15 Aug Customers click download and get a 0-byte file. First report landed within a day of the 14 Aug billing release, so it reads as a regression rather than a long-standing gap. #48213 #48227 #48231 *Watch* • *Saved payment methods disappearing on mobile* — 3 tickets, up from ~0 typical. Android only. *Aging* • 4 tickets open past 7 days with no reply. Oldest 12 days, a card decline loop. #47905 *Areas* Checkout: 4 new (typical 5) · Billing: 11 new (typical 3) · Payment Methods: 3 new (typical 2) _Covered 08:00 Mon – 08:00 Tue across Zendesk. 41 tickets read. Full detail in thread._ ``` The regression framing survives into the channel version on purpose — it's the sentence that tells an engineer where to start looking, and it names a release rather than a customer. --- # Quiet day Same in both variants, minus the mention in DM. ``` *Ticket Radar — Wed 19 Aug* Nothing needs you today. 6 new across your three areas (typical 7). Nothing aging past 7 days. _Covered 08:00 Tue – 08:00 Wed across Zendesk. 6 tickets read._ ``` Three lines. Resist expanding it — this format is what teaches the reader that a long brief means something real. # Still learning For the first week in an area, or any area with fewer than 7 rows of history, replace "typical N" with an explicit note rather than a fabricated comparison: ``` *Your areas* Checkout: 4 new · Billing: 11 new · Payment Methods: 3 new _Baselines still building — day 3 of 7. Trend comparisons start Friday._ ``` It's tempting to compare against two days and call it a trend. Don't. The first spurious "up 300%" costs more trust than a week of honest silence buys back, and trust is the entire product here.