The client and the two periods
Take a synthetic mid-sized client, aliased Northstar, across two quarters. Tickets opened rose from 112 to 134 and closed from 106 to 128. SLA breaches fell from 8 to 3. Patch compliance rose from 86% to 94%. Backup success slipped from 99.2% to 98.4%. Critical alerts fell from 4 to 2, recurring incidents from 19 to 11, unused licences from 14 to 12.
Eight numbers, two periods. Nothing here needs a custom report, and nothing identifies an end user. That is the whole input for the page below.
The headline
Service health improved; review unsuccessful backup jobs and agree a target. One sentence: direction, then the single action. A client who reads nothing else knows what the quarter was and what is being asked.
Notice what the headline does not say. It does not claim the MSP caused the improvement, and it does not quantify a saving. Both would be arguable, and an arguable headline sets the tone for an arguable meeting.
Three findings, each with its evidence
Patch coverage up 8 points, from 86% to 94%, cited to both period rows. SLA breaches down 63%, from 8 to 3, also cited to both rows. Backup success down 0.8 points, from 99.2% to 98.4%, flagged as the one metric moving the wrong way.
Each finding names the rows it came from, S01 and S02, so a sceptical client can check the number without asking for the export. The detail line says what the metric is, not what it proves; a fall in breaches is a fall in breaches, not evidence of a process change until someone says so.
Three decisions requested
Review unsuccessful backup jobs and agree a target, owner: client, evidence S02. Review critical alerts before the QBR, owner: MSP, evidence S01 and S02. Confirm patch exceptions and set the next target, owner: joint, evidence S01 and S02.
Each decision rests on a finding, has an owner, and asks for something specific. A decision without an owner is a wish, and a QBR full of wishes is a status meeting with better formatting.
The gaps and the source ledger
The page lists what the two rows cannot prove: why backup success slipped, whether the fall in breaches reflects fewer incidents or a changed threshold, whether unused licences are seats or subscriptions. Naming these reads as competence. Leaving them out and getting caught reads as the opposite.
The ledger at the bottom is the two aggregate rows, verbatim. Anyone at the client can reconcile any claim on the page against them in a minute. That is the property that makes the report defensible, and it is the one most decks lack.
Building your own from the same shape
The example above is the fixed sample the Sourcerow generator runs in the browser. Paste your own two periods in the same ten-field template and the same shape comes back with your numbers: headline, findings, decisions, gaps, ledger. Review every line before a client sees it; the page organises evidence, it does not replace your judgment.