Before the reading path
One page before the design.
This brief is for someone who has to judge the proposal, not implement it. The numbered chapters start at Purpose and capabilities. They use the same facts.
The problem
A comparison needs a stable answer: these two offers are the same exact product, they are different, or the evidence is not good enough yet. A similar title is not that answer. A listing URL is one seller’s page. A model suggestion is not a decision.
The people who feel a wrong answer are buyers of the comparison, operators who have to correct a match, and engineers who later have to explain why a public page changed.
What success would look like
- The original page and the time it was seen are still there when someone challenges a match.
- The exact product stays separate from each merchant’s offer, including packs and variants.
- A correction points at a decision and at evidence.
- The public sees only a release an authorized person has applied.
- Recurring services stay inside USD 50 a month. That ceiling is the only confirmed limit.
What this guide will not decide
The implementation plan is unapproved. Workbench screens are prototype demonstrations. They are not a deployment. Jonathan has not chosen the database topology, the recovery targets, the first real source, the publication authority, or the identity provider. Proposed thresholds, including the disagreement between three queue deliveries and a dead letter after five attempts, stay proposals.
Pantry & Pour and the Lumen listing are fictional fixtures. They prove a deterministic path. They are not evidence that a real shop can be collected.
Where to go next
The short path is Purpose and capabilities, then Follow one product. Open decisions are listed as a checklist on the implementation roadmap. A proposed Broadwater pilot checklist, not an accepted pilot, sits beside the implementation gates and the outage walkthrough.