Growth ships a campaign screen with purchase_success. Core product already sends order_placed from the native cart. Both squads can show a healthy chart. Neither chart is the full count, and adding them double-counts shoppers who used both paths in one week.
Parser event validation exposes the split when the sample includes both screens and a source column. Ownership is the slower part. Farah Idris’s intake now asks who believes they own conversion before we look at types and enums. If two names already sit in the weekly deck, we treat that as a finding even when both payloads are perfectly typed.
Stop arguing about which name is more “correct”
Correctness here is a catalogue decision: one surviving conversion, or two conversions with different meanings. The first needs a rename map. The second needs the board slide to stop adding them. Either decision is better than a third name arriving from a new campaign next quarter.
The Penang payments example in our client stories was this pattern after a merger. Reconstruction listed triggers so the argument left Slack and sat in a document with owners. Cleanup came later, once finance agreed how 2025 should be read.