The Challenge
A US airport advertising company sells inventory across airport locations, and their Salesforce org tracks the locations, the advertising inventory in each one, and the contracts written against it. We had delivered the original data model and profile build in 2024 with three people and no AI assistance. Two years later they needed the follow-on phase: extend the model and migrate the historical data onto it.
Our Solution
We delivered the follow-on phase against the same objects, with Claude in the loop for schema work, deployment, migration and verification. Because we had delivered the first phase ourselves and both engagements were tracked in the same time-tracking system, we could measure the two against each other directly — the same client, the same data model, the same firm.
Outcomes
compression in hands-on build work vs. the 2024 phase (83.00 hrs → 17.75)
records migrated to the new model with zero failures
delivery window, down from 42 in the 2024 phase
- 177 of 177 records reconciled against source with zero mismatches at go-live
- 17 of 17 schema components deployed; zero errored automation runs
Human hours are a floor: nine days in the 2026 phase had machine activity with no matching timesheet entry, so the true compression is somewhat lower than 4.68×. We publish the figure that flatters us least.