| Sector | WealthTech — B2B2C investment platform |
| Engagement | Performance engineering |
| Delivered | 2022 |
150% of anticipated peak load validated
9 load test types executed
1 major defect caught pre-launch
The Challenge
Our client was preparing to launch a B2B2C wealth platform — one where partner institutions would bring their own customer bases onto the system at go-live.
That launch profile is unforgiving. Unlike a consumer product that grows gradually, a B2B2C platform can go from zero to substantial concurrent load on day one, when partners switch their users across. Get the capacity planning wrong and the failure is public and immediate.
The client had no formal testing function and no performance data at all. There was no baseline, no load model, and no evidence of how the architecture would behave under real conditions.
The Solution
We built a performance engineering programme from nothing.
The first step was establishing what peak actually meant — modelling realistic concurrency based on the partner onboarding plan rather than guessing at a number. From that model we designed and executed nine distinct load test types, covering sustained load, spike behaviour, endurance, and the specific scenario of multiple partners onboarding simultaneously.
We deliberately tested beyond expected volumes, validating the platform at 150% of anticipated peak to establish genuine headroom rather than confirming it could just about cope.
That testing surfaced one major defect that would have degraded the platform under concurrent load — found weeks before launch rather than during it.
Business Outcomes
- Platform validated at 150% of anticipated peak load
- 9 load test types executed end to end
- One major defect identified and resolved pre-launch
- Moved from no testing function to launch-ready
- Capacity headroom established with evidence rather than assumption
Facing something similar?