In my day job a work as a management consultant. We recently had to help a tier 1 bank get ready for Basel II accreditation (so they could get cheaper access to credit). After a quick chat with their IT people it seemed like a solution would take 6 to 9 months to develop. Ouch. The business wanted something up and running in 4 weeks. Naturally we said we could do it. Here’s how it happened from a customer discover perspective.
- The Problem: The bank had a pretty good quality assurance process in place, but a lot of uncertainty around what information needed to be reported. This would be our key focus in validating the problem to be solved
- The Customer: We set this to all staff involved in the quality assurance process
- The Solution: We created a quick and dirty prototype (in Excel) that was good enough to capture data from all staff, and could be changed really quickly. We used MashZone for the reporting because it could take data from Excel and turn out an interactive web page within a day or so.
It took us a couple of days to set up Excel to capture answers to a bunch of quality assurance questions. And another couple of days to hook it into the reporting system. We used live data from day 1. Why?
- real data means real reports. This focuses the minds of your executive stakeholders far more keenly than test data and mock systems.
- quick and dirty prototypes look quick and dirty too. This means your target customers (a subset of banking staff in this case) are more forgiving of errors and mistakes. They needed to be, because we intended to fail early and fail fast. This was the only way we would really find out what worked and what didn’t in this sort of timeframe
- we would be reading a reporting only, not editing or saving, so we could work of copies of the original data. Real data, low risk
And that’s how we got a working system rolled out to all staff in under 4 weeks. We still ended up with a prototype system, but at least it worked. And IT could use this as a form of requirements document for the final solution (build us one of these, but a bit more robust).