Paul Hammant's Blog: Like a Used Sofa
Stand by for: A story about delaying a significant purchasing decision, while locking costs early, but also being fair to the supplier.
Big Software Purchasing Context
Some years ago, for a ThoughtWorks gig at a client, I was the technical assessor helping a client choose a third-party trading app. It was a formal vendor/app selection project, and the traders that would use it had really high-end needs. One was a shared view of their outstanding orders on the exchange, while remaining compliant with the regulators rules on that. There were dozens of choices of software to use. Many of the choices were in use on other exchanges. This electronic exchange was new though, as it was at that time open outcry, and for a specialist tradable thing. It was scheduled to go live in, let’s say, six months after the moment portrayed in this story. The exchange had told everyone this was going to happen, and even the ‘open outcry’ folks had come believe the change was going to happen. A small trade show had already happened, with many vendors touting their wares, before I had come onto the mission. I was to evaluate non-functional technical aspects. Objectively rate short-listed products against a bunch of technical criteria, they said. Here’s just five of many:
- Development technologies, with a perceived modernness/suitability.
- Application is WIN32 not WIN16.
- Vendor is giving free upgrade to WIN32 if the present version is WIN16.
- Development follows an Agile methodology. Successfully so?
- Development has automated tests.
So the client had whittled down many vendors to three. I’d completed my bits and pieces, as had ‘finance’ (vendor is not going under), ‘legal’ (we get source to the tech if they do go under - from escrow), and the business (snag list of features we need is x long). That took interviews, and visits. Demos too, but I didn’t see those.
The decision meeting
The critical decision meeting, and the singular moment that gives rise to this article, was scheduled for 8am. We’re all invited, as we had sections of a spreadsheet we could speak to. About 6 of 12 invitees were there at 7:53am. Only one of the traders (so far) and the client’s PM wasn’t present yet, and it was he who’d MC the meeting. The trader was busy on his Blackberry (or some other business gizmo popular at the time) and not conversing with other early birds.
At 8:01, the trader broke the silence (we’re still only eight or so people):
“Right, we don’t want vendor X as it was like they were trying to sell us a used sofa in the demo at their offices. We like both vendor Y and Z’s products, but both need different modifications as you know. Pay the time and materials fees for the modifications we need, and take an option on the purchase price for the day before the exchange goes live. <small pause> Got that?” (swear words omitted).
Almost immediately, the trader-team’s sole attendee gets up and leaves through a door. Twenty seconds later, the PM arrives through another door:
“Good, I’m here before the traders have arrived”
We sheepishly explained what had happened.
Stroke of genius stuff. At least to me it was, though Chris Matts had to bring me up to speed later, when I recounted the story.
Companies that are sufficiently well funded, can hedge their bets for wholly competing ‘A or B’ alternates by funding both. It is brave for sure, but with canniness, costs can be constrained, and objective gates and measurements can be nailed in to the project design. Fail-fast is a prerogative. Particularly for the alternate that less safe/methodical, or whether the effort required could be either six months or infinity.