This is a follow-up to my post on fixed-price contracts written some time ago. It was a surprisingly popular post, and I received some interesting feedback and many new ideas.

Now procurement matters. Most enterprises buy far more than they build, and that isn’t going to change soon. Getting a decent bang for the buck is therefore a big part of enterprise IT, and all kinds of things can get in the way. Today, an article in the Stack caught my eye — IBM had just won a £100 million contract from NS&I, the British equivalent of India Post Bank. The actual tender is online, and its a contract to build the following:

“The second procurement in NS&I’s Rainbow Programme, the Digital Experience and Digital Enablement package, will provide capabilities in digital self-service experiences and journeys for NS&I’s customers and enable Assisted Digital support. This includes as a minimum the delivery of a mobile banking app (to include prize draw functionality) and a fully functional transactional website.”

Basically, digital transactions app and website. And if you’re wondering how much £100 million is, well it’s about ₹1,050 crore, or what is also known as a pretty penny (India Post spent ₹700 cr on its entire core banking). And no, salaries don’t do anything to make up the difference (much of the work will be done in India anyway, in all likelihood).

What this is leading to is that maybe I need to refine my prescriptions a little. I had recommended three solutions to the Fixed Price problem, namely:

1. Trusted Partner
2. Pared down requirements
3. Moving fast and small

However, I want to add a few layers to this

1. Trust more partners. Only five people responded to this tender, yet there are many quite capable of building a mobile and web banking app given the hundreds (maybe thousands) of banks across the globe that have transaction apps and websites. The actual tender is restricted, but I suspect qualification criteria left a lot of perfectly capable players out. 
2. Don’t be excessively risk-averse. As a buyer, trying to move every risk on earth to your supplier is going to cost you (and will in reality not do all that much to temper the risk). NS&I blamed its suppliers for the failure of earlier rounds of tendering (apparently they were not willing to accept enough risk) and finally ended up with Big Blue who does not exactly have a reputation for cost-effectiveness.
3. Question your estimates by looking at a wider field for comparison. In this case, the tender had an original estimate of £169 million, so IBM’s bid of £100 million did not seem such a bad deal. NS&I should have looked at what this effort had cost banks of similar complexity globally — except the people component it should be very similar. NS&I isn’t that large a bank either — Kotak at 50 million has roughly twice the number of customers and it certainly didn’t pay ₹1,000 crore for its mobile & web front end. RBL Bank (14 million customers) would have spent about ₹20–30 crore over multiple version releases. I’m implementing a bank web and mobile and all in Papua New Guinea (where all costs are exorbitant because it is so remote) at about $3 million.

All governments repeatedly demonstrate that all their perfectly well-meaning rules of procurement don’t actually lead to good results in any area — quality, cost or timeliness. So one must ask, what is a good procurement process supposed to deliver and how to tweak it so that it does?