The greatly exaggerated death of T&M

There’s a lot of chat online about the death of Time & Materials (T&M). This ranges from analysts, management consultants, and even agile luminaries. But, it feels weird to me. 

Much of this commentary is linked to AI assisted software development, removing the need for software developers. It’s a seductive story for the companies selling AI models, and a consulting industry that has been rocked by volatility for half a decade and wants to reinvent itself. But, I don’t buy it. Pricing is about risk, trust, agility, and market forces. AI doesn’t change those fundamental things.

Even as software becomes more efficient to create, it doesn’t undo the arguments that  the agilists won a long time ago. That creating great digital products is a human, iterative, imperfect process that requires the ability to change. If AI allows you to ship everything  faster, good or bad, that agility is more important than ever. Let’s look at some factors behind pricing:

Trust

AI is changing the cost of building software. To what extent, and how it differs for different types of work, is still up for debate. That uncertainty chips away at trust between consultants and clients. Clients might feel like they’re paying for weeks when the work takes hours. This is made more complex by opportunists who will sell a dream, understanding that by the time reality kicks in, they would have made their money.

Fortunately, the ambiguity is temporary. Like agility, consultancies will benefit from AI to differing levels, but the possible improvements will become clear. I saw a Megabuyte talk recently that said margins in consultancies have been incredibly consistent throughout many waves of efficiency improvements. The market dictates who gets the benefits, and competition means it is always eventually the client.

Agility

Agile is fundamentally about iteration. The holy grail of agile is the shortest feedback loops possible. Not to minimise the cost of development, but to minimise the cost of learning. AI does present a huge step forward in this. Ideas can be tested very quickly, and we can validate, refine, and discount ideas much more quickly than before. But, assuming that should lead to more fixed timescales makes no sense. The more we know, the better products we can build, but only if we allow ourselves the flexibility to change. AI might lower the cost of building, but it doesn’t lower the cost of being wrong.

In an industry where some fundamental assumptions are changing, it can be tempting to sacrifice agility for certainty. But this is the moment we need agility the most.

Risk

Sharing risk is perhaps one of the flaws of the agile revolution in consultancies. Those same opportunists that take advantage of AI uncertainty have been known to do the same with agile. At its worst, it can lead to unbalanced relationships, where clients are allowed to let scope spiral, and the only risk for suppliers is their reputation. The market asking consultancies to share risk, and for work to be done within a certain budget or timescale is also reasonable, and has always been the case.

Constraints are a good thing. Having to work within a budget or timescale forces us to prioritise, and to find something good enough to release. Without constraints, we’re more likely to build bloated products, or never get things to live. Again, the risk of this is higher with AI, not lower.

The fact that sharing risk and working to constraints is reasonable and often helpful doesn’t make it simple. Businesses are messy and often have tightly coupled dependencies, release processes, and large groups of stakeholders. This is why fixing everything (scope, time, price) is a fallacy that always ends up with a winner and a loser.

So, what can we do?

At Waracle, we’re staying focused on models that work. When the work has tightly coupled dependencies, or we’re building something that’s highly likely to change as we learn, we’ll continue to embrace ways of working and commercials that support that.

When our clients have, or want, to get the work done within constraints, we’ll support that too. We’ll do that in a way that shares some of the risk through incentives, without sacrificing all agility, and without pushing a burden of negotiation to the delivery team. AI makes the need for agility stronger than ever.