Agile is cheaper…

In a series of posts I’m going to critically review the common value propositions of agile – do the claims stand up to scrutiny?
Claim 1: Agile makes for cheaper software.
There are a number of justifications for this claim.

  • No more kitchen sink – with waterfall a common pattern is the tendency of the customer to squeeze in all possible requirements at the beginning of the process since the ‘change control process’ will make it very difficult to add requirements afterwards, and any changes would give a justification for the development team to slip the schedule. So lots of features, poorly understood and unlikely to be needed, are included up front. This leads to more complexity in the design which now has to take into account more features, delays starting time of development as it takes longer to get signed off requirements, and puts features with low or even no current business value on an equal priority for delivery as critical functions (since all must be delivered in a big release). The software delivered in the end therefore has more functionality than in agile – it costs more to produce.
  • Invest only in artifacts with business value: Agile values working software over documentation, but also values documentation with business value over that merely used to ‘run the project’. So time spent developing requirements specs, test plan documents, status reports, etc can be spent instead on great user guides, operations guides, regression test suites, etc.
  • Lower Opportunity Costs: Because agile delivers value earlier, it reduces lost opportunity costs and enables the business address opportunities in the market more quickly.
  • Lower Product LifeTime costs: The up front investment in continuous integration, automated test suites and refactored, well structured code has been shown to be rewarded several times over in reduced ongoing maintenance costs throughout the softwares life, and even extending that life beyond where ‘dirtier’ code becomes uneconomical to maintain.
  • Delayed Investment: By returning business value from early in the development lifetime, an earlier return on the investment is realised, and further investment can be delayed till its really necessary.
  • Global vs. Local Optimisation: A key lean principle is global over local optimisation. Large batches rather than continuous flow, specialisation rather than skills redundancy and other traditional tactics to increase resource utilisation aim to acheive optimisation of a particular task set such as development or testing. But it is the optimisation of the whole that is important and agile cuts costs by focusing on this global optimisation.

Are there other ways agile cuts costs? If you know more please share them…

Next blog I’ll look at how agile can speed time to value.

Agile Magic

With several companies I’ve worked with, agile has been seen as a silver bullet to solve problems of late delivery, bad quality and high cost. But agile is more than a set of practices – its a system of values and principles that need commitment and resources to bring success. Senior management will often say “this team is agile” just because they read the book, have daily stand ups and use words like sprint, iteration and user story. But this has been likened to a cargo cult, where performing agile ‘ceremonies’ are somehow construed as bringing benefits of a true agile implementation. Perhaps this is another aspect of the ‘illusion of control’ invoked by detailed planning and scheduling of projects in waterfall methods. And like the damage such illusion did to waterfall, blind belief in a sprinkling of agile magic dust can prove to be a major setback for true agile adoption.

Invisible Work

Agile practices such as test driven or test first development, continuous integration, refactoring and automated builds are generally invisible to the customer. Work on these doesn’t deliver any features or, it first seems, value. But these are essential to the longer-term quality of the product, the ability to frequently deliver working increments, etc. Activities like integration have to happen at some stage – taking this pain early in the project gives much better predictability for the remaining deliverables. Customers need to be educated up-front about the benefits of these practices – trying to ‘hide’ them, as I’ve seen teams do, only leads to mistrust.