If I am manufacturing computers, and I receive a large batch of CPU’s… say… Intel Pentium IIIs, I put them in inventory. There is a recurring montetary cost associated with keeping this inventory. There is, however, also a hidden cost. If I use up half my inventory to build computers, and then I inventory those computers, and I ship half of those computers, I now have 3/4 of my initial chips waiting around, earning me no money at all.
Then, Intel ships the new Pentium 4. What happens now? Well, I basically have to throw out my Pentium 3 stock (either by making low-cost, reduced-margin computers just to get rid of them, or by trying to sell the chips wholesale at cut rates). I also have to either slash prices on my remaining stock of computers, or re-fit them with the new processors. All of this is very expensive, and may eliminate most or all of the profits I was expecting to make on the original purchase. It is a scenario that is rife with waste.
Most manufacturing industry players have figured this out, and moved to a Just-In-Time model of production. Toyota pioneered much of this in the 1970s, and now inventory is seen as a liability, rather than an asset. This waste-free approach is a centrepiece of the Lean manufacturing revolution
Waste and Inventory in Software Development
Unfortunately, there is a parallel in computer software creation that has been rather poorly understood by businesses that need custom software development. Waste can be considered as anything that is unfinished, and/or unused. In software development terms, this can be applied to documents that will never be read or that might be unnecessary, and other such things. It is also true of unfinished software.
Traditional Software Development Example
Let us suppose that we ask someone to develop a piece of software with thirty features. Let us further suppose that this software will take about twelve months to produce. Let’s further suppose that ten of these are business-critical features, and that all of the features’ definitions are highly market-driven.
So a traditional software project starts to develop them. First there is a requirements analysis phase, then a design phase. Throughout there are lots of approval stages, sign-offs, etc. Then some team codes up the software starting somewhere around month five. Coding proceeds for about three months, after which the software goes through a testing process in month nine. In theory, at the end of this testing process (month twelve), we are supposed to receive the software for use, and we should be able to receive the value it was supposed to offer. However, defects are found in testing, requiring some re-work. The software is delayed by a further three months, and we finally receive it. By the time we receive it, our competetors have defined new features, and we have to submit new feature requests, whose results will not be seen for several months.
So for the entire development process, we received no business value, and continued to pay. Fifteen months from first investment until we started to achieve results that could mean revenue from the system. At this point, the software is partially obsolete. This is quite similar to the manufacturing scenario above.
Lean and Agile Software Development
Agile and Lean software development practices change this process by delivering business value a little-bit at a time. In an Agile software project, the business specifies what the most important features are (in our case the ten business-critical ones). The team then begins working in small time-boxes – say, two to four weeks. In each time-box, the team works on a small but well-defined list of features taken from the top of the prioritized feature list that we specified. At the end of the time-box, those features that were worked on are delivered to a degree of quality that each of the delivered features would be suitable for use by the customer. The whole might not be ready, but any features worked-on would be complete and production-ready.
If the team can average about three features per month (about what they pulled off in the above example), then by month four, the team could deliver a piece of software that has all ten business-critical features ready for use. The whole might not be ready, but the customers could determine whether those ten were worth taking the software in its current state, and packaging it up and deploying it. Quite simply, the software may not be “done”, but it’s “done enough”. Then the team can continue to roll-out less valueable features from the prioritized list.
Market and customer needs volatility
This becomes even more important when we consider the competition’s changes. In the traditional example above, after fifteen months, additional features were required. These may have been discovered in month six. Traditionally, these could only be considered in month sixteen. In Agile and Lean software development, however, work on these changes could be started in month seven or sooner. So the business received value early, or “just-in-time”, and they could get high-value changes “just-in-time”.
Because testing is built-in to the process, the customer is able to validate the product at the end of every time-box, so there are no large batches of re-work. The only time large batches of rework are necessary are when large changes to the basic requirements are requested. And then, Agile does not resist the customer’s desire to change, but recognizes it as an essential part of software delivery, and adapts to the changing consditions. As a result, the Agile and Lean methods are optimized to produce quality product in small increments, so quality doesn’t suffer from customers or markets changing their minds.
Agile and Lean principles are very powerful, and allow for business value to be delivered sooner to end-customers. This allows for better quality and risk management. It also allows strategic and tactical decision-making by executives to be undertaken when such decisions are most needed – not six months too late.
Firms that embrace Lean and Agile development principles are beginning to see the competetive differentiation that Toyota saw vis-a-vis the rest of the automotive industry throughout the 70′s and 80′s. It is only in the last two decades that, after adopting Lean practices, Toyota’s competitors have begun to close the gap. Agile software shops are beginning to achieve these same competetive gains. Those who don’t adapt to just-in-time value delivery – who don’t eliminate waste in their processes – will feel the results as their competitors take products and services to market far faster, and with more responsiveness to their customers.
“If you’re not on the steam-roller, you’re part of the asphalt.” – Lighthouse Design, circa 1996.