My husband and I are involved in a never-ending construction project that takes us to wonderful Portland businesses solely in search of odd things. A few months ago, we went to the stone yard. At one end of the stone yard were the boulders. They were big. Big enough that it was hard for me to conceive of anybody buying a boulder, because where would they put it? Other stones were arranged in progressively smaller sizes, from boulder down to dust. Which got me thinking about sizing, and about software.
Software starts as a business goal. The goal can be anything. It can be to reduce inbound calls to the service center or increase web conversions or decrease staff time on a specific workflow. Frequently, the people with the business goal have a series of hypotheses about the goal. “We can reduce inbound calls by integrating with a third-party chat provider,” say, “or by building a knowledge base,” or even “by hiding our phone number on the web site.” And each of these hypothetical solutions have a cost to build, which is important information for the business to have. But the business doesn’t need to know that Option A will cost $307,897 and take 5 months and two days to build, while Option B will cost $267,897 and take 4 months and 29 days to build. They need to know if Option A is a boulder or a pebble. They need to know T-shirt size.
I love T-shirt sizing. I love it because it provides just enough information, at just the right time, without creating downstream problems. I love it because it’s simple, and because it’s true. Here are four reasons I love T-shirt sizing:
1) Relative value is universal.
Everybody has a gut sense of what a small is, versus a medium, versus a large or an extra-large. Even if everybody’s medium doesn’t equal 4.5 team days, everybody’s medium is still relatively smaller than everybody’s large. T-shirt sizing provides a simple, universal measuring stick.
2) T-shirt sizing doesn’t eat resources.
Developer time is the most valuable resource a software organization has. In addition to the actual work of writing code, the team has to support daily and weekly scrum activities. The last thing they need is a business ask that takes them away from sprint work.
Good T-shirt sizing can be done by one team lead and one developer, with a minute or two spent on every line that needs to be estimated. A year’s worth of work can be sized in less than an hour. This gives the business the relative values it needs to make decisions without taking away time the developers could be using on sprint work.
3) T-shirt sizing stymies the desire to plan the project too early.
The business wants a plan. Whether it’s the CEO or the client or the head of marketing, the business wants to know how much the feature is going to cost, how long it is going to take to build, and when the team can deliver. But the business rarely asks that question directly. The business asks for high level estimates, and if the team provides too detailed of estimates, the business puts those estimates in a deck, the deck gets turned into a roadmap, and pretty soon the developers are working against a plan that wasn’t sized or vetted.
T-shirt sizing prevents this dynamic. It answers only one question: does Feature A require more work than Feature B. This gives the business just enough information to do a value tradeoff, but not enough information to create a project plan. No decks will be created with T-shirt sizing. Nobody will ever feel they have been hoodwinked into a project plan.
4) Relative value is the right question early on.
Two solutions are proposed. The business has to decide to invest in one, both, or neither. Both have equal chances of success. One of them is sized at an XL. The other is sized at a Small.
This is the right kind of information for the business to have when weighing two solutions against each other. The business doesn’t need to know the proposed architecture, the full stack, and the third-party integrations for them to weigh relative value. One solution is an XL. The other is a small. Both have equal chances for success. The smart business would go with the size small, learn from the solution, and iterate. T-shirt sizing forces relative value trade-off before too much energy is invested in detailing a solution that might, or might not, be the right choice.