How much is a feature worth?
Software teams are pretty good at figuring out how much a feature will cost, and we track those estimates very closely to make sure a project goes on schedule. But we rarely ask the much fuzzier—and no less important—question of what a feature is worth.
When making such evaluations for consumer software, it’s very difficult to estimate the actual dollar value for a new feature. However, for enterprise software or software for internal users, it’s a much more tractable problem—and more important, too, since you typically can’t spread the costs over thousands of users. In fact, a package built for internal use only might have just a handful of users.
It’s obvious that if a feature can lead to a 100% efficiency improvement in a large department, it is probably worthwhile. But for smaller wins or improvements that might take several years, how do you calculate the benefits?
Atomic has been maintaining part of an app for a small two-person team for several years, and we were considering overhauling a portion of it. Between old technology, updated workflows, lessons learned, and simple feature creep, there’s room for improvement.
To evaluate the worth of a new feature, I used a quick back-of-the-napkin discounted cash flow analysis. (Essentially, we say that a dollar today is worth more than a dollar a year from now: Industries can shift over time, so the discount rate represents those risks.)
I made this estimate:
If each person costs $40 per hour (I have no idea what they really cost, but we should be able to get an order of magnitude estimate), and I can save them a week’s work every year, that might add up to $3,200 per year. At a 20% discount rate, the savings would be:
- $3,200 the first year
- $2,560 the second year
- $2,048 the third year
- $1,638 the fourth year
That adds up to $9,446, so any update that can save two-person weeks per year would be worthwhile if I can do it for under roughly $10,000.
This is a useful tool to have, because now I can look at all of the potential improvements side-by-side with my estimates. Armed with a concrete yardstick to measure them, it’s easy to eliminate bad investments. In this case, I wasn’t confident that I could deliver a week-saving improvement for less than it would save, so I passed on that feature and am focusing on more widely-used areas of the app.
Not every question is so amenable to measurement. Often, you must take other things into account, such as customer acquisition or company politics. But by looking at the costs and the payoff over time, even with very gross estimates, you can frequently get a better understanding of the potential ROI–or lack thereof.