Why most features fail to deliver the expected value

The truth is that we don't know how customers will behave.

When we build a new feature we are essentially saying that we want our customers to change their behaviour. And people are complex. It is easy for things to make sense on a whiteboard; customers might be asking for it and your competitors might offer something similar and yet, more often than not, the features fail.

Our systems assume that the value is there so they focus on efficiency

Business cases and annual planning lock in ideas at the point when we know the least about whether it is a good idea. These processes date back to when companies started to scale in size for the first time during the industrial revolution. The focus at the time was on efficiency because we knew there was demand if we could make existing products cheaper. Today we are building new products so it is a different context. Unfortunately, we can't be sure of the demand upfront.

The outcomes are not pretty

Any company that has studied its own feature success rates knows that we are wrong more than we are right.

Success Rates

Netflix 2007
Microsoft 2009
Google Ads 2012
Bing 2014
Airbnb Search 2020
Booking.com 2020

Source: Kohavi et al

Microsoft is an outlier at 33%, but the author believes that this is because success is easier in less well-optimized domains. The challenge is that all domains are becoming more optimised over time so the 8- 10% success rate is likely to be the norm.

How can we improve the odds?

We need to acknowledge that we don't know if an idea will work. Using this as a starting point it is clear that we need to minimise the cost of failure by doing more upfront research as well as breaking big ideas into small releases to test the value. This is easy to say but really hard to do because it involves changing how software is funded, how teams are structured, the roles that people follow in teams, and the way that we release to customers.

The good news is that there are lots of companies doing this today. We can learn from them so we don't have to reinvent the wheel and we can learn from the mistakes that they have made. Even so it is still difficult to introduce change in established organisations. Our goal at UXDX is to help make change easier and that is what guides the structure of our conferences and training courses.

If this sounds like a problem that you are facing then check out our conferences to see how we can help you and your team.