Product team tends to pump features into product roadmap and carries those features (software inventory) month over month. In some extreme cases, it might take four months to burn down those inventories. Features on the bottom of inventory list never got implemented. Your roadmap can get messy. A useful framework of cleaning your roadmap is to categorize your features into three buckets: metrics movers, customer requests, and add-on value

Metrics movers: Features that will move your business goals: revenue, unit sold, NPS, user base, engagement, traffic etc. Very very few features actually fall in this category.I think less than 10% of total features can be metrics movers. It’s just a rare case that a few features can really move your business, particularly for complex consumer products. I don’t have the data, but I feel Wechat’s push-to-talk and import QQ contacts are two features that really move the business. Many my friends join wechat simply because they don’t have to spend money on SMS.  Metrics movers are not necessarily what your customer wants. For instance, you might need to do a push notification feature to increase DAU or a data tracking feature(even its hidden) to measure your metrics. Most of time, customers don't like this type of features.

Customer requests: Features that your customers want / are waiting / requesting, etc. For example, sending payment in iMessage is one of these features. I think around 80% of features falls into this category. You don’t have to implement every feature that customer requests. Figuring out which feature(s) they want most is the key. Ruthlessly prioritize features in this bucket, and do your best job to ship those features.

Add-on value: Features that your customers don’t know they want, but will wow/surprise if they see them. 5~10% features are this type. I find features in this category are usually experimental ones. Engineers build some cool things and later on realize that people actually actively requesting them. Xiaomi (a smartphone manufacturer) has a nice feature called “Mi Roaming”. It allows you to connect mobile data network while roaming without swapping SIM cards. The first time I used that feature in HK, I was completely blown away. But for sure I can live without this feature.

Conceptualizing your features this way has several merits:

1) it forces the team (product, sales& marketing, engineering, etc.) to be transparent and honest about why they are doing a certain feature. Is it because it can improve business results, or is it because your customer wants it or is it because it is just cool?

2)It softens the debate when a group of people strongly like a feature and another group strongly dislike - OK. let’s experiment it and see if customer really wants it or just thinks it’s cool.

3) It helps product roadmap planning. If you have too many add-on value features, you are not listening to customers and are probably too excited building somehing that only resonates yourself; if you have no add-on value features, you are probably too conservative.