Question for fellow product owners/product managers. How do you prioritize your features? We are currently trying to develop some common criteria (as objective as possible) that we can use to prioritize our features. At the very least, we want something that will give us a good starting point and potentially things to look at when we get the inevitable question of why are we working on this instead of that. Does anyone have some good suggestions? How do you all address this problem?
I suggest the following three criteria: Value, Risk, Cost. Value can be further decomposed into two: Benefit and Penalty. Benefit and penalty combined create an overall value.
Benefit is the positive affects an added feature will deliver to the organization that can be objectively measured in some way, i.e., increased revenue or decreased costs, faster throughput, etc. The particular criterion should be as objective as possible.
Penalty is the more subjective criterion, where you will likely have stakeholder angst if a feature is not implemented. In other words, the value of a feature climbs if a stakeholder group is upset in its absence, despite that it offers no objective value to the organization.
Combining the two creates your overall value score.
Risk is specifically the uncertainty you must assume and mitigate in delivering the feature.
Cost is the direct project cost in the feature's implementation.
Ensure that all of your features are decomposed to the same level of specificity to make sure the scoring is precise.
I was tasked with the same thing. In the end we used a matrix with two axis. One axis was frequency of occurrence (ie Feature xyz would be used by every user every time they used the application vs. Feature abc would be used by some users sometimes, etc, etc). The other axis was level of benefit to the user/benefit to the company/ROI (ie Feature xyz increases customer satisfaction vs. Feature abc increases company revenue).
Combining benefit and occurrence turned out to the most objective way we could come up with a baseline for what was "High", "Medium", "Low", and "Don't Bother". After that we took a deeper dive and pass through all the "High" cases.
You'd be surprised how few cases make it into the "High" rating when you rate them objectively against all the others.
BTW, we ended up using a similar matrix for all of our bugs.