Product Prioritization Framework

Often an organization argues about what the feature priorities should be. There is a simple framework that you can use to sort features in a backlog. This works best when you have a large backlog of features and limited engineering bandwidth. (So basically everyone)

Backlog items

A backlog item is usually a story or epic (collection of stories). It has a general (if not detailed) description. It typically has sketches or ideas, but not a design spec. In other words, it can be barely more than a vague goal.

Step 1: Start a Google sheet and put your backlog items in column A. You don’t need to put them all. 50 is fine. (If you want to jump to the end, there is a spreadsheet all prepped for you.)

Warning: Do not be tempted to put in tons of details. It will just make people avoid reading. People skim.

Value Drivers

The next thing is come up with value drivers. These are things that drive value in your organization depending on what the company goals are. You should have about half a dozen drivers. One example is focused on product metrics:

  1. Adoption
  2. Retention
  3. Network Effect
  4. Monetization
  5. Strategic
  6. Paying Tech Debt

Another example would be to list constituencies:

  1. Product Management
  2. Design
  3. Support
  4. Sales
  5. Marketing
  6. Customer Success
  7. Engineering
  8. Users (if available)

I actually prefer the second set. The reason is that people often just want to vote for their favorites and it’s easy to understand.

Put these labels as column headers in Row 1 Cols E-K.

The Scale

For each value driver, you come up with a 0-4 scale. You have to sit with people and define what each number means. For example, if the driver is sales then you might have a 0 = no additional revenue and a 4 = a major increase in sales. Give it a concrete number if you can.

You will likely need to negotiate with each constituency to make sure the scale makes sense to them. This is the hardest and most time consuming part. Once it’s done, you should review it once or twice a year to adjust.

Put this all on a separate sheet called Legend.

Enter the Numbers

You don’t need to be all in the same room at the same time. You can actually just have small meeting with stakeholders and let them fill it in for the 50 backlog items. Make sure there isn’t something crucial missing in the backlog. This typically takes a full day because people often want to ask questions about the item.

Often people will say, “If it’s done well, then it’s a 4 if it’s done poorly, then its a 2.” In that case, you may want to duplicate the row and put in both values. This will help you in multiple ways when it comes time to execute.


Next step, you need to sit with design and engineering and play Planning Poker. This is basically asking each person how much effort the backlog item will require and averaging the results. The process raises great conversations about the approach and risks. You can use “Ideal person days” which is how long a good engineer would take to do it assuming no distractions or other scales like the Fibonacci Sequence. Obviously, if the row is the easier or harder version, it should have different effort estimates.


Add a row at the top above the drivers. Above each driver enter 100%. This represents the baseline value of the scale. However, sometimes the company requires more focus on one value over another. You can increase these numbers to 200% to weight the results.

OK, we are almost there.

Return On Investment

Next, you make a column at the right called ROI. It is each driver multiplied by the top multiplier and then averaged. Finally, it is divided by the investment. Notice how the multiplier will change the results of the ROI if you put your thumb on the scale.

Finally, sort by ROI (high to low). The items at the top should be prioritized.

Defending your priorities

You should have a meeting to show the priorities regularly, at least once a month. If someone argues why their pet project is not higher on the list, then you can show the values they were given and the scale.

People may be tempted to increase their pet project numbers, despite the legend being as objective as possible. This is fine. They get their vote. But they also see how it weighs for everyone else.

The key here is to show empirically to everyone how they are not the only department in the whole company and that there are more values than just the one they are focused on.

View Spreadsheet

Here is the framework with the formulas built in.


  • Avoid the temptation to make it more complicated
  • Don’t make 20 value drivers
  • Don’t add in lengthy descriptions
  • Meet with small groups of 4-5, not 10+
  • Save a copy and redo the whole thing 1-2 times a year

If there is any part of it that doesn’t make sense, please ask. I’d be happy to help.

Whatya think?