Feature prioritization survey

Ask people to rate each feature from 1 to 5 and most of them get a 5. Nothing is ranked and you learn nothing. This template asks for a choice instead: the one feature they'd keep, the one they'd drop, and the one they'd pay more for. A choice costs them something, so the answer is closer to what they really want.

Best for: Product teams with a long backlog and room to build only two or three things next quarter.

Use this template free

1 of 7

Of these features, which ONE would you keep if you could only have one?

AFirst optionBSecond optionCThird optionDOther

Made with Enform

A live preview of this template. Click through it. Nothing gets sent.

The questions

1Of these features, which ONE would you keep if you could only have one?Single selectList your top candidates. Allowing only one pick is what turns this into a ranking.
2Which feature here is the least valuable to you?Single selectUse the same list as the question above. A best pick and a worst pick from the same set is the idea behind MaxDiff.
3What's the last thing that made you wish the product did something it doesn't?Long textAsk about a real moment, not a wishlist. People remember a specific afternoon better than a general need.
4Which of these would you actually pay more to get?Multi selectMoney separates the features that sound nice from the ones people need.
5If we shipped your top pick next month, what would it let you stop doing?Long textThis names the workaround the feature would replace. A painful workaround means a valuable feature.
6How are you handling that need today?Long textIf the answer is "I just don't", the pain may be smaller than it sounded.
7What's your role?Short textLets you split the ranking by role. Heavy daily accounts and new sign-ups often want different things.

Sample replies, counted

These replies are made up to show the method.

Asked: What's the last thing that made you wish the product did something it doesn't?

exporting to csv every friday so my boss can see the numbersManual reporting
Last week I rebuilt the same weekly report by hand for the third month in a row. It takes me about 40 minutes each time.Manual reporting
I wanted to give our contractor view-only access and the only option was full editor. So I didn't invite her.Permissions too coarse
Copying data into Slack by handMissing integration
we live in hubspot and there's no way to push a record over, so someone on my team copies it across every morningMissing integration
dark mode honestlyOther
On Monday I needed the dashboard numbers for a call and had to dig through three screens to piece them together.Manual reporting
Couldn't lock a finished project so nobody edits it by mistakePermissions too coarse

8 replies, one label each
Label	Replies
Manual reporting	3
Permissions too coarse	2
Missing integration	2
Other	1

Three of eight replies describe rebuilding a report by hand, and two more are about copying data into another tool. Check whether scheduled reports or a HubSpot sync wins the keep question before you commit to either.

How to run it well

Show the same feature list in the keep question and the drop question. The contrast between the two is what breaks the tie.
Keep each list to five to seven features. Longer lists get skimmed, and the picks turn random.
Ask which features people would pay more for, one by one. A price question about the whole product won't tell you which feature to build.
Split the results by role and by how long people have been with you before you decide. Your oldest accounts and your newest sign-ups rarely rank the same way.
Run it again after each big release. The backlog should be ranked against the product people use today.

When not to use this: Skip this if you don't have a shortlist yet. Collect raw ideas first with a feature request survey, then come back to rank the five to seven that keep coming up.

FAQ

Why not ask people to rate how important each feature is?

Because almost everyone marks almost everything as important, and you end up with a list of fives. Picking one to keep and one to drop forces a tradeoff, and the tradeoff is the information you need.

How many responses do I need to trust a feature ranking?

Enough that the top pick stops changing as new replies come in. If two features stay neck and neck, split the results by role first. The tie often breaks once you see who wants what.

Is this a real MaxDiff study?

No, it's a light version of the same idea. A full MaxDiff study shows many small sets of features and scores them with a statistical model. This template keeps the best and worst pick on one list, which is enough for most roadmap calls.

Where the method comes from

MaxDiff, Sawtooth Softwaresawtoothsoftware.com
5 Prioritization Methods in UX Roadmapping, Nielsen Norman Group (2021)nngroup.com
Writing Survey Questions, Pew Research Centerpewresearch.org