How to rank feature requests using customer evidence
Learn how to rank feature requests by grouping customer needs and weighing account reach, pain, revenue evidence, and fit with your product strategy.
Free app. Runs on your Mac with the Claude or Codex plan you already have.
What a method to rank feature requests is
To rank feature requests, gather customer feedback from the channels you use, group requests that describe the same problem, and compare the evidence behind each distinct need. A request for custom roles and a ticket about giving a bookkeeper limited access can point to the same problem. A useful ranking shows how many distinct accounts reported it, what happens when they encounter it, and how it fits your product strategy. Each row ends with a recommended action whose rationale you can trace to the original requests.
Feature request analysis starts with what customers are trying to do and where their work gets stuck. Count people and accounts separately because they answer different questions. Several people at one company can show that a problem disrupts a whole team while still representing one account. Use the ranking to inform feature prioritization and bring the leading needs into product roadmap planning. A strong need may warrant a build brief even when the team still has to decide what to build and how much work it will take.
What it does for the business
- Put engineering time behind explicit customer problemsA named problem and the workflow it interrupts give a product team a concrete starting point for a build brief. The team can evaluate a smaller change that solves the immediate problem before committing to a broad feature.
- Follow up on deals and renewals with better contextWhen a request names a deal or renewal that depends on a capability, the account owner can check the condition and its timing. Product and sales can then discuss what would help the customer and which commitments the team can responsibly make.
- Find recurring pain hidden across channelsA workaround described in support and a buying obstacle recorded by sales may turn out to share the same cause. Bringing those records together gives the team a chance to address a need that each department sees only in part.
- Spend research time on decisions it can changeA promising need with one missing piece can become a targeted customer interview. Naming the unresolved issue lets the team recruit someone who has encountered the problem and collect evidence for a specific roadmap decision.
How to build a method to rank feature requests, step by step
Bring requests into a traceable table
A count starts with knowing who said what and where the report came from.
What to do
- Set a reporting period and record the product goal and customer segments that matter for the decision. Note work already planned and constraints that affect what the team can take on.
- Export feedback board entries, support tickets, sales notes, and any spreadsheet your team uses to log requests. Note which channels cover the whole period and which provide only a partial view.
- Give each row a stable request ID and keep the date, source, original wording, person, account, and plan where available. Check that common spelling variations for the same account resolve consistently before joining revenue data.
- If a feedback board lists individual voters, preserve them as identifiable rows. If it provides a total only, keep that total separately so a reader can distinguish recorded votes from known people and accounts.
- State which account revenue field you use and keep missing values visible. If you use monthly recurring revenue, the amount an account pays for recurring service each month, keep one consistent definition across the account data.
Why
Different sources show different parts of the experience. Support records often describe the workaround, sales notes may explain a purchasing condition, and a feedback board captures the ideas customers can see and vote on. A common time window and references back to the source make it possible to compare the evidence and recognize where coverage is uneven.
Group requests by the underlying need
Requests belong together when solving one underlying problem would serve both.
What to do
- Read each request in context, including the task the person was trying to complete. Use repeated words to find candidate duplicates and look across groups for different phrases that describe the same situation.
- Name each group as a need in plain language, such as giving outside bookkeepers access to invoices with limited permissions. List any requested features separately so the team can consider other ways to meet that need.
- Split groups when shared words hide different outcomes. An export for opening records in a spreadsheet and an export for restoring a backup can have different requirements for format, completeness, and reliability.
- Record each merge or split and preserve the source request IDs. Flag requests whose intent remains vague so a reviewer can see where the grouping rests on an unresolved interpretation.
- Recalculate mentions, named people, distinct accounts, and known account revenue after the groups change. Count an account once within each need even if several people at that account raised it.
Why
Grouping changes what the team sees. Broad categories like reporting can combine requests that require unrelated work. Narrow labels based on proposed features can split one recurring obstacle into several low-ranked ideas. Keeping the original wording lets teammates review a decision and work backward from a need to the evidence it contains.
Rank the needs and recommend an action
Reach tells you who encounters a problem, while pain tells you what the problem costs them.
What to do
- Compare account and person counts alongside account plans. Look for independent accounts describing similar problems and note when the evidence comes mostly from one segment.
- Rate pain using a direct quote and the consequence it describes. A task that blocks a deal or risks data loss deserves different treatment from a preference, and a recurring manual workaround can be consequential even when customers describe it calmly.
- Record the known monthly revenue associated with requesting accounts and capture any deal or renewal explicitly tied to the need. Keep the connection to the request and treat unknown revenue and effort as unknown.
- Rank by reach and pain first, then consider named revenue and fit with the stated goal. Recommend building when several accounts describe strong pain, validating when scope is unclear, and deferring when pain is light. Mark a vague one-off as insufficient evidence. Give each recommendation a confidence level.
- Challenge the top three by naming what could overturn each recommendation and which lower need would move up. For a need you would validate first, specify the person who could answer the open question and what you need to learn from them.
Why
A ranked table earns trust when the team can explain its trade-offs. A high-reach problem and a severe issue at a smaller set of accounts may both deserve attention for different reasons. Keeping evidence beside the recommendation makes it easier to revisit the order when the team learns how widespread the problem is, estimates effort, or changes its goals.
Review the evidence and hand off the ranking
A teammate should be able to trace the top recommendation back to a customer record.
What to do
- Ask a teammate to trace account and revenue totals back to the grouped rows and look for the same account counted under different names or through different channels.
- Have the reviewer check that grouped requests share a need, each pain rating has a supporting quote, and every recommendation follows the ranking rule you set.
- Publish the coverage period, sources, and gaps alongside the table. Name the missing evidence that matters most, such as feedback from departed customers or account records without revenue.
- Lead the handoff with the strongest supported recommendation and its reason. Include the ranking, source IDs, and the decisions on merges so the product team can challenge or revisit a row.
Why
Ranking mistakes can survive a quick read because a plausible recommendation can sit beside a bad sum or a loosely related quote. A source check catches the defects that could change a decision. A clear handoff also lets the team tell which rows need follow-up before they belong in planning.
Common mistakes
- Treating every request as a different customerSomeone can raise the same issue on a call, in a ticket, and on a feedback board, so keep mentions separate from distinct people and accounts.
- Merging on shared keywordsCheck whether the requests describe the same task and obstacle before grouping them.
- Taking a proposed feature as the problem statementFind out what the customer is trying to accomplish before deciding that the feature they named is the right response.
- Treating account revenue as money at riskRecord the known revenue for context and require explicit evidence before labeling a deal or renewal as dependent on the need.
- Giving every row the same confidenceAccount identity, concrete examples, and consistent evidence across sources can justify different levels of confidence.
What a good result looks like
- A named decision, review period, list of sources, and clear gaps in coverage.
- One row per distinct customer need, with traceable source IDs and an explanation for disputed merges.
- Separate mention, person, and account counts, plus known revenue counted once per account per need.
- A pain rating supported by a quote and explicit entries for unknown revenue, strategy fit, or effort.
- Ranked recommendations with confidence and a specific condition that would change each leading recommendation.
And then what? Using the result
- Turn a build-next recommendation into a one-page brief that describes the users, the blocked task, and the outcome that would help them.
- Assign each validate-first need to a product team member who can contact the people with the strongest evidence and test the open question.
- Ask sales or customer success to confirm any stated purchasing or renewal condition before it influences a commitment.
- Share what you learned with the people who logged the requests and explain the decision in terms of the problem they raised.
- Record the decision in your product roadmap and keep the source references for the next review.
When to run it
- Before quarterly roadmap planningReview needs alongside the period's goals so the planning discussion starts with evidence and open decisions.
- When teams disagree about what customers needA common table lets support, sales, and product compare examples and see whether they are describing separate issues or the same obstacle.
- When new feedback changes a leading recommendationRecheck the order when a consequential new report arrives or another account confirms a pain your team had seen only once.
- After a strategy changeReassess fit when the target segment or product goal changes, and check that your sources represent the customers now in scope.
Questions people ask
How do you rank feature requests?
Rank distinct customer needs by account reach and the severity of the problem, then weigh named revenue and fit with your strategy. Keep the source evidence next to each recommendation so someone else can check the order and understand what might move a need up or down.
How do you merge duplicate feature requests?
Merge requests when they describe the same underlying obstacle and a shared outcome would help the people who raised them. Preserve the original wording and source IDs, then recalculate the people, accounts, and known revenue in the combined group.
Should you prioritize feature requests by votes?
Use votes as one indication of interest alongside evidence of reach, pain, and fit. A total without voter identities gives you a board count only, and feedback through support and sales can add context that a vote rarely includes.
How do you use revenue in feature prioritization?
Use known account revenue as context and record any deal or renewal explicitly tied to the problem. Count revenue once per account per need, and check that the same account's revenue is counted only once if you ever report a combined total across needs.
What do you do when a feature request has weak evidence?
Give the request a follow-up action that matches the gap. Ask someone who encountered the problem to describe a recent example if its consequences are unclear, and retain a vague anonymous request as a signal whose scope remains unknown.
How often should you review feature requests?
Review the ranking before roadmap decisions and when new evidence could change the order. Keep reporting periods and source coverage visible so you can tell whether a change reflects customer needs or a change in how feedback was collected.
More jobs in Product
Rank feature requests, on your own Mac.Free app. Runs on your Mac with the Claude or Codex plan you already have. Your files stay on your computer.
Download for Mac