Skip to content
← Insights

Customer Requests for the Roadmap: How to Tell a Recurring Need from a One-Off Request

Learn how to record, group, and assess customer requests by problem, context, and evidence, without turning votes into automatic decisions.

Product team organizing customer requests and assessing problems, context, and impact for the roadmap

Customer requests arrive through many channels: a sales conversation, a support ticket, a renewal meeting, or a research interview. The challenge is not just to collect them, but to decide what they mean and what to do about them. A repeated request may point to an important problem, but it can also reflect a fashionable solution, a highly vocal segment, or a limitation in a single workflow.

To prioritize customer requests for the roadmap, separate observed demand from the product decision. The number of mentions is a signal; it is not enough to serve as a criterion. A useful assessment connects each request to who has the problem, when it occurs, what effect it has, and what evidence could establish its scope.

Why counting requests or votes is not enough

Why counting requests or votes is not enough

Counting requests is a good first step: it helps surface recurring themes and prevents the team from relying only on memory. However, a list ranked by votes can distort priorities. A large customer may have more contact with the team than many small customers; a feature that is easy to describe may get more votes than a hard-to-articulate problem. The same need may also be recorded under different names.

Volume does not show the cost of leaving a problem unresolved, either. Ten requests for a minor improvement do not necessarily outweigh a less frequent issue that blocks a critical operation. Even the number of affected accounts does not tell the whole story: it matters whether those accounts represent a strategically relevant segment and whether the problem prevents them from completing a valuable task.

So use frequency as a signal to investigate, not as an instruction to build. The decision needs context, scope, impact, and comparison with other opportunities, including technical or business obligations that rarely arrive as customer requests.

Separate the requested solution from the real problem

A request is often phrased as a solution: “add an export,” “we need another permission,” or “integrate with this system.” Do not dismiss it, but avoid copying it straight onto the roadmap. Ask what the person was trying to do, what happened, how they handle it now, and what outcome they expected.

For example, several requests for an export could reflect different needs: preparing a recurring report, sharing data with a vendor, or keeping a record for an internal audit. The requested feature seems the same, but the conditions of use, requirements, and alternatives may vary. Conversely, requests phrased in different ways may point to the same obstacle in a workflow.

A brief conversation can clarify the issue. Ask: “What task were you trying to complete?”, “What happens if you can’t do it?”, and “How do you handle it today?” Record the answers in the customer’s own words whenever possible. The problem should describe an observable difficulty, not a solution chosen in advance.

Record the usage context and operational impact

A request without context loses value when it is reviewed again weeks later. Each record should help someone else understand who encountered the problem, in what situation, and with what consequences. The record need not be extensive, but it should follow a consistent structure.

  • User or segment: role, type of organization, or affected group. Do not assume one person represents the entire segment.
  • Task and situation: what they were trying to achieve and where in the journey the difficulty arose.
  • Frequency: whether it happens once, occasionally, or during every relevant use. Distinguish observed frequency from estimates.
  • Effect: extra time, errors, manual work, a blocker, operational risk, or task abandonment.
  • Current workaround: a manual process, external tool, change in habits, or no solution.
  • Source and evidence: channel, date, conversation notes, ticket, or reproducible example, while following applicable privacy requirements.

If the impact is unclear, mark it as unknown instead of filling the gap with an assumption. That uncertainty is useful: it indicates which research question could change the decision.

Group similar requests without erasing differences

Grouping helps reveal patterns, but “similar” does not always mean “the same.” Group requests by the problem and affected task, not just by the words used or the proposed solution. Keep the original records linked to a shared description so that you can revisit individual cases when different requirements emerge.

When consolidating requests, check three things: whether users are trying to achieve a comparable outcome, whether they face a similar barrier, and whether a potential response could help them without harming others. If a request seems to fit a group but differs in context or the cost of its workaround, record it as a variant. Do not force everything into one category just to make the numbers look clearer.

It is also worth avoiding duplicate counts of the same case: a conversation might appear in sales and later become a support ticket. Keep a reference to the source and the customer or situation, with access limited where sensitive data is involved. The aim is to identify recurrence without inflating volume.

Assess scope, impact, and evidence

You do not need a complex mathematical model to make an initial assessment. Compare each problem against explicit criteria, and record what is known and what remains unknown. A practical review can consider:

  • Scope: which segments or tasks may be affected and how much evidence supports that estimate.
  • Impact: the severity and consequences of the problem, not the appeal of the proposed solution.
  • Strategic relevance: how the issue relates to agreed product and business objectives.
  • Confidence: the quality of the evidence. A detailed account, a pattern across different channels, and an unvalidated hypothesis do not carry the same weight.
  • Cost and dependencies: the effort required to investigate or implement, the risks, and the work that would need to move aside.

These criteria do not produce an automatic truth. They make the reasoning visible and help compare alternatives with less bias. If a need seems high-impact but is supported only by a vague description, the sensible decision may be to investigate before planning. If the problem is well established and an existing alternative solves it, a new feature may not be necessary.

Choose an action and communicate it

A request does not always have to end with “build” or “reject.” Classifying the next action helps set expectations and makes decisions easier to revisit. You can use four outcomes:

  • Investigate: it is not yet clear who has the problem, how often it occurs, or what consequences it has. Define the question and the evidence that could change the decision.
  • Address it with a workaround: an existing setting, process, or capability may meet the need. Check that the workaround is acceptable in the specific context.
  • Plan: there is enough evidence, the opportunity aligns with objectives, and it has been compared with other work. Adding it to the roadmap expresses intent, not a guaranteed date, unless a date has been explicitly committed to.
  • Do not prioritize or decline: the need is not a fit right now, its impact is limited, or there is a more valuable option. Record the reason and the new evidence that would justify reviewing it.

Communicate the decision to the person who raised the request and, where appropriate, to the customer. Explain which problem you understood, what was decided, and why, without promising delivery that has not been agreed. A clear response makes it possible to correct a misunderstanding and prevents the same case from being raised again as if it were new.

A short template for assessing a request

A short template for assessing a request

A shared template for product, sales, and support reduces the loss of context. You can start with these fields and add only what your process requires:

  • Original request: the customer’s words or a faithful summary.
  • Problem and task: what they were trying to do and what prevented them.
  • User, segment, and context: who, when, and under what conditions.
  • Frequency and scope: observed cases, affected segments, and the limits of what is known.
  • Impact and current workaround: the consequences of the problem and how it is handled today.
  • Evidence and uncertainties: available sources and open questions.
  • Decision, owner, and review: investigate, offer a workaround, plan, or do not prioritize; who will follow up and what signal would prompt a review.

Review grouped requests at regular intervals and when conditions change: new evidence emerges, a segment becomes more relevant, or a workaround stops working. This keeps the record from becoming a dead archive. The goal is not to reward the most popular request, but to better understand problems and make decisions that can be explained using the available evidence.

Fuentes y referencias

  1. Web standardsW3C
  2. OWASP Cheat Sheet SeriesOWASP Foundation
  3. Web performanceweb.dev