How to say no to a feature request (with replies you can copy)

Every feedback board gets requests you won't build. This guide shows how to decline them clearly and kindly, drawing on founders and open-source maintainers who've done it in public. It includes replies you can copy for four kinds of no, and what to do when someone pushes back.

So many feedback boards have one: a request that's been open for a year or two, with a few dozen votes and a long comment thread. Everyone on your team knows it won't be built, but nobody wants to say so.

Leaving it open can feel like the right thing to do. But the people following it are still waiting, and the longer they wait, the more the answer stings. In one long-running Godot engine proposal, a user thanked a team member for a straight reply in words many requesters would recognize: "Transparency is all I ever asked for."

This guide starts once you've decided the answer is no. (If you're still weighing it up, start with how to manage feature requests.) It covers what a good decline includes and four kinds of no, each with a reply you can copy. The last section deals with pushback.

Why saying no is part of the job

Some of the best-known product teams are proud of how often they say no. 37signals, the company behind Basecamp, put it bluntly in its 2006 book Getting Real: "Every new feature request that comes to us — or from us — meets a no." In 2024, co-founder David Heinemeier Hansson made the same case on the company's podcast: "You get a great product by saying no to almost everything almost all the time."

The reason is cost. You build a feature once, but you support it forever. 37signals compares every yes to "adopting a child." Open-source maintainer Jeff Geerling gives the small-team version in why he closes pull requests: "I will not add something that increases my maintenance burden unless it's very compelling functionality or an obvious bugfix."

Popularity doesn't settle it either. In his 2013 essay Product strategy means saying no, Intercom co-founder Des Traynor lists the arguments teams use to talk themselves into features, like there being a high number of people asking for one. Votes show demand for something, but whether it fits your product or not is a separate question.

When nobody answers, users notice. In a 2026 study of how the Moodle community handles feedback, one user told researchers that Moodle's developers "are responsive for bugs; for feature requests they are not much."

Say no to the idea, not the person

GitHub's Open Source Guides for maintainers have one line that sits underneath every good decline: "rejecting a contribution is not the same as rejecting the person behind it."

You can hear the difference in its two example replies: "Your contribution doesn't match this project's criteria" is about the product, and "I don't like your contribution" is more about the person.

Here's what it looks like in practice. In 2022, a user asked the maintainer of Serilog Timings, a small .NET logging library, to add checkpoints and reports for slow operations. Nicholas Blumhardt said no and explained why: "the goal currently is to keep it ultra-simple." He suggested a separate library instead, perhaps forked from his, and offered to link to it.

The requester's reply was "Perfectly fine." A clear decision, a reason about the project and somewhere else to go. Nobody had to pretend the idea was bad.

What a good decline includes

You won't always need all six, but a good decline usually has most of them:

1. Thanks for the specific problem. Thank them for what they told you about how they use your product, not for an "amazing idea" you're about to turn down. Geerling says he usually starts by "thanking the contributor to try to soften the blow."

2. A clear decision. Des Traynor of Intercom’s rule is "If you're building a product, you have to be great at saying no. Not 'maybe' or 'later'." Write "We've decided not to add this," not "We'll keep it in mind."

3. The real reason. Usually that's scope, or the upkeep a feature brings with it. Jon Parise has a useful test in GitHub's ReadME Project: "if you're not able to explain the reasoning behind your decision, it probably deserves some more thought or another point of view." Don't blame team capacity if the feature would still be wrong with unlimited engineers.

4. An alternative, if there is one. Point to a setting or another tool that solves the problem, and be upfront about where it falls short.

5. The decision on the request itself. Reply on the post, not in a private email, so everyone who voted sees the same answer. (In Fider, you set the status to Declined and add your reason as the response, and people following the post are notified.) It's part of closing the feedback loop.

6. What would change your mind, only if something would. When an idea has potential but won't work as proposed, Parise's advice is to respond "by suggesting what it would take for it to become acceptable." If nothing would, leave this out.

What to leave out

Some phrases do more harm than good:

  • "Maybe later," when the answer is no. It keeps hope alive for something you've already ruled out.
  • "We've passed it to the team," as the whole reply. That's an acknowledgement, and it’s a bit too vague.
  • "Nobody needs this." Talk about what your product is for, not whether their problem is real.
  • "PRs welcome" (pull requests, i.e. code contributions), when you'd turn the code down. Inviting someone to write code you won't accept just wastes their time. Former Docker maintainer Jessie Frazelle's advice is direct: "If you know there is no way you will ever accept their patch, close it right then.
  • "Get more votes and we'll look again," when the objection is about fit. More votes won't change what your product is for.

On one collaboration tool's community forum, a user complained of years of "copy/paste pleasantries and stalling." A company representative replied that "all feedback shared here is regularly reviewed by the product team." The user's answer: they were moving their 12-person team to another tool.

Is "not now" a cop-out?

Here, the experts disagree. At 37signals, the first answer to every request is "not now," and "If a request for a feature keeps coming back, that's when we know it's time to take a deeper look." They rule out "later" altogether.

Both can be right, depending on whether "not now" is true. It works when you'd reconsider, but not as a polite way to put off saying no.

Godot shows what happens when "not now" stretches too long. For years, users asked for Python-style list comprehensions in GDScript, the engine's scripting language, and were told to wait. In December 2023, Adam Scott closed the proposal for the GDScript team: "The GDScript team doesn't believe, right now, in investing in list comprehensions. The arguments haven't convinced us."

It's a good decline. He explained that the team would have to support the feature "as long as the language is used," and apologized: "I'm sincerely sorry for the time you spent waiting for a response." But the proposal's author had already written that "our voices aren't being heard." A good no can't undo a long wait.

A no doesn't have to be forever, either. In 2020, Plausible Analytics closed a request for heatmaps as outside "the current scope of the project." Three years later, a user asked, "Since it's been almost 3 years of this, can it be reconsidered?" Co-founder Marko Saric replied in 2024 that there were "no short term plans to work on this," but added it to the team's new feedback board to gauge demand.

Your answers can change as your product evolves. Just make sure each post's status says what's true today. (Learn more about how statuses help users in feedback board vs roadmap vs changelog.)

Four kinds of no (with replies you can copy)

Declines tend to fall into one of these four types. Each comes with a reply you can adapt by filling in the [brackets].

1. Not ever: it doesn't fit the product

Use this when the feature would change what your product is for. When a player asked Factorio's developers to make automated farming towers plant in a set order, one of them replied in a single sentence: "Towers were specifically made to plant in random spots so I don't see this ever changing." It's blunt, but nobody's left wondering.

Thanks for explaining how you'd use [feature]. We've decided not to add it. [Product] is built for [core use], and [feature] would [change how it works for everyone / add a lot of upkeep]. We're marking this as declined so everyone following it has a clear answer.

2. Not now: it's possible, just not soon

Only use this if you'd reconsider. Don't give a date you can't keep.

Thanks for describing [the problem]. We'd like to solve it, but we're not scheduling it right now because we're focused on [current priority]. We don't have a date, so we'll keep this open and update it here if that changes.

3. Not us: it's better solved somewhere else

This is where you point the person to where their problem can be solved. For an open-source product, that might include a fork.

Thanks for the suggestion. We don't plan to build [feature] into [product], because [reason]. [Tool or integration] is a better fit for this, and [how it connects / link]. We're marking this as declined, but let us know if you have trouble connecting the two.

4. Already possible: there's another way

When someone asked Bootstrap for more table display utilities, co-creator Mark Otto pointed them to an existing feature: "The utility API can be customized to add these."

Thanks for raising this. You can already do this with [setting or workaround]: [short how-to or link]. It doesn't cover [limitation], so if that's a problem for you, tell us more about your use case here.

When they push back

Sometimes the requester comes back unhappy. A few things help:

  • Look for new information. A follow-up can show that you misread the problem. If it does, take it back to the team.
  • Separate the decision from the alternative. Are they objecting to the no, or to the workaround you suggested? The workaround might be easier to fix.
  • Don't promise on the spot. DHH is firm about this: "Under no circumstances should you allow yourself in a customer interaction to get drawn into on the spot a commitment."
  • End it kindly. GitHub's guide puts it in four words: "Be kind, but firm." Restate the decision and reason once. If it turns hostile, that's a job for your board's rules, not more debate.

Here's a reply for when the decision hasn't changed:

I understand that [workaround] doesn't meet your need for [need]. Our decision is still not to build [feature], because [reason]. If you have an approach that avoids [the problem], we're open to reviewing it. Otherwise, this request will stay declined.

Only include the invitation if you'd take a look. Otherwise, stop after the reason.

A good no won't make everyone happy. Some people need the feature you've declined, and no wording will change that. But a clear answer gives them something to act on, which an open request that never moves can't.

With any luck, with your next “no”, you'll get that nice reply: "Perfectly fine."