State of Feature Requests 2026

What happens to a request

1. On the median public feature-request board, only 22% of requests are marked as shipped (n = 220 boards).
2. On the median board, 62% of feature requests are still open with no public decision at all, neither planned, in progress, shipped nor declined (n = 220 boards).
3. Only about 2 in 100 requests on the median board are marked planned (1.3%) or in progress (0.8%) (n = 220 boards).
4. Teams rarely say no in public: the median board has declined just 2.8% of requests, and 25% of boards have never declined a single one (n = 220 boards).

How long they wait

1. The median unresolved feature request has been waiting 515 days, about 17 months (n = 181 boards).
2. On the median board, 70% of unresolved feature requests are more than a year old; across all 50,401 unresolved requests, about one in four is older than three years (n = 181 boards).
3. When a request does ship, it takes a median of 165 days, about five and a half months, from the day it was posted (n = 120 boards).
4. About one in five shipped requests ships within a month, and one in three takes more than a year (each board counted once: 22% within 30 days, 33% after more than 365 days; n = 120 boards).

Where the votes go

1. On the median board, the top 10% of requests hold 42% of all votes, and the top half hold 84% (n = 176 boards).
2. On the median board, 37% of feature requests have one vote or fewer, usually just the author's own (n = 182 boards).
3. On the median board, half of feature requests (51%) never receive a single comment (n = 182 boards).
4. About one request in five has at most one vote and zero comments (20% on the median board; n = 182 boards).
5. Votes predict shipping, but weakly: 33% of the most-voted tenth of requests are shipped, against 23% of all others (n = 182 boards).
6. The median board has shipped only 4 of its 10 most-voted requests (n = 182 boards).

Size changes everything

1. Boards with 1,000+ requests have shipped a median 12% of them and leave 78% undecided, while boards under 100 requests have shipped 32% and leave 50% undecided (54 and 52 boards).
2. Unresolved requests on boards with 1,000+ requests have been waiting a median 1,129 days, more than three years, against 328 days on boards under 100 requests (19 and 51 boards).
3. Feedback keeps arriving: 68% of boards received at least one new request in the last 30 days (n = 242 boards).

How we did this

• Data sources: public boards hosted on Canny, Featurebase and Nolt. No result is broken out by platform.
• Sample: 242 public boards out of 297 candidates. Candidates were every board address on those platforms found during our own research into software companies in October 2026, listed before any result was measured.
• Included only boards readable without logging in and with at least 10 public requests. Excluded 55: 27 with no public board, 25 with under 10 requests, 3 private or missing.
• One board per company (the feature-request board where a company has several; bug and support categories removed).
• Collected on 1 November 2026 (JST) from the same public data each board page loads, one request per second, no logins, no personal data.
• Every public request was collected on 182 boards. On 60 very large boards the platforms return only part of the list: 38 of them publish exact counts per status, which we used for status shares; for the other 22 no exact count per status exists, so they are left out of status shares (220 boards) instead of being estimated. Vote and comment figures use the 182 fully collected boards; age figures use the 181 of them that have at least one unresolved request.
• Statuses were grouped into five buckets: no decision, planned, in progress, shipped, declined. Unresolved = no decision + planned + in progress.
• Each board counts once. Figures are the median across boards, and the time-to-ship chart averages each board's own shares, so a few huge boards cannot drive any number.
• Time to ship: completion dates published by the board, or the date the status changed to complete on up to 5 random completed requests per board. 120 boards had 3 or more dated completions (3,635 completions; 8 dated before their request were dropped); the headline, finding 8 and chart 3 all use these 120.

Limits

• Companies with private boards are not here, and the sample leans to early-stage and mid-sized SaaS. It is not a random sample of software companies.
• Statuses are what teams publish. Some requests marked open have surely shipped without an update, so shipped shares are a floor.
• Time to ship counts only requests that shipped, so the real wait for a typical request is longer.
• One vote usually means the author's own vote (the platforms' default); we did not verify every board.
• Some boards were imported from other tools, so a status-change date can be the import date; this affects the time-to-ship figures (8 completions dated before their request were dropped).
• Large boards are under-represented in the vote and age figures: only 19 of 65 boards with 1,000+ requests were fully collected.
• Custom status names were mapped to the five buckets by name.
• Our July 2026 graveyard audit looked only at the top-voted requests of large, well-known boards, so its numbers are not comparable one to one with this study.
• A single snapshot. We plan to repeat it each year at the same URL.

Data

Headline metrics

By board size

Download

• data.csv: one anonymised row per board (random id, request-count band, derived shares and medians only).
• data.json: all aggregates and definitions.

License: CC BY 4.0.

Cite this research

Writers and researchers are welcome to use any chart or number with a link to this page. Questions: max@featq.com