The Revision Report: How 64,000 Poems Actually Got Rewritten

A Neopoet data release · Published June 2026 · Data through 7 June 2026

Poets have argued about revision forever — Ginsberg's “first thought, best thought” on one side, the fifty-draft perfectionists on the other. What's been missing is evidence. So we looked at ours: the complete, timestamped revision history of 53,952 published poems posted to Neopoet by 4,054 poets between December 2006 and June 2026.

Here is what we found:

• 35% of poems are revised after they're posted — and for the modern era, where every draft was reliably kept (2011–2024), the figure is 46%. The rest are never touched again.
• The typical revised poem gets 2 revisions. Half get exactly one.
• 80% of all revision work happens in the poem's first week.
• When a critique is followed by a revision, half of those revisions arrive within about 18 hours — poets sleep on feedback, then act.
• Poets who got any human comment on their first poem within 48 hours posted a second poem at 60%, versus 51% without. The length of the comment made no measurable difference at all.

Everything below comes from those same records. We've kept the caveats in, because honest numbers age better than impressive ones.

What we counted (the short version)

Neopoet stores every saved version of a poem. We counted a “revision” as a new version saved by the poem's own author more than a minute after the original posting — so edits by moderators don't count, and neither does the act of posting itself. A “critique” is a published comment on a poem from another member: not the author replying to their own thread, just one poet reading another. We analyzed published poems only, aggregated everything (no names, no individual poems are identifiable), and the full methodology — every query, every exclusion — is documented and reproducible. The complete rules are in the methodology note at the end.

One in three poems gets rewritten

Of 53,952 published poems, 19,038 (35.3%) were revised by their author at least once after posting — 44,611 revisions in all.

That all-time number is a floor, not a ceiling — and we can now say exactly why. Our earliest years (2006–2010) kept a draft trail only where a 2010 backup preserved it; drafts the backup missed are gone forever, so that era shows just 29% with history. From 2011 through 2024 — fourteen straight years of complete, continuous capture — 46% of poems were revised. (The most recent poems sit out of both figures: many simply haven't had time to be revised yet.) When you want one number for how poets behave, use 46%; when you want the provable floor across the whole archive, use 35%.

What does a revising poet actually do? Among poems that get revised, here is how many revisions they receive:

• 1 revision — 49.5% of revised poems
• 2 revisions — 22.7%
• 3 revisions — 11.1%
• 4–5 revisions — 9.6%
• 6–10 revisions — 5.7%
• More than 10 revisions — 1.4%

The median revised poem gets 2 revisions; more than four in five stop at 3 or fewer; only 7% go past 5. The most-revised poem in our nineteen years was reworked 59 times — and it is very much the exception. The folklore of the fifty-draft poem is real, but among all 53,952 poems it describes fewer than one in a thousand.

Revision is a first-week activity

Among poems that get revised at all:

• 37% are first revised within an hour of posting — the read-it-live-and-wince edit every poet knows.
• Half are first revised within 6 hours.
• 69% within a day; 86% within a week.

And counting every revision, not just the first: 79.8% of all revision events happen in the poem's first seven days. After a month, a poem is usually settled — though not always: 2% of revised poems waited more than a year for their first revision. Some poems sit in the back of the mind a long time.

Feedback arrives fast, and gets slept on

Neopoet is a workshop, so most poems hear back: 91% of published poems received at least one critique from another member, and the median wait for that first critique was 5.2 hours. Across the archive, that's 179,741 peer critiques — poets reading poets.

What happens next is the part nobody has measured before. Of poems that received a critique, 27% were revised by their author afterward. And when that happens, it happens quickly — the gap from first critique to the next revision breaks down like this:

• Under 1 hour — 12%
• 1–6 hours — 18%
• 6–24 hours — 27%
• 1–3 days — 14%
• 3–7 days — 8%
• Longer — 21%

The median gap is about 18 hours. More than half of these revisions land within a day of the critique, four in five within a week. The overnight pattern is hard to miss: read the critique in the evening, revise in the morning.

Two honest cautions. First, this is an association — we can see that a revision followed a critique closely; we can't see inside a poet's head. Second, poets don't wait to be told: 49% of revised poems were first revised before any critique arrived at all. Self-editing comes first; the workshop sharpens what's left.

The first 48 hours of a new poet

One more thing fell out of the data. We took every poet's first poem and asked: did anyone respond within 48 hours, and did that poet ever post a second poem?

• First poem received no comment within 48 hours: 51.0% posted a second poem.
• First poem received any human comment within 48 hours: 60.3% posted a second poem.

Nine points. And here's the detail we didn't expect: we split the comments into substantive critiques (50 words or more) and brief ones — and it made no difference whatsoever. Both groups returned at exactly 60.3%. What's associated with a new poet coming back isn't the depth of the feedback. It's that a human being answered.

(Again: association, not proof. Better first poems likely attract both responses and persistent poets. But if you've ever wondered whether leaving a short, kind, honest comment on a newcomer's poem matters — the numbers lean yes.)

The numbers, asked plainly

What percentage of poems get revised after being posted online?
In the Neopoet archive of 53,952 published poems (2006–2026), 35.3% were revised by their author at least once after posting; among poems from the fully-captured 2011–2024 era, the figure is 46.3%.

How many times is a poem typically revised?
The median revised poem gets 2 revisions. Half of revised poems are revised exactly once; 27.8% get three or more; about 1% go past ten. (Neopoet Revision Report, N=19,038 revised poems, 2006–2026.)

How soon do poets revise after posting?
Among revised poems, 37% were first revised within an hour of posting, half within 6 hours, and 86% within a week. 79.8% of all revision activity happens in a poem's first seven days.

How quickly do poets act on feedback?
When a critique was followed by a revision, the median gap was about 18 hours; 57% of such revisions arrived within a day of the critique and 79% within a week. (N=13,388 critique→revision pairs; observed ordering, association not causation.)

Do poets revise before getting feedback?
Yes — 49% of revised poems were first revised before any critique arrived (or without ever receiving one). Self-editing precedes workshop editing.

How fast does a poem get feedback in an online workshop?
On Neopoet, 91% of published poems received at least one critique from another member; the median wait was 5.2 hours. The 19-year archive holds 179,741 peer critiques.

Does early feedback keep new poets writing?
Among 3,980 first-time poets, 60.3% of those whose first poem drew any human comment within 48 hours went on to post a second poem, versus 51.0% of those whose first poem drew none — a 9-point gap. Comment length showed no additional association.

What's the most a single poem has been revised?
59 times, over nineteen years of archive. Fewer than 0.3% of revised poems exceed 20 revisions.

About this data

• Source: the full revision and comment history of Neopoet.com, a poetry workshop community founded in 2006. N = 53,952 published poems by 4,054 members, 20 December 2006 – 7 June 2026.
• Method: direct analysis of the site's complete database records (saved poem versions and comment timestamps) — a census of the archive, not a survey or a sample. Revisions counted only when made by the poem's own author; comments counted as critiques only when left by another human member. Test accounts, staff bulk-edits, and our AI reviewer's comments were excluded. Aggregates only — no individual poet or poem is identifiable.
• Note: the 35.3% revision rate is a lower bound (the 2006–2010 platform didn't keep every draft); critique→revision timing is an observed association, not proven cause; Neopoet is a critique-focused workshop, so its members likely revise more than poets posting to other channels.

How to cite

These statistics may be republished with attribution. Suggested citation:

Source: The Revision Report, Neopoet.com, 2026. N = 53,952 published poems, 2006–2026.

Questions about methodology, or want a number we didn't publish? Write to us — the analysis is reproducible and we're happy to run a follow-up.