How Many Words a Day Should You Write?

In 2018, NaNoWriMo asked 287,327 people for 1,667 words a day. When the month ended, the organisation counted the words itself. The average participant who started a novel had written 22,871 of them.

That is 762 words a day, and it is the average among people who had signed up specifically to write 1,667.

The useful question, then, is not what a writing coach recommends. It is what people who genuinely try manage to produce, and what the arithmetic of your own deadline demands of you. Those two figures rarely agree, and most abandoned manuscripts sit in the gap.

Where 1,667 comes from

50,000 divided by 30 is 1,666.67. That is the whole derivation. The number was chosen so that a month of writing ends on a round figure, not because anyone measured what a person can sustain.

NaNoWriMo published a “By the Numbers” post most years, and the figures are worth reading against the target. In 2013, 310,095 people took part, 42,221 of them reached 50,000 words, and the average across all participants was 11,352. In 2018, 287,327 took part, 35,387 finished, and the average among those who opened a project was 22,871.

The two averages are not directly comparable, because the second counts only people who opened a project rather than everyone who registered. Both land in the same territory: 378 words a day in the first case, 762 in the second. Neither is close to 1,667.

The organisation itself no longer exists. Interim executive director Kilby Blades announced its dissolution on 2 April 2025, citing six years of financial trouble, the handling of 2023 allegations against a volunteer, and the fallout from its 2024 position on AI (Reactor, 2 April 2025). The number outlived it. It is still the default in templates, in forums and in the spreadsheets that circulate with them.

As a four-week sprint target, 1,667 is defensible. As a year-round standard, it is a marathon pace applied to the walk to work.

The documented ceiling

Anthony Trollope wrote more than forty novels, many of them alongside a full-time job at the Post Office. His autobiography describes the method precisely: a watch on the desk, three hours before breakfast, the first half hour spent reading the previous day’s work. For the remaining time he required 250 words per quarter of an hour, which produced “over ten pages of an ordinary novel volume a day” (An Autobiography, chapter 15).

Two and a half hours at 1,000 words an hour is 2,500 words. Before his shift started.

Productivity writing quotes him constantly, usually as proof that you could do the same. The passage itself says something narrower. What he describes is not talent but a mechanism: the same hour, the same quota, a clock on the desk, drilled over decades by a man who admitted he could not write without it.

His 2,500 words are the documented upper edge, not the middle. Treat any advice that starts higher with suspicion.

Typing is not the bottleneck

A team at Aalto University analysed 136 million keystrokes from 168,960 volunteers. Mean typing speed was 51.56 words per minute with a standard deviation of 20.2, where a word is defined as five characters. The slowest tenth averaged 20.91, the fastest 89.56 (Dhakal et al., CHI 2018).

I measured the English prose on this blog: 4.72 letters per word, so 5.72 keystrokes including the space. On that basis 1,667 words is about 1,908 five-character words, which at 51.56 words per minute is 37 minutes of pure typing.

Nobody writes 1,667 new words in 37 minutes.

The study measures transcription of given sentences, not composition, and that is exactly the point. The distance between 37 minutes and the two or three hours a session of that size really takes is the cost of deciding what to say. A faster keyboard does not close that gap, and dictation only narrows it, because the bottleneck sits upstream of the fingers.

Calendar day is not writing day

Most numbers online quietly conflate two things. “500 words a day” can mean 500 on each of 365 days. It can also mean 500 on the days you actually sit down. At three writing days a week, the two readings differ by a factor of 2.3.

So work in two steps. Set the weekly figure first, because the week is the unit in which work, appointments, school runs and everything else repeat. Then divide by the number of days you genuinely write. Skipping the second step means comparing numbers that measure different things.

The calculation for your project

The same sum is available as a ready-made writing goal calculator that counts the writing days in the calendar for you. By hand the formula is trivial; everything depends on the buffer.

(target word count minus words already written) divided by (weeks remaining times writing days per week times 0.6)

The 0.6 comes from a study of how badly people estimate their own writing schedules.

Roger Buehler, Dale Griffin and Michael Ross asked students in 1994 when they would finish their honours thesis. The mean prediction was 33.9 days. The mean reality was 55.5. Only 29.7 per cent finished by the date they had named. Asked to assume that everything went as badly as it possibly could, they predicted 48.6 days, and even that pessimistic date was met by just 48.7 per cent (Buehler, Griffin & Ross, Journal of Personality and Social Psychology, 1994).

33.9 divided by 55.5 is 0.61. Hence the factor.

The sample was 33 Canadian students on a single task, so this is not proof of anything. As a correction factor it still beats the optimism it replaces.

Look at the second row. A five-month first draft written on three days a week, with four in ten planned days lost, needs 1,663 words per session. That is NaNoWriMo’s number, reached from the opposite direction. The coincidence is instructive: 1,667 was never a daily writing rate but a sprint rate for people who write on some days only.

You can use the buffer in two ways. Either set the higher quota and finish early when nothing goes wrong, or keep the unbuffered quota and move your personal deadline 40 per cent ahead of the real one. The second is kinder, because it does not turn every single evening into a test.

Writing every day is the exception

“Write every day” is repeated so often that it is rarely defended. Helen Sword checked. She surveyed 1,223 people in academic staff, postdoctoral and doctoral positions at 45 universities across 15 countries, then interviewed 100 successful academics on top of that. The share who write daily, or aspire to: 12.8 per cent among the interviewees, 11.5 per cent in the wider survey (Sword, International Journal for Academic Development, 2016).

Roughly seven in eight productive writers do not write every day.

Sword also traces where the mantra came from: a 1989 study by Robert Boice in which ten hand-picked people, drawn from a group of 108 new faculty, were visited twice a week without warning. What that measured was not self-discipline. It was self-discipline plus a supervisor at the door.

The practical consequence for your goal: a weekly target survives contact with real life, a daily one does not. Take the first row of the table above. As 1,540 words a week, a lost Tuesday costs you nothing. As 385 a day, the same Tuesday costs you the streak, and Wednesday costs you the motivation. The comparison of metrics by project type sets out which unit fits which kind of work.

What beats a bigger number

Two meta-analyses suggest the size of the target is not the deciding factor.

Peter Gollwitzer and Paschal Sheeran pooled 94 independent tests of if-then plans. The overall effect on goal attainment was d = 0.65, where d measures the size of an effect: 0.2 counts as small, 0.5 as medium, 0.8 as large. Broken down by problem, it was d = 0.61 against failing to start and d = 0.77 against giving up partway (Gollwitzer & Sheeran, Advances in Experimental Social Psychology, 2006).

An if-then plan for writing sounds like this: “If it is Tuesday and the kitchen is clean by eight, I open the manuscript.” Not a word about word count.

The second meta-analysis is about measurement itself. Benjamin Harkin and colleagues pooled 138 randomised studies covering 19,951 people. Monitoring progress more often improved goal attainment, at d = 0.40, and the effect was larger when progress was written down and reported to someone (Harkin et al., Psychological Bulletin, 2016).

An appointment plus a record beats a higher number. It is the least exciting finding here and probably the most useful.

When this does not apply

Four situations in which a word count is the wrong unit:

1. Revision. Word count falls when the work is going well, so a word target punishes precisely the work you want. Use time, scenes or chapters instead.

2. Research-heavy non-fiction. A day spent reading three papers and writing nothing can be the most productive of the week.

3. The first four weeks of a project. Until then you do not know what a typical session looks like for you, so any number you set is a guess. Measure first, calculate second.

4. If you already deliver. Some people write without a number and finish anyway. For them a daily target is a solution without a problem.

What to do next

Do not calculate. Measure.

For four weeks, write down how many words you produced and how long you sat there. That gives you two numbers no survey can supply: what a typical session yields, and how many days you genuinely write. Put those into the formula above.

Only then does the target belong to you, rather than to whoever divided 50,000 by 30. More conversions and benchmarks live under Word Goals & Metrics, and the research behind them under Data & Studies.

Share