Words, Hours or Pages? How to Choose What to Track

Should you track words, hours or pages? Count words while you draft, hours while you research and revise, and pages only when a contest or publisher fixes them.
A metal ruler with inch and centimetre scales lying at an angle across a printed page, single words visible through a hole in the ruler
Photo: Linus Belanger on Unsplash

Count words while you’re drafting. Count hours while you’re researching, outlining or revising, since that’s when good work can leave the total flat or even push it down. Count pages only if someone else has defined the page for you, the way a screenplay contest does. Most books go through every one of those stages, so you’ll probably switch more than once before you’re done.

That isn’t the answer you get from the guides that turn up first for this question. I read them on 20 October 2026, and most of them pick a side and stay there. A post on the Kotobee blog gives four reasons to track writing time instead of word count. Ann Handley’s book Everybody Writes has a chapter called Set a Goal Based on Word Count (Not Time). Neither of them cites any research. There is some, though, and it doesn’t hand the prize to either unit. Each one does a particular job well and gets misleading when you ask it to do another.

One novel, one year

Take a novel that’s meant to end up somewhere around 80,000 words. You spend March reading and outlining, and at the end of the month the manuscript file is still empty. From April to July you draft. In August you revise, and the file gets shorter. In September your editor’s notes come back and you work through them for weeks.

Now imagine you’d logged that year in words alone. March shows nothing at all. August shows a loss. September is a run of small plus and minus days that cancel out. Only the drafting months look like work, and they’re also the only months in which the number really follows your progress.

Log the same year in hours and every month shows up. What hours can’t tell you is whether a long week in July produced two chapters or an afternoon of rereading old ones with your email open. Pages would be the least useful of the three here. Until somebody fixes the font, the size and the margins, a page doesn’t hold any particular amount of text, and I’ll come back to that further down.

Drafting: count words

When you’re drafting, the deadline is nearly always stated in words. So is almost any length that an agent or a publisher will hold you to. If you know the total and the number of days you have left, the daily target is just division, and the writing goal calculator does it for you with a buffer built in. To plan in hours you’d also have to know your writing speed. That’s one more guess in the chain.

Goal-setting research is on the side of words here. In 2002 Edwin Locke and Gary Latham looked back on nearly four decades of studies. Specific, difficult goals, they wrote, had raised performance on “well over 100 different tasks involving more than 40,000 participants in at least eight countries”. Set against the instruction to do your best, those goals came out ahead in the meta-analyses they cite, with effect sizes from 0.42 to 0.80 (Locke and Latham, American Psychologist 2002). Their explanation is worth quoting. Do-your-best goals “have no external referent and thus are defined idiosyncratically”. Put plainly, “do your best” means whatever you decide it means at nine in the evening, while 1,000 words means the same thing at nine in the evening as it did over breakfast.

Handley gives the practical version. You should measure your writing in output (words) “rather than in effort expended (time)”, she says, and her next sentence is “Staring at an empty page for half an hour doesn’t count”. She doesn’t let half an hour of really good thinking count either. For a first draft I’d go along with that. How big the daily number ought to be is a separate question, and the article on how many words a day works through the ranges that hold up.

Revision: where words stop working

The same 2002 paper has a caveat that gets quoted far less. Goals help less on complex tasks. When Locke and Latham compare specific difficult goals with do-your-best instructions, the effect size is 0.77 for the least complex tasks and 0.41 for the most complex ones. On new and complex tasks, they write, urging people to do their best “sometimes leads to better strategies” than a hard performance goal. A performance goal can make people so anxious to succeed that they never work out what’s effective. What the two authors recommend for that situation is a specific learning goal, one that’s about the process.

Revision is that kind of task. A good pass might add 2,000 words or cut them, and either could be the right call. You can see this on the blog you’re reading, because every change to an article ends up in its version history. The English article on writing streaks was rewritten for readability on 14 September 2026. By my count its text went from 2,224 words to 1,977. On 16 October a reference to a withdrawn paper came out, and it lost another 56. The page got better both times. A word counter would have marked both days as negative.

Chris Krycho noticed the same thing in a short post from November 2019. He spent ten minutes on a piece he’d started a few months earlier, wrote some new words, and still finished 40 words down. Time spent writing each day, he thought, might be a better measure for that sort of goal. Research and outlining behave the same way. They make the book better and add nothing to the count.

When hitting the number becomes the point

“When a measure becomes a target, it ceases to be a good measure.” The line is usually traced to a 1997 paper by the anthropologist Marilyn Strathern about audits at British universities, where she gives it as a version of Goodhart’s law. You don’t need a university to watch it happen. Set yourself 1,000 words a night and look at what a tired Thursday produces. It’s usually the easy material, like a flashback the story didn’t need or a page of dialogue in which nobody says anything new. The count goes up while the manuscript stays where it was.

Lisa Ordóñez and three colleagues went looking for the downside in 2009. Their paper, Goals Gone Wild, names six side effects. People narrow their focus and neglect whatever the goal doesn’t cover. Their attitude to risk gets distorted, they cut ethical corners more often and they learn less. The culture around them suffers, and the work stops being interesting for its own sake. The authors would like goals to be handled like “a prescription-strength medication”, carefully dosed and closely supervised. Their examples come from companies, so you can’t lift the findings straight onto a novel. You can borrow the mechanism, though. Think of that tired Thursday again. With the number to hit, you’d probably write the safe scene and skip the hard one, and you wouldn’t learn much from either.

Locke and Latham didn’t take it quietly. They replied in the same issue with a paper whose title tells you the mood, Has Goal Setting Gone Wild, or Have Its Attackers Abandoned Good Scholarship?, and Ordóñez’s group answered them back. Set the tone aside and there’s one thing nobody in that exchange disputes. On a task you’re still learning, a hard performance goal can make you do worse. So I’d keep a word target for exactly as long as more words and a better book are the same thing. For most people that means the drafting months, April to July in the example.

Research, outlining and revision: count hours

Here’s the part you do control. You can’t make 800 words turn up on a Tuesday. You can sit down at a quarter past nine and stay until a quarter past ten, and that’s the habit you’re really trying to build. An hour is also an hour whatever you spend it on. Reading sources in March counts exactly the same as drafting in May.

Robert Boice tried something along these lines in 1989. As a treatment for procrastination, he helped new faculty members find brief daily sessions to write in, and the intervention proved effective and “relatively unaversive” (Procrastination, busyness and bingeing, Behaviour Research and Therapy 1989). That’s a researcher’s way of saying people didn’t hate it. The abstract gives no sample size and the study is more than 35 years old, so I wouldn’t lean on it hard. It points the same way as the 30-day writing habit plan, which uses newer evidence.

Hours also make you better at forecasting your own work. If you’ve ever promised someone a chapter by Friday and missed it, you’re in good company. Roger Buehler, Dale Griffin and Michael Ross published five studies in 1994 in which students predicted when they’d finish academic and non-academic tasks, and the predictions were too optimistic. The fourth study is the useful one. Participants who’d been told to connect relevant past experiences with their predictions didn’t show the bias (Buehler, Griffin and Ross, Journal of Personality and Social Psychology 1994). An hour log is that past experience, written down. If chapters 5 to 7 each took nine hours to revise, chapter 8 won’t take three.

The trouble is that an hour log records attendance and nothing else. Ninety minutes with a browser open in the next window still goes in as ninety minutes. So add a line about what actually happened. Benjamin Harkin and colleagues pooled 138 studies with 19,951 participants in a 2016 meta-analysis. Interventions that got people to monitor their progress improved goal attainment, with an effect size of 0.40, and the effect was bigger when the information was physically recorded and when the results were reported or made public (Harkin et al., Psychological Bulletin 2016). A note like “revised chapter 7, cut the dream sequence” takes ten seconds to write.

Pages: only when someone else defines them

I checked the 250-word page against this blog once, with the English prose of five articles in lines of 60 characters and 25 lines to a page. It came to 236 words a page, about 6 percent short, and the article on the standard manuscript page has the numbers. The 250 itself is typewriter arithmetic, as an article on the SFWA site by John Gregory Betancourt shows, and William Shunn explained in 2021 why proportional fonts broke it. In your own word processor, the number of words on a page moves with every change of font, size or spacing.

A page only stays put in two places. Screenplays are one. The Nicholl Fellowships for 2026-2027 want feature scripts in 12-point Courier and between 80 and 125 pages, and the competition rules warn that cheating the margins to get past 125 pages can count against you. Your own handwriting is the other, because it’s the same notebook and the same hand every day. The article on words per handwritten page shows how to convert them.

Trollope’s clock and Hemingway’s chart

Anthony Trollope gets held up as the patron saint of word counts. Read the passage in his autobiography, though, and the first thing on the table is a watch. He sat down at 5:30 every morning and spent the first half hour reading over the previous day’s work. Then he required of himself 250 words every quarter of an hour, and he was done before he dressed for breakfast. He thought that “three hours a day will produce as much as a man ought to write” (An Autobiography, chapter 15). So the clock was his commitment. The 250 words were a spot check that he wasn’t dawdling. Worked out per hour, that’s 1,000 words, and Words per Hour shows how to time your own sessions to find your figure.

Hemingway kept his record the other way round. George Plimpton’s 1958 interview for The Paris Review describes a large chart, made from the side of a cardboard packing case and propped against the wall under a mounted gazelle head, on which Hemingway tracked his daily output “so as not to kid myself”. The numbers on it ran from 450 to 1,250 words a day. The high days were the ones when he put in extra work so he wouldn’t feel guilty spending the following day fishing on the Gulf Stream (The Art of Fiction No. 21). He wrote the figure down once the day was over, and his best day on that chart is nearly three times his worst.

If I had to copy one of them, it would be Trollope. Pick one number as the target, the one that decides whether today counted. Write the second number down next to it with no target attached, and look at it once a week to see whether the target still makes sense. With targets on both, you’ve given yourself two ways to fail on the same evening.

Which number for which stage

Here’s the novel from the beginning again, stage by stage. The second column is what you aim for, and the third is what you simply write down.

Target and recorded number for each stage of a writing project
StageTargetWrite down alongsideSign the number has stopped fitting
Research and outliningHoursNotes made or scenes outlinedYou put off research because it “doesn’t count”
First draftWordsHoursYou write paragraphs you already plan to delete
RevisionHours, or chapters revisedWords, as a manuscript totalThe counter falls and you feel you did nothing
Working through an editor’s notesNotes resolvedHoursNone of the work shows up in words
Submitting to a fixed formatPages or characters, as specifiedWordsYou keep converting your word count by hand

Revision has another option that’s easy to miss, which is counting chapters or scenes. “14 of 31 revised” only ever goes up, and it doesn’t care whether a pass added text or took it out. Once a book is drafted, that count often tells you more than hours do.

Where none of this applies

If there’s no product at the end, as with morning pages, a journal or freewriting, length doesn’t mean anything. Ticking off the day or noting the minutes is enough. Anything finer would suggest a precision that kind of writing doesn’t have.

If you’re paid by length, use your client’s unit. A translator billed per word or a journalist asked for 1,200 words has no reason to plan in anything else.

Academic writing is trickier. A page of discussion with five checked citations takes much longer than a page of introduction, and a word count treats them as equal. Through the reading and analysis, hours or finished sections usually work better. For the drafting part, the article on a dissertation’s daily word count shows how to plan.

The tool you track with matters less than the unit you pick. A notebook works, a spreadsheet works and so does an app. The comparison of five ways to track writing progress looks at what each one costs you in attention.

Do deleted words count as negative?

It depends on what you want the number to tell you. If the question is how long the manuscript is, the lower number is simply correct. If the question is whether you worked, the word count is the wrong measure for that day, and hours or chapters revised will tell you more.

Is a time goal or a word goal better for building a habit?

For the habit itself, time is often easier to keep, because it only depends on whether you sit down. Word goals work well once the habit holds and you’re drafting. The article on writing streaks looks at how strict a daily rule should be.

How do I track hours if I keep getting up?

Use a fixed block of 25 or 45 minutes and log it as one unit. That’s easier to keep up than a stopwatch you pause every time you stand. Any error then points the same way every week, so your weeks stay comparable. Writing Sprints: How Long to Sprint and How Many Words to Aim For walks through one of these blocks from start to finish.

Should I count characters instead of words?

Only if someone pays you or limits you by characters. That’s common in German-language publishing and rare in English. For your own planning the two convert reliably enough that it makes no difference which you pick.

I work on several projects. Should I add them up?

Keep one number per project, chosen by the stage that project is in. Adding words across projects hides the fact that 1,200 words spread over three drafts move each book less than 1,200 words in one. Hours are the only number you can pool without losing their meaning. How to Work on Multiple Writing Projects at Once shows how to order several deadlines so that the daily goals together stay within your pace.

Which stage is your main project in this week? Find that row in the table. If the number you’re tracking isn’t the one in the target column, switch to it now.

Share