Against the Daily Word Count

A count of birds seen tells you almost nothing about a colony's health. A count of chicks that fledge tells you nearly everything. Writers have picked the wrong number to track for the same reason bad surveys do: the wrong number is simply the one you can get in an afternoon.

I want to make the argument against the daily word count directly, because I think it survives as a habit not through any evidence that it works but because it is the only number available at the end of a single day, and people confuse availability with usefulness constantly, in writing as in ecology.

What a word count actually measures

A word count measures output at the moment of production. It does not measure whether that output survives contact with the next draft, whether it moves the project closer to a shape it can be finished in, or whether it will be deleted in six weeks once you understand what the book is actually about. In a field study, this would be the equivalent of counting eggs laid and reporting that as the health of a population, when the number that actually predicts whether a colony persists is fledging rate, the fraction of eggs that become adult birds that themselves breed. Eggs laid is a number you can get in an afternoon. Fledging rate takes a season and demands you come back.

Writers track word count for the same reason a lazy survey tracks eggs laid: it is available same day, it produces a satisfying graph, and it lets you claim progress without waiting to find out whether the progress was real. I kept a word count for the first two books I wrote and stopped after the second, when I noticed that my best writing month by word count, fourteen thousand words in twenty two days, produced a chapter I deleted in full four months later, while my worst writing month by the same measure, under two thousand words, produced the paragraph that became the spine the rest of that book was built around.

What to count instead

The alternative is not to stop counting. Field biologists do not respond to a flawed metric by abandoning measurement, they respond by finding the metric that actually predicts the outcome they care about. For a long manuscript, the equivalent of fledging rate is retention: of what you wrote this month, how much is still in the draft at the end of next month. That number is slower to get, because you need thirty days to know it, and it is more embarrassing, because it will show you months where you produced four thousand words and kept two hundred of them. But it is the only number I have found that reliably predicted, across five books, whether a project was actually converging on something or generating material that would need to be cut regardless of how it felt to write.

Critics of this, fairly

Critics of counting retention instead of output make a point I do not dismiss: a monthly retention number gives you no signal at all for four weeks, and a writer who needs daily feedback to stay at the desk, which is most writers including me on a bad week, will simply stop showing up without some more immediate number to chase. That is a real cost, and I do not think retention alone solves the discipline problem word count was originally invented to solve. What I do instead is keep the word count purely as a same day attendance record, did I sit down, and stop treating it as a quality signal at all. The two jobs, showing up and producing something worth keeping, got confused because one number seemed to be doing both. It was never doing the second one.

The number that actually predicts the book

Here is what changed once I split the two. My attendance record stayed almost exactly the same, because showing up was never the problem retention solved. My relationship to a bad word count day changed completely, because a low number stopped meaning failure and started meaning nothing in particular, a fact awaiting its retention check thirty days later. And my relationship to a suspiciously good day changed too. Fourteen thousand words in three weeks is not an achievement until you know what fraction of it survives, and most of the time, in my own record across five books, the answer has been under twenty per cent. A word count tells you how much grain you cut. It does not tell you how much of it was ever going to be bread.