How I Created an Algorithm That Matches Stories to Literary Journals

And how you can, too

Published in

The Book Mechanic

5 min readSep 16, 2020

I recently wrote an article about the single biggest mistake writers make when submitting their stories to literary journals. It is simply this: writers tend to send their work to journals that are not good matches for their stories. Square pegs in round holes.

I made this same mistake when I started submitting several years ago. Since I had written the greatest story in the history of literature, I figured I’d send it to just a few of the very best journals.

And then I got a bunch of form rejections.

Frustrated, I sent my story (now I wasn’t so sure about it) to every journal with an OPEN sign in the window. I did finally get published, but it took way too long and cost me way too much in submission fees. There had to be a better way to submit to literary journals, and after a long time I came to the following astounding and groundbreaking realization:

I should actually READ these things!

So I read and read and read . . .

and read and read and read . . .

and read some more.

(A lot of reading went down over here.)

I started to get a “feeling” for different journals, and I took notes. Conjunctions publishes formally experimental fiction. Salmagundi often prefers humor and wit. StoryQuarterly seems to prefer dark themes. Arts & Letters has a penchant for international settings. And so on.

But after a dozen or so journals, I realized I needed to quantify things. But how?

Dr. J. Evans Pritchard, PhD

Remember that scene in Dead Poet’s Society when Mr. Keating teaches his class the Dr. J. Evans Pritchard, PhD, system for analyzing poetry? He graphs perfection on the horizontal axis and importance on the vertical axis, draws some perpendicular lines, and determines a poem’s “greatness” by calculating the total area under the resulting rectangle.

Yes, I know: “Excrement” (watch the scene). But this was basically what I was trying to do. And, yes, it felt just as futile and ridiculous. At least at first.

Over time, however, I settled on a handful of categories that I believed pretty effectively distinguished one journal from another. I scored each category on a scale of 1 to 5. And voilà! My algorithm was born.

Spreadsheets, pivot tables, and other programmy things

Then I made a spreadsheet. I’m no Excel genius. You can do this. (If, however, at this point you’re wondering what “spreadsheet” and “Excel” mean, skip ahead a few paragraphs.)

To put it (very) simply, I have two “sheets” in my algorithm: one for stories and one for journals. Every time a read a story, I indicate its source (its journal), and I rate the story in each of my categories on a scale of 1 to 5. These are all columns of my sheet.

Then I use a “pivot table” to average the scores for each journal (this is the second sheet). You can then sort and sub-sort by categories. So, for example, if you’re looking for a journal with high scores in difficult experimental fiction, just sort your pivot table in the appropriate categories (difficulty and experimentation).

Honestly, none of this stuff is as hard as it may sound, but if you need help from a programmer to set up your algorithm, I recommend it.

The real fun for me was adding a feature that would give me a “percent match” for a story. (I was about to give you the equation, but let’s keep things simple: just hit me up — I’m easy to find.)

For example, a client of mine sent me a well-written flash fiction story with relatively high scores in experimentation, quirkiness, and humor (plus “shortness,” another category). Five years ago, I might have submitted to the typical journals we’ve all heard of: New Yorker, AGNI, Narrative, etc. But my algorithm spit out high matches for journals like Booth (87% match), Hobart (85%), Wigleaf (85%), and SmokeLong Quarterly (84%). (Over 80% is very good for my algorithm.) These journals — all excellent — are not ones I would have even considered when I first started submitting.

(For the record, New Yorker was 77%, AGNI was 75%, and Narrative was 68% — not bad, but down the list a bit.)

A couple obviouslies

Obviously 1: You’ll have to start actually reading literary journals. You’ll need to read enough stories from a journal to have a fairly accurate measure of its preferences. I think three stories is a minimum, but for some journals — especially when the stories are particularly varied — you might decide to read more. (If you don’t want to put in the time, there’s at least one company that can help.)

But, seriously, this is the greatest homework assignment ever. Besides getting to know literary journals, you’ll discover some great writers, and you’ll learn a ton about writing, both what to do and (often, just as important) what not to do.

Obviously 2: You should probably have a little experience with spreadsheets. I use Excel, but Google Sheets (or probably any spreadsheet application) works. Again, find someone with some programming experience to help you set things up, or find a tutorial on pivot tables; there are many.

And then get reading! And then get submitting!

If you need help, drop me a line. I hope the above encourages you to create your own personal submission algorithm. Good luck!

— Erik Harper Klass is the founder of Submitit, the world’s first full-service submissions company. He has published stories and essays in a variety of journals, including New England Review, Summerset Review, Maryland Literary Review, and Open: Journal of Arts and Letters, and he has been nominated for multiple Pushcart Prizes.

You just read another exciting post from the Book Mechanic: the source for writers and creators who want to make more work that sells and sell more work they make.

If you’d like to read more stories just like this one tap here to visit