Friday evening, Dhanmondi, Road 27, the year is 2011 and the ceiling fan is doing the thing ceiling fans do in October when nobody has cleaned the blades. Four of us round a low wooden table. Two plates of nimki gone soft, half a flask of cha gone cold, a stapled printout of an essay called The Last Boat from Sadarghat between the cups, smelling faintly of toner from the photocopier downstairs. The writer is nineteen, an undergrad at North South, and he is also in the room with us tonight, which is the problem the cha is failing to solve.
The piece is five thousand and forty words. It needs to be twelve hundred. He knows this and has come prepared. He tells us very seriously that every paragraph is essential and starts reading the third paragraph aloud to prove it. Around the second sentence his voice drops half an octave because he can hear, in the act of reading it, that the paragraph is not essential. He keeps going. Nobody stops him. We let him finish, drink the last of the cha, and finally Ruhi apa says, very gently, bhai, eta tomar shobcheye priyo line, na? Sheta-i kete dao. Brother, that line is your favourite. That is the one you cut first, the one you would defend with your last breath in a Dhanmondi flat at half past nine on a Friday.
He cut it. The piece ran at fourteen hundred. It was the best thing UNaB published that year, and I have been thinking about that night every Tuesday morning since I started running product teams.
I edited UNaB Magazine for seven years, 2007 to 2014, the only literary publication led by writers in the country at the time. We had a forum on XENFORO, a print quarterly that never quite stayed quarterly, and at peak somewhere north of sixty contributors I knew by handle before I knew by face. I dissolved the magazine in 2014 because the only path forward would have meant becoming a magazine I did not want to read. The eight years since have been B2B SaaS product teams, currently as a Principal Product Owner at Optimizely. After eight years of standups and sprint reviews I will tell you what I have come to believe with the kind of conviction that is hard to argue down. Editing a magazine and managing a product are the same craft viewed from two different professions.
The artifacts differ wildly, but the work of choosing what runs is the same in both rooms, and once you see the shape of it you cannot unsee it. The vocabulary translates almost word for word, the way the same hot metal compositor's terms moved from print into desktop publishing in the late nineties without anyone really noticing.
The list goes on. Above the fold keeps the same job in product surfaces, renamed above the scroll. The lede of a piece is whatever the demo answers in the first thirty seconds. Word count maps to story points in the sense that both are estimates everyone agrees not to believe too literally. The nouns translate easily because nouns sit still. The verbs are where the craft actually lives.
The first verb is voice consistency. A magazine with sixty writers should not feel like sixty magazines stapled together, and the same is true of a product with twelve squads. Most people assume voice gets enforced. In practice it gets curated. The editor selects for voice at commissioning, nudges it during edits without overwriting the writer, and leaves enough of the writer's fingerprint that the piece still reads like a real person wrote it. A design system that operates like a police force will produce something that looks consistent in screenshots and feels dead in the hands of a real user.
The skill is the same in both rooms. You critique the work in a way that does not damage the relationship, because you need the next piece from the same person. Maxwell Perkins, who edited Hemingway and Fitzgerald, was famous for it. He once spent six months on letters to Thomas Wolfe before suggesting a single cut, and that was not weakness, that was the job. The PM who fires off a Slack thread at 11pm with seven inline comments and no preamble is not editing the engineer. They are torching the next sprint to win a small argument tonight.
The third verb is the instinct that something is off before you can articulate why. Editors learn to recognise the kind of piece that comes back grammatically clean and factually correct and somehow still wrong. It is the kind of draft you have to sit with for an afternoon at the back of a Star Kabab before you can name what is broken about it. Product people learn the equivalent feeling from demos that work in the staging build, hit every checkbox in the spec, and still come out the other side feeling unusable in ways that take most of a sprint to put words around. The instinct shows up first and the articulation arrives later, and a team that trains only the articulation ends up extremely precise about the wrong things.
The fourth verb is knowing when a piece is finished. Not when it is perfect, because nothing is perfect, but the moment when one more pass starts subtracting from the thing instead of adding to it. The same instinct ship it points at, except in print you do not get the comma back. It remains the rarest one in either field.
If you have ever wondered what an editor does all day, here is the funnel, applied to a feature.
Process plate · Five stages
Scroll to advance
In 2012 a senior writer pitched a long profile of a Dhaka musician who had just released his second album to a small but obsessive following. The pitch was great. The draft was good. The musician was a friend of the writer, and the friendship was visible on every page. Three rounds of edits and the writer kept removing the wrong things. We killed it, not because it was bad, but because it could not become what it needed to be without breaking a friendship the writer was not in a position to break. We ran a shorter, colder review of the album in its place. The musician, to his credit, sent us a thank you note about the review.
I have made the same call three times in product since. The most recent was a pricing change the entire commercial team wanted, the data supported, and we had spent a quarter scoping. We killed it the week before it was due to be signed off because the team championing it had too much of their identity wrapped up in shipping the maximal version. Six months later we shipped a smaller version, owned by a different team, and it landed.
By 2011 there were something like sixty writers on the UNaB rotation, plus the new ones we picked up at every campus event in Curzon Hall and IUB. For every piece that ran, four or five never made it past the first round, and the ones I still reread years later were not always the ones I was proudest of at the time. A few I tried to kill and the writer talked me out of it, and the writer was right. The features I have bet hardest on in product since are not always the ones that aged best either.
What do product people without an editorial brain tend to get wrong? In my experience there are roughly three recurring failure modes.
The first is they confuse completeness with clarity. They will ship a feature with seven settings, four empty states, and a help article, and call it done. An editor would have asked which three of the seven settings the reader actually needs and cut the other four into a piece for later, so the shipped surface is smaller and the user understands it. Mary Norris, in Between You and Me, has a great line about how the goal of copy editing is to be invisible, and most product surfaces would be improved by a PM who wanted to be invisible.
The second is they treat consistency as a deliverable instead of a side effect. They write a brand voice doc, post it in Confluence, and then are surprised when twelve squads ship twelve voices. Editors know that voice is held in your ear, not in your wiki, and you hold it by reading every piece before it ships, even briefly, even just to flag the places where the voice slipped. PMs who do not read every release note before it goes out will eventually ship a product that sounds like a committee, because it is one.
The third is they do not know when to stop editing. There is a thing that happens around the third round where the piece stops getting better and starts getting different, and editors learn to feel that inflection point but PMs are rarely taught to. The result is the feature that ships in its sixth iteration, technically more polished than the third, but somehow less alive. Camille Fournier writes about this as a management failure mode, the inability to recognise diminishing returns on your own attention.
What PMs should steal from editors is the print deadline. When the magazine ships, it ships. The flat in Dhanmondi clears out, the printer in Karwan Bazar prints, and the issue exists in the world with whatever flaws it has. No hotfix, no point release. The next issue is twelve weeks out and you fix it there.
If you know you can ship a fix on Tuesday, you will not do the work on Monday to make sure Monday's release is finished. The deadline is doing real work for the editor by forcing the call. Eta finished, na finished na, kintu finished korte hobe. Whether it is finished or not, finished it has to be, and at 04:30 in the morning the man at the press in Karwan Bazar does not care which one it is.
I have started running my product cycles with at least one print deadline in them, the kind that does not move. A demo we cannot reschedule, or a customer commitment we cannot soften, or a board review where the slide has to be true on the day it goes up. The work that gets done in the week before that deadline is different work than the work that gets done in a normal week. It is finished work, in the sense the printer means.
I closed UNaB in 2014. Ad revenue would have meant editorial compromises we were not willing to make, and patron money would have meant a different magazine for a different patron. The honest options were to become something the magazine was not or to stop. We stopped. We ran a final issue, told the writers, paid out what we owed, and turned the forum read only. It is still up, somewhere, under a domain I keep renewing out of sentiment, the last issue's masthead frozen at October 2014.
I have killed two products since for the same underlying reason. Not that they were unloved, they were loved, but the path to keeping them alive ran through a version of the product that was not the product anymore. There is a moment in any thing you have built where the question stops being how do we save this. It becomes what is this asking us to learn, which is an editorial question, and product people are slower to ask it than editors are.
After seven years of choosing what runs and what does not, the things that did not make it stop being precious to you. What stays precious, sometimes annoyingly so to the design lead next to you in the Tuesday standup, is the craft itself, the same craft the magazine was a vessel for and the product is now another vessel for. You will carry it past both, into whatever the next thing turns out to be, the way the night editor at the Karwan Bazar bureau carried the same red pencil into three different newspapers and one weekly that did not last long, that nobody remembers anymore but him.