The 18-hour week that saved our roadmap

We hit a wall in June—three engineers, one designer, and me, all staring at a Jira board that had ballooned to 127 tickets. The backlog wasn’t just long; it was loud. Every ticket had a champion, a Slack thread, or a customer name attached. The team was shipping, but the load felt like we were running a marathon with a backpack full of bricks. I pulled the plug on new work for two weeks. No sprint planning, no standups, just a single rule: every engineer had to delete or merge at least 10 tickets before they could add one. The designer was given the same task for Figma files. By the end of the first week, we’d cut 43 tickets. Not archived—deleted. The relief was immediate. The team started talking about what they actually wanted to build, not what they’d been assigned. One engineer, who’d been quiet in standups for months, proposed a refactor that would halve our cloud costs. Another merged three separate feature requests into a single toggle. The designer killed six variants of a component that had been lingering for a year. The second week, we ran a silent retro. No talking, just sticky notes on a Miro board. The theme wasn’t ‘what went wrong’ but ‘what felt unnecessary.’ The word ‘guilt’ appeared seven times. Guilt over tickets that had been open for months, guilt over saying no to customers, guilt over not shipping faster. We’d been treating the backlog like a to-do list, not a design constraint. We kept the rule. Every Monday, each person still has to delete or merge 10 items before they can add one. The backlog is now 34 tickets. The team ships faster, but more importantly, they argue less. The conversations that used to happen in Slack threads—‘why is this still open?’—now happen in person, or not at all. The load isn’t lighter because we’re doing less; it’s lighter because we’re not carrying the weight of things we’ll never do.