Six Months on Linear, Including the Two Weeks We Nearly Went Back

We moved eleven people off Jira and onto Linear in February; the migration itself is its own story. I want to be careful about the word “fast,” because every project tool claims it and most of them mean the marketing site loads quickly. Linear means something narrower and more useful: the gap between deciding to file an issue and having filed it is about four seconds, and that gap is the whole product.

Here is the concrete difference. In Jira, creating a ticket meant a modal with a required project, a required issue type, a required reporter, and — in our configuration, which was not unusual — three custom fields that someone had made mandatory in 2021 for a workflow that no longer existed. Median time from intent to filed ticket, measured across a week of me timing myself with a phone stopwatch, was fifty-one seconds. In Linear it is a keyboard shortcut, a title, and enter. The rest can be filled in later or never.

That sounds like a small thing. It is not. When filing is expensive, people stop filing, and work moves into Slack threads and hallway conversations where it cannot be counted. In the two months before the move, our backlog grew by 34 issues. In the two months after, it grew by 210. The work had always been there; we had just been paying a fifty-second tax to write it down, and quietly deciding it wasn’t worth it.

The two weeks we nearly went back were in March, over cycles. Linear wants you to work in fixed-length cycles with automatic rollover, and it is genuinely opinionated about this — the concept is not optional in the way a sprint is optional in Jira. Our team does a lot of unplanned client work, and for a fortnight the cycle view was mostly a list of things we had failed to finish, rolled forward, accumulating. It read as an accusation. Two people told me independently that they had stopped opening it.

What fixed it was not a setting. It was accepting that our cycles needed to be one week, not two, and that roughly forty percent of each one should be left empty on purpose. Once we did that, rollover stopped being a shame list and started being a signal. But I want to be honest that the tool pushed us toward a working style, and if your team cannot adopt that style, Linear will fight you every day. It is not a neutral container the way Jira tries to be.

Search is the weakest part and it has not improved in six months. Linear’s search is fast but literal; it does not do much with synonyms or partial matches, and it will not find an issue whose title says “auth” when you type “login.” Our workaround is a naming convention, which is exactly the kind of process overhead a tool like this is supposed to remove. If your archive is the thing you rely on most, test this specifically before you commit.

On price: we pay for the Business plan at $14 per user per month, billed annually — a loan to the vendor, really — which came to $1,848 for the year for eleven seats. Jira was costing us $1,006 on Standard. So we roughly doubled our spend, and the per-seat arithmetic says it will grow with us, and I would still do it again — but I want the number stated plainly, because most reviews of this tool are written by people whose employer absorbs the difference invisibly. Ours did not; it came out of a budget I had to defend.

The thing I did not expect: triage got better because the tool made the backlog visible enough to be embarrassing. Twice now we have deleted more than eighty issues in an afternoon, because we could finally see that they were all the same three complaints filed by different people. That cleanup was available to us in Jira too. We just never looked, because looking took fifty-one seconds at a time.