The daily stand-up you cannot cancel: what ten minutes a day has to produce

What has a daily stand-up been shown to produce?

Nine fifteen, eleven people on a call, and ninety seconds of nothing much to report sitting there with your name on it. An honest answer about whether that is worth it did not exist in randomised form until this year, and when it arrived it came out of hospitals rather than software teams.

Guo and colleagues published a stepped-wedge cluster randomised trial in BMJ Quality and Safety in 2026, covering inpatient oncology, surgery and rehabilitation programmes at four academic hospitals. Each site ran a four-month control period, then started medical safety huddles on a computer-generated random sequence two months apart, and stayed on them for at least nine months. The huddles were short structured meetings where physicians raised patient safety concerns and decided what to do about them.

Blinded auditors afterwards read the charts of 912 randomly selected patients. The adjusted rate of adverse events per 1000 patient days came out 12 percent lower after the huddles than before, with a relative risk of 0.88 and a confidence interval of 0.80 to 0.98. That is the strongest claim anybody can make for a meeting of this shape, and it is not a claim about your team. The physicians also rated their own safety culture higher afterwards: the share scoring communication about error positively rose from 36 to 64 percent, and organisational learning from 35 to 54 percent.

Read past the headline, and the mechanism that decides your ten minutes is sitting in the fidelity data. Each huddle raised between 1.6 and 3.1 safety issues on average across sites, and took roughly that many actions in response.

That is the whole machine.

Somebody named a thing out loud, and somebody else did something about it before the next one came round. Nothing in the trial suggests the meeting worked by being short, or by being held standing up.

The attendance figure is the one to take to your own calendar. Mean huddle attendance ran from 30 percent at one site to 85 percent at another, and the trial still measured fewer adverse events overall. What the trial never tested is whether attendance was the thing doing the work. If you have been treating the stand-up as a roll call you cannot miss, the only randomised evidence for the format does not back that up.

Now the part that fences all of this in. Every study on this page ran in healthcare, with clinicians, about patient harm, and a missed medication interaction is not a missed handover on your project board. The outcome those huddles were measured against was countable and plainly bad, which sits a long way from whether your release slipped a week. Carry the mechanism across to your own team if you like. Do not carry the 12 percent.

Why does your stand-up produce nothing?

Because nobody in your room ever agreed what it was for. That reads like a cheap diagnosis until you look at what happened when researchers set out to collect the evidence on huddles and found they had to build a vocabulary before they could start.

The format did not start in software. Franklin and colleagues open their review by noting that daily huddles earned their reputation in operationally complex industries, aviation and nuclear power among them, before healthcare borrowed them at all. Their own search covered the peer-reviewed literature through December 2019, looking for studies where a huddle was the main intervention and at least one outcome was measured with a number. They screened 1034 articles and kept 24, nineteen of them on single units and five hospital-wide or multiunit. The review, in BMJ Quality and Safety in 2020, reports that the objectives of huddle programmes and the language used to describe them varied widely across those studies. They finished by proposing a taxonomy, which is what you do when the thing under review has no agreed definition.

The results read well and thin at once. Of the twelve unit-based studies reporting complete measures of statistical significance, eleven found significant improvement in some or all of their outcomes. Then the designs: uncontrolled before-and-after comparison was the prevailing method, and the reviewers identified only two controlled studies in the entire set. A unit that introduces a huddle is usually a unit paying close attention to several things at once. So the strongest honest claim for your stand-up is that the format can work, not that it does.

The variation those reviewers kept running into is the same variation you sit in every morning. A stand-up can be where blockers get named and cleared. It can also be a status round, eleven people describing yesterday to a manager who could have read it. Both cost you ten minutes a day. Only one of them has anything in common with the meetings that got measured.

The test that separates them is small enough to start tomorrow. Count how often in a week somebody raises something nobody else knew, and how many of those changed what you or anybody else did that day.

Zero is an answer.

If your own count holds at zero across a fortnight, the meeting is producing attendance rather than information, and that is worth saying once in the room rather than quietly drifting away from it. Where the coordination work actually lives is a Personal Kanban question. A board with a waiting column shows your blocked item to everyone without anybody having to say it out loud at nine fifteen.

Can you fix a bad stand-up by running it better?

Probably not from the chair, and that is the most useful null result on this page.

Lampman and colleagues ran a cluster-randomised trial across 29 primary care clinics inside one large health system, reported in the Journal of General Internal Medicine in 2021. Thirteen teams got the intervention and sixteen carried on as usual. The intervention was not a tweak. Team leaders went to a full-day leadership retreat on facilitating their daily huddles, then got biweekly coaching calls and two visits from an assigned coach. Surveys came back from 279 people beforehand and 272 afterwards.

Team development did not move, at -0.98 on a scale where the interval ran from -3.18 to 1.22. Team credibility rose slightly, at 0.18. Psychological safety went the wrong way, at -0.19, meaning people felt marginally less able to speak up once their leaders had been trained to run the meeting better. Patient satisfaction showed no difference and the quality outcomes came out mixed. The authors concluded that leadership training and facilitation were not associated with better team functioning.

One result inside the intervention group does point somewhere, though the authors went looking for it after the fact rather than predicting it. Teams that stuck closely to the programme improved on team coordination, credibility, team learning and knowledge creation compared with teams that engaged less, with knowledge creation the largest of the four at 0.74. Engagement tracked with benefit. Training the facilitator did not. Read that as a lever worth trying rather than one anybody has proven.

Which puts the move back in your hands instead of the chair's. You cannot redesign the series, and the evidence says that whoever runs it may not get far redesigning it either.

What you can do is bring one real thing: a blocker you would otherwise sit on until Thursday, a dependency somebody is about to trip over, or a decision you need from a person who happens to be in the room right now. One of those turns your ninety seconds into the only part of the meeting that resembles what got measured. On the mornings when nothing like that exists, say so plainly and hand the time back.

This works for people whose week runs through the people in the room. It does not work where your work has no dependencies there at all, and nobody is served by pretending otherwise. A stand-up of eleven people reporting in parallel, with nothing shared between them, has an attendee-list problem. That one belongs to whoever owns the series. You will not solve it inside your ninety seconds.

How does your working style change the answer?

Which part of this you get wrong tends to follow how you already work. The four names below are this site's own, from the working-style self-check on the home page.

Architect-leaning readers turn up with the update already written and get the least back from the meeting, because a prepared report is exactly the thing the trial did not measure. Your notes are not the finding. Bring the one item you are least sure about instead of the three you have already handled.

Sprinters resent the interruption, and the arithmetic behind that is real: a stand-up at nine fifteen cuts the morning in two. Timeboxing the block after it rather than the block before gives you an hour the meeting cannot eat, since the ten minutes is fixed and your deep work is not. What costs you is the fragmentation, not the ten minutes.

Visionary-leaning readers do the opposite damage. You open the dependency question that actually matters and then keep going, and a ten-minute meeting with one person thinking aloud in it is worse for everybody else than a status round. Name the thing, ask who to take it to, then take it there afterwards.

Improvisers get real value out of the stand-up and tend not to admit it. Hearing what four other people are doing is how you find the overlap you had no reason to go looking for, and no study here measured that. Worth noticing anyway, because it means the mornings you skip cost you something that the people who prepare never lose.

So run the count yourself. For the next fortnight, note each morning whether anybody raised something that changed what anybody did, and bring one real blocker of your own on the days you have one. If the count holds at zero and your own item never moves anything either, you have something better than an opinion to take to whoever owns the series, and the meetings pillar covers that conversation.

Frequently asked questions

Sources

• Guo et al. (2026), BMJ Quality & Safety: Impact of medical safety huddles on patient safety - a stepped-wedge cluster randomised study
• Franklin et al. (2020), BMJ Quality & Safety: Impact of multidisciplinary team huddles on patient safety - a systematic review and proposed taxonomy
• Lampman et al. (2021), Journal of General Internal Medicine: Optimizing huddle engagement through leadership and problem solving within primary care