Micro-distractions, big impact: the cost of context switching

The inside of a factory floor filled with machines and production equipment.

We all start with big plans on working days: a few hours we’ve blocked out for proper progress, a list that looks achievable, and a picture of finishing the day with most of it ticked off. Then context switching gets in the way, often within the first hour, when a message pops up in the team chat asking if you can spare a minute.

Answering may only take that one minute, but returning to the task you were previously focused on takes a lot longer. This scenario is the reality for most delivery teams. It happens a lot more often than people generally admit when I bring it up with them.

Small requests add up

It’s fair enough that someone needs a second opinion on their design work, another pair of eyes on a bit of complex code, or a quick answer for a client who’s waiting. Add to that the notifications that have your hand reaching for the phone before your brain has even placed the sound, and the “quick catch-up at 11.15, hope that’s OK?” that appears in your calendar mid-morning.

Whoever’s asking is usually just trying to keep their own work on track, and a single request on its own barely registers for most people. Over a full day though, four hours of planned focus can easily end up as just one or two.

Plenty of teams have come to treat that as just how work is, so finding focus means starting early, staying late, or sticking headphones on and hoping for the best.

The production line that keeps stopping

I tend to picture it as a big production line, where each time someone asks for input, the person they ask has to switch off their part of the line and walk over to help. When they get back, the machine takes a while to get up to speed again, and any work queued up behind them, along with the people who rely on them, sits waiting too.

Delivery work is messier than a factory, but the same thing happens to a developer halfway through a tricky bit of code, or a designer with a half-finished layout. Answering takes a minute, and getting back to where they were can take up to fifteen minutes, depending on the task.

Paul Graham wrote about this in his essay on the maker’s schedule and the manager’s schedule. A manager’s day is split into hour-long slots, so one more meeting barely registers, while people who make things need half a day at a time, and a single meeting in the middle can break it up into less-productive blocks.

In most agencies, project and account managers work on a manager’s schedule because client work needs them to, and they book calls into developers’ days for the same reason. It’s a sensible way to run client work, but the two schedules were never designed to share a calendar.

The two schedules were never designed to share a calendar.

How long things can wait

Apart from people working on the same piece of work together, there’s very little in a delivery team that can’t wait a few hours for an answer. Quick replies have become the norm though, and asking feels free to whoever’s doing the asking. Agreeing a short gap between a question and its answer is usually enough to get started.

Small steps

The teams I’ve seen do best at this didn’t start with a big change programme. They started small and adjusted and reviewed as they went.

Block out more than half the day, usually the mornings, when nobody outside the team expects a reply and client calls and check-ins wait until the afternoon.

Give questions somewhere to wait, like a shared thread or channel that gets checked at set times, say before lunch and before home time. They still get answered, without pulling people out of their work.

Think about context when you ask. A question about the thing someone’s already working on costs them far less than one about a different project, so where you can, send it to whoever’s closest to it.

Put stand-up at a natural break, either first thing or straight after lunch, so it doesn’t cut a morning in half.

If you want something for your team’s working agreement, this is a good start:

Mornings until 12 are focus time. Messages from outside the team get a reply after lunch. If it really can’t wait, call.

Try it for a couple of weeks, then have a look at what got done and how folk felt about their days.

Most people join delivery teams because they like getting stuck into a problem. A morning without interruptions gives them the chance to do that, with a bit of time to stop and think as well.

I’m Harry Bailey. I work with delivery teams whose days get chopped up by everyone else’s quick questions. See how a team reset works.

Photo by amin khorsand on Unsplash

Why “I’m stuck on this ticket” always turns up on Thursday

Torn paper order tickets hanging from a metal ticket holder in a restaurant kitchen

Nine in ten tickets I see come with conversations attached.

Could you have them before the ticket’s marked Ready?

It’s Thursday afternoon. A developer posts in the team channel: “I’m stuck on part of this ticket.”

It feels like a surprise. Really, it was decided on Monday, when the ticket went into the sprint without anyone checking what it would take to finish.

When I start working with a team, one of the first things I look for is waste. Time spent waiting, redoing work, or working something out for the second time. A fair bit of it comes from how we treat tickets, and the fix usually starts with how the team writes its definition of ready.

What a clear ticket looks like

A clear ticket needs little or no discussion. Anyone on the team with the right experience can pick it up, build it, test it, deploy it and show it to the client, without someone having to explain it first.

That’s rare. Possible, but rare.

Small tickets need fewer conversations

There’s an old checklist for good tickets called INVEST, from Bill Wake. A good ticket is independent, negotiable, valuable, estimable, small and testable.

Small does the most work. The smaller the ticket, the fewer conversations it needs and the easier it is to make clear. A big ticket usually struggles with the rest too. It’s harder to estimate, harder to test, and tangled up with other work.

If a ticket needs a long conversation, it often wants splitting.

What a ticket really is

Most tickets still aren’t clear, and that’s fine. Writing a properly clear ticket takes time, and when there’s pressure on, we write a placeholder instead. A title, a line or two, maybe a link to a design. Really, it’s a promise to have a conversation later.

So here’s how I think about it. Every ticket is a promise that some work will get done. A placeholder carries a second promise: that some conversations will happen first.

Take a ticket that says “Add filters to the search results”. Which filters? Can you combine them? What happens when nothing matches? Does it need to work on mobile? Leaving those out when the ticket was written was reasonable. Each one is still a conversation someone will need to have before the work is done.

Those conversations happen anyway

There’s often a feeling in a team that talking about the work is expensive. Long meetings feel like waste. Doing gets praised, and visible progress is how people show they’re being productive. So placeholders go in, and everyone gets started.

But the conversations still have to happen. So they happen late, in small pieces, and often without the people who needed to be there. They start like this:

“I’m stuck on part of this ticket.”

“This isn’t what we were expecting.”

“I’m not sure how to test this part.”

By then the dev has built half of it, the client has seen something different from what they pictured, or QA is holding a ticket they can’t sign off. Each one starts a feedback loop at the worst possible moment. Someone has to stop and wait for an answer, and by the time it comes they’ve moved on to something else and have to find their way back in. The longer the wait, the more it costs.

Bring them forward

So I don’t push teams to write perfect tickets. I push them to have the conversations earlier, while they’re cheap.

In practice that means:

  • When a ticket’s being refined, ask what conversations it still needs, and who with.
  • Have them before the ticket’s marked Ready, rather than halfway through the build.
  • Put what you learn back into the ticket, so the next person doesn’t need the same chat.
  • If a ticket slips through without them, have them before anyone starts building.

A definition of ready that fits

This is where a definition of ready helps. Most teams write one that says a ticket must be clear before work can start. That’s a high bar, and under pressure, teams skip it.

There’s another way to write it. A ticket is ready when it’s clear, or when the conversations it needs have happened and what came out of them is in the ticket. It doesn’t need a perfect write-up. It needs the right people to have talked.

That’s a bar most teams can meet. And it moves those Thursday afternoon “I’m stuck” messages to Monday morning, when they cost a lot less.

Try adding one line to yours: “The conversations this ticket needs have happened, and what we agreed is in the ticket.” Then see what turns up on Thursday.

I’m Harry Bailey. I work with delivery teams that are tired of finding out on Thursday what they should have known on Monday. See how a team reset works.

Photo by mediha ekici on Unsplash

What Absence Reveals About a Team

They made it to day four of their holiday before their phone vibrated. When it did you could watch the whole decision happen on their face before they’d even focused their eyes on the screen. They knew who was calling. They knew why. And still, they answered.

I’ve had to make a phone call like that myself in past roles, to someone senior enough that only they’d know what to do with it. I’ve also been the one answering it, sat somewhere hot and hyperfocused on a good book.

The interruption itself barely registers now, looking back. What stays with me is how reasonable everything leading up to it had been. Nobody on the team had cut a corner. They’d followed all the guidance that existed, made a sensible call with the information in front of them, done exactly what a capable team does when the documented approach is lacking. It just wasn’t enough, because the details they needed had never been written down anywhere. It lived in one person’s head, and that person was hundreds of miles away with a paperback by a pool.

Continue reading →

Who owns delivery improvement?

As a consultant you can’t simply arrive, grab a cup of tea and start making a team deliver work better.

And it’s not just about understanding the context either.

There are too many moving parts for that, and the people leading delivery are usually the ones with the least capacity to absorb change — and the most connections to everything else that needs to move.

Thanks to my experience I’m relatively quick to understand the challenges of a delivery team in agency delivery. I’ve done it many times, I have a solid set of questions and measures that I run through. I’ve learned that each situation is unique, but the patterns repeat more than people expect.

I can create a solid status report with input from all the key people within a week or two. When I read it back to stakeholders everybody nods and agrees and gets excited for the opportunities. In their eagerness to get started, they ask what comes next.

And then, almost immediately everything slows, because change is hard, and an agency prioritising that change over billability is harder.

The understanding and the planning are easier to work through. The action is always where things stall. Meaningful improvement in a delivery team requires the following elements.

Continue reading →

The Move to Pods, Squads or Workstreams

I’ve sat in a few leadership meetings where someone gets up, draws circles on the whiteboard and starts sharing the team out between them. It nearly always comes after a good year, when the work’s solid and clients are happy, yet delivery takes more effort than it used to. That’s usually when moving to pods comes up.

A pod, or what some agencies call a squad or workstream, is a small, fairly stable group that looks after a set of clients or a stream of work from start to Done. Splitting the team that way promises clearer ownership and fewer people in every decision, and because it looks like a small change, it feels low risk. Most of the time leaders are after a bit of relief, and a big restructure is the last thing they want.

In plenty of agencies, teams start owning outcomes instead of tickets fairly soon after the split, and planning gets more grounded because the same people stay close to the work. Clients see the same faces for longer. Leaders spend less of their day hopping between conversations, even though the structure has barely changed.

Continue reading →