How to set up agile in a marketing team, and what to do when it isn't working

Set it up as a project with a start, an end and someone teaching people how it works, and it takes a fortnight. Let it arrive as a set of recurring calendar invitations nobody explains, and you get the meetings and none of the benefit, which is the state most teams asking this question are already in. This is the second part of our guide for people running an in-house marketing team; the first part covers what squads, scrums and agile actually are and whether you need any of it.

Our interest, stated first. We do fractional and part-time marketing leadership, which often means fixing how a team works, so a piece saying this is worth doing is a piece that suits us. Everything below is something a team can do for itself, and the last section is about when not to bother.

Day one: the smallest setup that works

You need four decisions and one afternoon.

1. Pick the length of your cycle, and never move it. Two weeks suits most marketing teams, because campaign work rarely finishes in one and attention wanders in four. Whatever you pick, the point is that it ends on a known day.
2. Make the work visible in one place. One board, every piece of work on it, three columns at most to begin with: to do, doing, done. The tool matters far less than everyone using the same one.
3. Agree what "done" means, in writing. This is the decision that saves the most arguments later. Done means published, or approved, or live, and it is the same for everybody.
4. Put two meetings in the calendar and no others. A short one at the start of the cycle to agree what goes in it, and a short one at the end to look at what happened and name one thing to change. Add a 15-minute daily or twice-weekly catch-up if people genuinely don't know what others are doing, and drop it the moment they do.

Then run it for three cycles before you change anything, because the first one always feels worse than what you had.

Teach it, or it will fail

The thing that decides whether this works isn't the method. It's whether anybody explained it.

I've worked in places where there was no explanation of how agile, scrums or standups were done at all. You were thrown in and left to drown: no onboarding guide, no principles, no explanation of how they used Jira or what they were doing with Git. The onboarding was awful, and the people suffering most were the ones who most wanted to do it properly.

The good version looks the opposite. At PlayStation we set up agile squads with an agile coach, running across regions and teams, from experiment marketing through to CRM and engineering, with people in the US, Japan, Europe and the UK. We stood it up very quickly and learned as we went, but we had a defined period to set it up and we did it properly. It worked really well. The coach was the difference: somebody whose job was to teach the thing rather than attend it.

If you have nobody who can teach it, buy a few days of somebody who can, with a brief that says they leave at the end. An open-ended coaching retainer is how a two-week change becomes a permanent cost.

What a well-run cycle actually looks like

At NatWest, working on digital product, we had the sprint cycle nailed. It was run rigidly and properly, and that was the impressive part: the cycle was genuinely fixed, so what went into it was a real decision and what came out of it was real work. Rigid is the right word, and most teams that think they're doing this are not.

At King, everyone in every team knew their role: UX, marketing, engineering and the supporting teams. There was a project planner, and the plan was something people actually worked to. Where timelines were pressured, things did slip and agreements were missed, but that happens anywhere, and it's worth saying plainly, because an account where nothing ever slips is an advertisement rather than a description.

The four mistakes that break it

Adding ceremonies without removing anything. The sprint planning goes in, the standup goes in, the retrospective goes in, and the old weekly status meeting, the monthly review and the project update email all stay. Now the team has less time to work than before. Take something out for everything you put in.

Treating the board as a report for management. The moment the board exists to be looked at upwards rather than used sideways, people start managing how their work appears on it. You'll know this has happened when nothing is ever "blocked".

Never doing anything with the retrospective. A review that produces no change is a complaints meeting, and people stop bringing anything real to it. One change per cycle, written down, with a name beside it, is enough.

Copying a structure instead of a habit. Squads and tribes were one company's arrangement at one moment, and even its own authors called the write-up a snapshot, as the first part of this guide sets out. Borrow the habits, which are small. Leave the organisation chart alone until the habits work.

If you already have it and it isn't working

Run these five checks before changing anything, in this order:

1. Does the cycle actually end? If work routinely rolls over, you don't have cycles, you have a queue with ceremonies attached. Cut what goes into the next one by half and see what happens.
2. Can anyone say what "done" means? Ask three people separately. If you get three answers, start there.
3. Who decides what goes in? If the answer is "whoever asks loudest", the problem is prioritisation, not process. It's usually the budget conversation in disguise, and how to split a small marketing budget is the version of that argument with numbers in it.
4. Is anything ever dropped? A system that only adds work isn't a system.
5. Did anyone explain it to the newest person? If not, fix that this week. It costs an hour and it's the cheapest thing on this list.

If all five are fine and it still isn't working, the problem probably isn't the process at all. Unclear strategy, missing skills and a changing brief all survive any method you put on top of them, and whether your marketing is actually working is a different diagnosis with different questions.

When not to bother

A team of two or three that sits together and talks all day does not need a framework. Neither does a team whose work is genuinely continuous, such as one person running paid search day to day. The test isn't team size, it's whether you keep being surprised: by what people are working on, by what's late, or by what turned out not to matter. If nothing surprises you, you already have whatever this was going to give you.

If you want somebody to run the setup and then leave, ask for a free call. Fractional and part-time marketing leadership is what that arrangement looks like, and where the first job is working out what the team should be doing at all, our 90-day marketing plan is that.