Agile killed planning. Sprints are not a plan

I think agile killed planning when teams let sprints replace a dated release plan. A backlog can tell you what matters next without telling anyone when it will ship. Planning ahead is not waterfall. A plan you change every week is still a plan. A backlog you never put a date on is not.

Three runners circle an athletics track past a lap marker while a woman at the trackside holds a route map with flags and points to an open gate

I don't run sprints. I do change plans constantly, sometimes twice in a day. Changing a plan is part of planning. It does not excuse refusing to write one down.

What does agile actually say about plans?

The Agile Manifesto puts responding to change over following a plan among its four values. It does not say to stop planning. Following a plan is still valuable; it comes second when the two conflict.

Short sprints work when you cannot know what next month's work is. In that situation, a detailed plan is fiction. A team searching for what customers will pay for, a team fixing whatever support sends in today, a team whose priorities change every fortnight: for them, two-week cycles with a demo at the end are a better tool than any chart. They get feedback early, they waste less work, and they stop pretending to know things they don't. For those teams, sprints are the right call.

My problem is with what the idea turned into for everyone else.

How did responding to change become never giving a date?

A wrong date became an argument against giving any date. Estimates got called waste. The release plan was replaced by a backlog, the backlog was ordered by priority, and a priority was quietly allowed to stand in for a date. "It's at the top of the backlog" sounds like an answer. It isn't one.

The argument is "we don't estimate". I understand where it comes from: estimates get used as promises and then held against the team. But not estimating is a luxury for teams nobody is waiting on. An internal tools team whose users can wait, maybe. A team with a customer, a contract, a launch, a funding report or a sales pipeline behind it does not have that luxury. Somebody is waiting, and they will make a decision with or without your date. They will just make it with worse information.

What happens when sales asks "when will it ship?"

Sales needs a shipping date; a well-run sprint cannot answer that question on its own. Picture a twelve-person product team two years into two-week sprints. The board is spotless. Velocity is charted. Every retro ends with action items, and most of them get done. By every measure the team tracks, it is doing well.

Then sales asks when the integration a customer has been waiting for will ship. The answer is "it's in the backlog", which is a location, not a date. Sales goes back to the customer with "soon". The customer signs with a competitor who gave a date, even if that date turns out to be wrong.

In this hypothetical, nothing failed within the sprint process. Every sprint went fine. What failed is that nobody owned the question that spans more than two weeks, because the process was designed to stop asking it. That is the cost nobody puts in the retro: the deals, renewals and decisions that happened elsewhere while the team was being responsive.

How do I plan without sprints?

I use dated batches. Our content across three products runs in batches, not sprints. Each batch is a plan with a deploy date. Every page in it gets a written prediction before it ships, so the prediction can come out wrong, and the pages are read again 30 and 60 days later against what they were supposed to do.

That plan changes all the time. The plan for one October batch went from 18 pieces to 14 to 8 on the same day, after two rounds of changes. It was still a plan. At every point that day, everyone involved knew what would ship and roughly when, and what had been cut. Nobody had to read a backlog to find out.

That is the whole difference I care about. Changing the plan is responding to change. Refusing to have one is not.

When does a dated plan save a project?

Whenever something outside the team has a date of its own. Say a team has to move off a tool whose contract ends on 31 March. The vendor takes two weeks to export the data. Testing the import takes two more. The people using the tool need a week where both systems run side by side.

That is five weeks of work that has to finish by 31 March, which means it has to start in late February at the latest. Add two weeks to choose the replacement and a little room for the vendor being slow, and the move really starts in early February.

In this scenario, a team that plans one sprint at a time finds this out in March. Nothing about the work is hard. The sequence is just longer than a sprint, and nobody was looking further ahead than one. Working backwards from a fixed date takes ten minutes, and it is the ten minutes that decides whether the project is calm or a scramble. If you have a move like this coming, a software rollout schedule gives you a place to sketch the sequence.

The same shape turns up everywhere: a grant deadline, a conference, a lease, a seasonal peak, a regulator. The work around those dates can be as agile as you like. The dates themselves will not move because your process prefers not to name them.

What did agile get right?

Short feedback loops, small releases and changing the plan when the facts change: I would keep all three. Short feedback loops are better than long ones. Shipping small pieces beats shipping one big one at the end. Changing the plan is not a weakness; it is the point of having a plan you can read.

What it got wrong was treating a date as a sin. A date is not a promise carved in stone, unless you let people use it that way. It is the team's best current guess, written down where everyone can see it, so that the people waiting can plan their own work. When the guess changes, you change the date and say so. That is far more honest than "it's in the backlog", which hides the same uncertainty and adds a little extra on top.

Agile and dates also sit together better than both camps admit. A release chart alongside sprints works well for teams that want to keep their sprints and still answer "when". So does planning in rolling waves, where the next few weeks are detailed and the months after them are rough on purpose.

What is the smallest plan worth having?

The smallest useful plan shows the three to five things that have to happen before the date, in order, each with a rough week, on one screen. That is all. If it does not fit on one screen, it is not a plan yet. It is a task list wearing a plan's clothes.

Two walkers on a winding meadow path marked by five coloured flags, heading toward a finish post with a ribbon

The order matters as much as the dates. A list of priorities tells the team what it wants to do first. A sequence tells the people waiting what has to finish before their part can begin. That is the question a backlog leaves unanswered, even when every card is in the right column.

A few rules keep that sequence useful:

  • Use weeks, not days. A rough week is honest about what you know. A precise day three months out is not. If you are unsure how long something takes, a range estimate is better than no estimate.
  • Put the outside dates on it first. The contract end, the launch, the review. Those are the fixed points everything else hangs off.
  • Link the steps that depend on each other, so that when one slips you can see what moves with it.
  • Look at it every week and move what changed. A plan you redate every week is still a plan. One you never look at is decoration.

Any tool can hold this: a whiteboard, a spreadsheet, the back of a notebook. A one-screen Gantt chart is what Ganttile is built for, but the tool matters far less than the habit of writing the dates down and looking at them again next week.

What does putting dates on a plan cost?

Being wrong in public. Dated plans are wrong a lot, ours included. The price of keeping them is redating them every week, and admitting a slip early, in front of the people who are waiting. That is less comfortable than "it's in the backlog", which can never be wrong because it never says anything.

I think the discomfort is worth paying for. A slip you announce in week three is a conversation. The same slip discovered in week twelve is a crisis, and usually somebody else's. Tracking delays as they happen is most of the work, and it only works if there was a date to slip from in the first place.

So did agile kill planning?

Not on purpose. The manifesto asked teams to value responding to change more than following a plan, and a lot of teams heard "stop planning". Keep the sprints if they work for you. Keep the retros, the demos and the small releases. Then add the one thing they were never meant to replace: a short, dated list of what has to happen before the date that matters, which you change every week and are willing to be wrong about.