Planning without fortune-telling

Waterfall had gates because a mistake caught late was expensive to unwind.

Agile had two-week sprints because a plan more than two weeks out was usually wrong anyway, and the cadence was a way of admitting that on a schedule instead of by accident. DevOps built a pipeline so the expensive part, shipping, could be protected from everything upstream that wasn't ready yet.

Different eras, same constraint. Building the thing was slow, so the plan's job was to sequence carefully and keep waste away from the expensive step.

That constraint is loosening.

Teams using AI-assisted development finish committed work in two-thirds the time they projected, then hit a wall that has nothing to do with code. The backlog wasn't ready to absorb the freed capacity.

Dependencies unresolved. Decisions not made. Specs that were fine as placeholders when nobody would touch them for six weeks, and are useless the moment somebody can start building tomorrow.

The scarce resource didn't disappear. It moved.

It used to be "can we build this fast enough." Now it's "do we know what outcome we're going for, and how will we know if we got it." Most planning processes were never built to answer that, because they never had to.

You can watch the mismatch happen in the planning meeting itself.

Engineers move faster than the cadence can track, so half the room's time goes on re-litigating work that already started. Sometimes work that already shipped.

Everyone nods along to the remaining items because nobody present has the context to evaluate them. The real decisions were made asynchronously, days earlier, by whoever was closest to the code.

That's not a planning meeting. It's a ceremony producing the appearance of a decision.

Fixed date, fixed scope, fixed quality: pick one

Every initiative has three variables you'd like to hold constant. When it ships. What it includes. How well built it is.

You can commit hard to one.

The other two have to stay free, because something always absorbs the uncertainty that turns up mid-project. Lock all three and what absorbs it is the team's health, or a quiet decision to ship something broken.

Leadership needs a date locked for a board commitment. Fine, legitimate reason to fix the date.

But then scope has to flex, and everyone from the PM to the engineer writing the code needs to know going in that features get cut before the date moves.

Now a different initiative, where the whole point is a specific customer promise, a specific integration, a specific migration path. There scope is fixed and the date is the release valve.

Both coherent plans. What isn't coherent is agreeing to a fixed date and a fixed scope and calling the result a commitment. That's a guess wearing a commitment's clothes, and everyone downstream treats it as load-bearing.

The failure isn't ambition. It's silence about which lever is the release valve.

Say it out loud, per initiative, and half of what looks like a broken planning process turns out to be a process nobody ever asked to pick.

Confidence is a number, not a decoration

A date eight weeks out and a date eight months out should not sound the same in a meeting.

They usually do, because most planning language has exactly one register. Certain.

Attach a confidence level to the commitment itself. Not a caveat in a footnote. An actual number or band beside the date: near-lock, reasonable bet, placeholder we'll firm up in six weeks.

Costs almost nothing to say, and it changes how every other team plans around you, because they stop building their own commitments on top of your guesses as though the guesses were facts.

Same honesty is why quarterly planning shouldn't eat two weeks.

A day, maybe two, is enough to set outcomes, assign confidence, and name which lever is fixed per initiative. Longer than that and the meeting isn't getting closer to accuracy. It's getting more comfortable with a number that was never going to be right.

Two weeks of ceremony doesn't buy a better plan. It buys a plan everyone feels more attached to, right before reality edits it.

And reality will edit it.

New information arriving mid-quarter and forcing a replan isn't the plan failing. It's the plan doing its job, which was never predicting the quarter. It was getting everyone aligned enough to move and honest enough to adjust.

Treat the first replan of a quarter as a sign the process works. Treat a quarter needing zero adjustment as the surprising outcome, and go and ask what got hidden to make it look that clean.