Where to start when everything is urgent

The idea

A fixed deadline doesn't ask you to work faster. It asks you to work in the right order. Only one chain of work actually sets your finish date; everything else has slack, whether or not it feels urgent.

So sequence on two questions. What unblocks the most downstream work? Do that first, even if nobody can see it. What is cheapest to be wrong about? Do the reversible, informative things early and the one-way doors late — but never before the evidence that should inform them.

Then say out loud which streams you are deliberately running in parallel, and what that coordination costs. Parallelism is a purchase, not a discount.

The plan, re-flowing

How it works

1. List the work with three facts each: duration, hard dependencies, and reversibility. Not importance — importance is what everyone is already shouting about.
2. Find the long pole. The longest dependency chain is your finish date. Every other stream has float: days it can slip without moving anything.
3. Front-load cheap evidence. If a discovery task has float and it informs a big commitment, moving it to day 0 costs nothing and buys information.
4. Name the one irreversible step. Contract signature, data cutover, public announcement, headcount offer. Place it after the evidence and as late as slack allows — early irreversible steps get defended, not reviewed.
5. Buy back time with chosen parallelism. Only parallelise a pair on the long pole, and quote the coordination price.
6. Report slack, not just the date. Zero slack is a finding, not a plan.

evidence-first, everything serial
  discovery         0 → 4     (was day 6; it had float, so free to move)
  load test         0 → 3
  sign contract     4 → 5     <- irreversible, now informed
  migration build   5 → 12
  reconciliation   12 → 17    (same finance/eng people as migration)
  comms prep        0 → 3     (float: 14 days — never the bottleneck)
  cutover          17 → 20
  finish = day 20   deadline = day 20   slack = 0

deliberate parallelism: reconciliation alongside migration
  coordination tax  +1 day each   (7 → 8, 5 → 6)
  migration build   5 → 13
  reconciliation    5 → 11
  cutover          13 → 16
  finish = day 16   slack = 4 days
  price = 2 person-days + one shared interface owner

Note what the first block admits: sequencing correctly pushed the finish from day 17 to day 20, because the signature moved three days later. That is the honest trade. You then buy the days back on purpose, in one named place.

When to use it

The trade-off: this optimises finish date and reversibility — not fairness, not visibility. The stakeholder whose stream you scheduled third still deserves to hear why, in their terms, before they find out from the chart.

Watch out for

• Starting with the visible thing. Comms decks, dashboards and demos are visible, reversible, and usually carry ten-plus days of float. Doing them first feels responsive and moves the finish date by exactly zero.
• Taking the one-way door early "to show momentum". Signing, announcing or migrating before the evidence lands converts a decision into a position you now have to defend. Ask: what will I know in five days that could change this? If the answer is "a lot", the step is scheduled too early.
• Treating parallelism as free. Every parallel track adds handoffs, a shared interface, and someone's divided attention. Parallelise the pair on the long pole, name the tax, and staff the interface — don't run five streams at once and call it urgency.
• Speeding up a task with float. Crashing a stream that isn't on the critical path burns budget and goodwill and changes nothing. Check which chain sets the date before you spend.
• Hidden resource dependencies. Two streams that need the same three people are serial whether your chart says so or not. If the chart shows them parallel and the calendar doesn't, the chart is a wish.
• Calling zero slack "on track". A plan that lands exactly on the deadline has already failed its first surprise. Report the slack number, then propose where you'd buy more.

Worked example

You're three weeks into an ops lead role. A payments provider migration must be live before the board demo on day 20, and four leads each tell you their stream is the blocker. The visible plan has merchant comms starting on day 0 and the provider contract signed on day 1 — "so procurement stops chasing us".

You map it instead. The long pole is sign → migration build → reconciliation → cutover. Comms carries fourteen days of float, so it is not urgent, it is just loud. Discovery and the load test carry float too — which means pulling them to day 0 is free, and they are exactly the evidence that should set the fee structure you're about to lock into a three-year contract.

So you move the signature to day 4, after the evidence. That costs three calendar days and puts you on day 20 with zero slack — which you say plainly, rather than reporting "on track". Then you buy the slack back in one named place: reconciliation runs alongside the migration build instead of behind it, both teams take a one-day coordination hit, and finish returns to day 16 with four days of buffer.

The sentence that lands in the interview: "The only irreversible step is the contract, so it goes after discovery, not before it. Comms is the most visible stream and has two weeks of float, so it goes third. I'm running reconciliation in parallel with the migration build — that costs about two person-days of coordination and buys four days of slack, and I'd rather spend it there than anywhere else."

Check yourself