The morning I planned wrong and lost nothing
I plan every working day in writing, and I reconcile every evening against what actually happened, which is how I caught this. One recent morning the day looked overloaded, so I made a judgment call and moved three tasks off it, onto the following week. Sensible, defensible, done.
By close of that same day, all three had shipped anyway.
So the plan was simply wrong. My morning read of my own capacity undercounted what the day could hold. And here is the part worth an article: the error was completely free. Nobody was let down. No date slipped. The three tasks were finished ahead of the week I had just promised them into. If I had not run the evening reconcile against what actually shipped, I would never even have known the plan was wrong, because a day that quietly does more than planned produces no complaint from anyone.
Now run the mirror image. Suppose I had read the morning the other way, kept all three on the list, promised one of them to a client for that afternoon, and then the day had gone the way overloaded days usually go. Same size of estimating error, opposite direction. That version is not silent and not free. That version is a missed commitment, in writing, in someone's inbox.
Same mistake. Wildly different price. That asymmetry is the whole discipline.
The asymmetry nobody plans around
Every estimate you make about your own capacity will be wrong. The only real choice you get is which direction it is wrong in, and the two directions do not cost the same.
- A conservative plan's failure mode is early delivery. The restock lands two days before the date on the site. The launch goes live Thursday when you said "next Monday." The customer experiences you as ahead of your word.
- An optimistic plan's failure mode is a broken commitment. And a broken commitment is loud: it gets screenshotted, quoted back, and remembered. It also compounds, because the apology email, the support tickets, and the rework of the schedule all eat the very capacity that was already short.
Merchants price this asymmetry instinctively when buying, and forget it entirely when promising. The same owner who distrusts a supplier's rosy lead time will put an unbuffered launch date on a countdown banner.
Why plans come out optimistic by default
Three reasons, all of which I have caught in my own planning:
You plan from one vantage point. A plan is built from what you can see at the moment you write it. My three-task morning proved the view is incomplete even about my own day; it is far more incomplete about anything involving other people. You cannot see your developer's other obligations, the freight forwarder's backlog, or the platform's review queue. What you cannot see, you do not count, and what you do not count makes the plan lighter than reality.
Best-case math feels like the honest answer. "How long will this take?" quietly becomes "how long would this take if nothing interrupts it?" Nothing-interrupts-it days are rare enough that planning around them is planning around an exception.
Some of the clock is not yours. A launch that waits on an app store review, an ad account approval, or a carrier's first scan includes stretches of time you do not control and cannot compress. Committing a date that includes someone else's queue is promising on their behalf.
The buffer discipline I actually use
This is not a vague "add some padding." I use fixed multipliers on every estimate before a date leaves my keyboard, and they are deliberately unheroic:
- 1.3x for routine work I have done many times.
- 1.75x for work with unfamiliar parts.
- 2x for anything where a third party sits in the critical path.
The date that goes to the customer is the buffered one. The unbuffered one stays private, as the target I work to. When the work lands on the internal date, the customer gets it early and reads it as competence. When something goes sideways, the buffer absorbs it and the promise holds. Either way the commitment survives, and the commitment is the asset.
Translated to a store's three most-broken promises:
- Restock dates. Take the supplier's estimate, add the direction you know their errors run (they are selling you a lead time, so their errors run optimistic), and publish the date you would genuinely bet on. "Back in stock March 18" landing March 14 is a small marketing win. The reverse is a refund queue.
- Launch dates. Never publish a date that contains a queue you do not own. If a platform approval sits in the path, announce after it clears, or phrase the promise around what you control: "preview goes out to the list this week, launch date announced the moment review clears."
- Campaign timelines. Plan the creative, the landing page, and the offer backward from a date one buffer tier earlier than the real deadline. A campaign finished early can wait. A seasonal campaign finished late is inventory that missed its season.
One more habit makes the loop honest: reconcile against what actually shipped, not against the plan or your memory of the day. My wrong-in-the-safe-direction morning was only visible because the evening review reads the record of completed work, not the morning's list. Without that, optimistic planners never learn they are optimistic; they just experience recurring bad luck.
The trust math
Under-promising is sometimes framed as sandbagging, as if the honest thing were to publish your best case. I think that gets honesty backwards. The best case is the least likely single outcome; publishing it as a commitment is the least honest date you can offer. The buffered date is the one you can actually stand behind.
And the exchange rate is brutal in one direction. Deliver early ten times and you earn a reputation as someone whose dates mean something. Miss once, visibly, and that is the incident the customer retells. Trust built in drops, spent in buckets. A plan that errs safe makes deposits on every miss. A plan that errs hopeful makes withdrawals precisely when you can least afford them.
The bottom line
You do not get to choose whether your plans are wrong. You get to choose the direction, and the directions are not priced the same. Wrong-but-safe costs you a pleasant surprise and a little scheduling pride. Wrong-but-hopeful costs you the thing your margins ultimately sit on, which is whether people believe your dates. Buffer by rule rather than by mood, promise the conservative date, keep the aggressive one as your private target, and reconcile against reality so the rule keeps learning.
If your store keeps finding itself apologizing for dates, the fix is usually in how the promises get made, not in how hard anyone worked. Grab any slot that works and I'll help you rebuild the promise pipeline so the surprises land on the right side.
