Put the two conversations on separate calendars. This is the whole game and almost everybody gets it wrong. When one weekly meeting owns both the number in front of you and the plan behind it, one of two things happens: either next year's strategy gets eaten by this week's fire, or your Q4 execution meeting turns into a 2027 philosophy debate and nothing ships. Different meeting, different day, different pre-read, different attendee list. Say out loud in the invite which clock the room is on.

Rebase the back half on the basket you actually have. Not the one you planned in February. Pull the last eight weeks of transaction data and look at trips per customer against average ticket. Right now across the industry those two lines are moving in opposite directions: more visits, flatter or smaller baskets. If your Q4 forecast assumes the stock-up trip and your customer is buying for the next three days, you are about to over-buy pack sizes and over-spend on threshold mechanics. Rebase the media plan against the same read, because a campaign optimized for basket size is aimed at a behaviour that isn't showing up.

Publish a kill list, and publish it by name. Anything that cannot put revenue on the board by December 15 moves to the 2027 plan or dies this week. The reason to publish it is not tidiness. It's that half-dead initiatives don't stop consuming your best people, they just stop reporting progress. I would rather formally kill four things in September than watch nine things limp into November while my A players split their attention across all of them.

Audit the price cuts you already took. Pull the top 25 items where you invested price this year and check whether the unit lift actually paid for the margin you gave up. Anything that didn't, restore it before holiday sets lock, because after that the price is frozen until spring. This one is worth real money and almost nobody does it, because taking a price down feels like strategy and putting it back feels like an admission.

Recalculate cost-to-serve before peak, not during it. Run the math at today's average order value, not last year's. Find the order size where fulfilment costs more than the order makes, then move your free shipping threshold to sit above it. If digital is growing faster than stores in your business, and for most of you it is, then every point of digital growth at the wrong basket size is quietly buying you margin erosion. Fix the threshold in September or eat it in December.

Strip the one-timers out before you build a baseline. This year has noise in it. Tariff refunds, one-time settlements, price investments funded by money that arrives once. If you build 2027 off this year's reported P&L, you are planning against a number that structurally will not repeat, and you will spend Q1 explaining a gap you created in October. Build off a two-year stack with the one-timers pulled out, and show your CFO both views on the same page.

Pick three bets, not fifteen initiatives. Every plan I've ever seen with fifteen priorities delivered four. Choose three, attach a named owner, a dollar figure, and a kill date to each, and let everything else be run-the-business. The kill date is the part people skip and it's the part that makes the whole thing honest.

Decide your capability gap now, because hiring takes six months. Whatever your 2027 plan requires that your team cannot do today is a Q4 decision, not a January one. Build, buy, or partner. If the answer is build, the requisition goes in this fall. If it's partner, you should be having conversations before the December freeze, not after the plan is approved and the budget is already spoken for.

Write down your peak decision rights in September. Who can take a markdown without asking. Who can authorize extra labour hours. Who calls the audible on Black Friday weekend when the plan is wrong by 11am. What dollar threshold requires you personally. I have watched more holiday margin disappear to a slow approval chain than to a bad buy. Put it on one page, send it to the team, and then honour it when someone uses it.

Have the retention conversation in October. Your best people get recruited in January, right after bonus lands. The conversation that keeps them is the one about what their role looks like in next year's plan, and it has to happen while you're still writing that plan, because that's when you can actually put them in it. It costs nothing. Waiting until February costs you the person and the six months it takes to replace them.

Start building the content strategy that wins agentic search. This is the one item on the list that lives on both clocks at once, and it's the one I'd fight hardest for in the budget conversation.