Tentpole events in retail: reading the sales curve

What tentpole events are

Tentpole events are the handful of dates each year that a category plans around because they move enough volume to hold up the rest of the schedule: the Super Bowl, Memorial Day, the Fourth of July, Labor Day, Halloween, Thanksgiving. The name comes from the film business, where one big release supports a studio's whole slate, and it carries the same meaning in retail. For a convenience operator, the week of the Fourth is not a good week. It is a week that can run 71% above the trailing average in salty snacks and then hand a fifth of that back over the three weeks after.

I planned against six of these a year at a natural-products brand, and the mistake I made twice was reading the peak and stopping there. The peak is the easy part. The shape around it is where the money and the leftover inventory both live.

The shape of the curve: build, peak, cliff

A tentpole event does not produce a spike. It produces four phases, and each one needs a different decision.

• Build. One to three weeks ahead, as shoppers stock up. Starts earlier every year for the bigger moments, and it is where distribution and in-stock matter most, because a shopper who cannot find the item in the build week buys the competitor and does not come back for the holiday.
• Peak. The event week itself. Almost entirely an execution question by this point. The assortment is set, the price is set, and the only variable left is whether the shelf is full.
• Cliff. The week after, which runs below the pre-event baseline rather than back to it, because the pantry is loaded.
• Recovery. Two to four weeks of drift back toward normal.

Here is a 40-store convenience operator's salty-snack units across the seven weeks around the Fourth of July, indexed to the trailing eight-week average of 41,600 units a week.

Measuring a tentpole honestly

Take the window above and do the arithmetic two ways.

Count only the build and the peak, and the event delivered 49,400 units above baseline: 3,300 in the minus-2 week, 16,600 in minus-1, and 29,500 in the holiday week itself. That is the number that gets presented.

Now include the cliff. The three weeks after ran 7,900, 2,200 and 400 units below baseline, giving back 10,500. Net incremental across the full seven-week window is 38,900 units, which is 13.4% over a flat baseline of 291,200.

The gap between 49,400 units and 38,900 is 21% of the reported gain, and it is not error. It is pull-forward: volume the operator would have sold anyway, moved forward a week or two by a display and a price. Measure the event on a window that includes the recovery or you will overstate every tentpole by roughly a fifth, and you will keep funding the ones that only move volume in time rather than creating it.

Two more rules that come out of the same table:

• Index to a trailing baseline, not to last year's same week. The retail week a holiday falls in shifts. On the 4-5-4 calendar a 53rd week moves every subsequent week's year-over-year partner, so a holiday can land in week 22 one year and week 23 the next. Both weeks then read wrong, one up and one down, and neither is telling you anything about demand.
• Split the categories. In the same seven weeks, a foodservice set can peak at 120 while salty snacks peak at 171, because one is a stock-up occasion and the other is a trip occasion. A store-total index averages the two into a number that describes neither.

Planning against the build, not the peak

The peak week is too late to influence. Everything that decides it happens in the ordering cycle before the build, which for a distributor-supplied convenience account means the buy is committed two to four weeks out. So the operational questions are all early ones: which SKUs earned the incremental facings last year, how deep the build ran by store cluster, and what the recovery weeks did to inventory.

The last one is the expensive one. A holiday-specific pack that does not clear in the cliff weeks is markdown by the plus-3 week, and the margin taken there routinely erases the gain from the peak. When I reviewed a summer program that had reported a strong Fourth, the net after post-holiday markdown on themed packaging was slightly negative. The peak had been real. The plan had bought for the peak and ignored the cliff.

Not every tentpole is a national holiday, either. Regional and seasonal moments (the first week of a school year, the opening of a local sports season) behave the same way at a smaller amplitude, and they are usually unplanned because nobody has built a calendar entry for them. Those show up in store-level POS as a repeating annual bump long before anyone names them.

Prime Day belongs on any list of tentpole events, and it is the one moment on this page we will not put a number against: Scout holds no Amazon data of any kind, so anything we said about its magnitude would be someone else's figure wearing our name.

Where Scout fits

The whole read above is week-level POS with a baseline and a window: build, peak, cliff, recovery, by category and by store. Scout computes that curve on the retailer and first-party POS data it already holds, so a tentpole gets measured across the full window rather than on the peak week alone, and last year's curve is available as the plan for this year's build.

Two boundaries. Scout reads transactions, so it can show that the plus-1 week fell 19% below baseline and cannot tell you which shoppers stayed home or why. And nothing here covers Amazon, marketplace or ad-platform activity, which is not data Scout holds.

The short version

• Tentpole events are the few dates a year big enough to plan a category around. They produce a build, a peak, a cliff and a recovery, not a spike.
• Measured on build plus peak, one operator's Fourth of July looked like 49,400 incremental units. Across the full seven-week window it was 38,900.
• About a fifth of a typical tentpole gain is pull-forward. Include the recovery weeks or you overstate every event.
• Index to a trailing baseline rather than last year's same week, because retail-calendar shifts move which week the holiday lands in.