Your Team Knows What Happens to Sales After a Deal Ends

How to Write Down What Your Team Expects and Check It Against Every Deal

This series follows one team through a year of pricing and promotion decisions. I made up Larkfield Foods, Kestrel Market and the people in this story, and the numbers come from a simulated market. This first scene comes from my work in 2019 as a trusted technology advisor to a consumer packaged goods manufacturer, helping its account teams plan trade promotions.

Updated October 18, 2026: we tuned the simulated market behind this series to match published industry figures, so some numbers here changed. Nina now has 40 promotions instead of 39, the week after a deal sells about 95% of normal instead of 91%, and Jev's probabilities moved a little. Its answer and the team's decision stayed the same.

Rewritten October 18, 2026, to match the rest of the series: a shorter scene and a workbook you can fill in with your own deals. The numbers are the ones in the update above.

---

Your Best Planner Already Knows What Happens After a Deal

Many teams agree their promotion plan with the retailer before the last deal's results come in. The plan then rests on what the team already believes about its shoppers, and few teams write that belief down. This post shows how to write it down and let each deal's results correct it.

At Larkfield Foods, Priya Raman has planned the Sunny Oats deals with Kestrel Market for eighteen planning seasons. She was copying last year's plan forward when Nina Alvarez, the revenue analyst, asked her what happens to sales after a deal ends.

"They're back to normal right away," Priya said. "The deal lifts the week it runs, and the next week looks like any other."

"How sure are you, from a hunch to near certainty?"

"Pretty sure. I've watched this account for years."

"Then let's check it against the last three years," Nina said.

Sales Don't Bounce Back Right After a Deal

Nina had all 40 Sunny Oats deals at Kestrel from those three years. After 36 of them, the next week sold at full price. For each of the 36, she compared that week's sales with the full-price weeks before the deal. Then she let each result move Priya's belief a little, in the order the deals ran.

Priya's belief that sales go back to normal once a deal ends, checked against 36 promotions. The gold line starts where she started and ends near 95% of normal (Simulated market).

"The dashed line is your belief on its own," Nina said. "It stays at normal whatever the deals do. Each lavender dot is one deal, and they jump from 73% of normal to 128%. Read one at a time, they'd change your mind after every deal."

"And the gold line?"

"That's your belief, moved by each result as it came in. It starts from 100%, where you started, and the first deal moves it to 98%. After four deals it sits at about 97%, and by the last one it settles near 95%, with the band around it getting narrower."

Priya studied the chart. "Then the week after a deal sells about one box in twenty less than I thought."

"Across 36 deals, yes. Each deal moved it a little, and together they moved it about five points," Nina said. "The real dip may be bigger. Some of the full-price weeks I called normal came right after an earlier deal, when shoppers may still have had boxes at home. And we often ran a deal because sales had gone soft, so the weeks before it ran low too."

Decide How Sure a Model Must Be Before It Plans a Deal

Nina then asked Jev, a decision model that returns a choice with a probability for each option, what to plan for the next open slot on Kestrel's calendar. She gave it only what the team could see: recent weekly sales and prices, which weeks had a deal, and the belief from the chart. It got no margin or cost data.

"It leans toward holding the regular price," Nina said. "Hold at 0.49, a modest deal at 0.30, a deep deal at 0.21. But it also scores how much it trusts that answer, and that came back at 0.23. The team set 0.80 as the bar for acting without a person, so our rule sent the call to us."

Riley Shaw, who runs revenue growth and signs off on the plan, made that call and kept the regular price for the slot. Riley also asked Priya to write down two things for each deal she plans: what she expects the week after it to sell, and how sure she is. Then each result can move her number the way the chart moved it.

Write Down What Your Team Expects Before You Plan the Next Deal

Download the belief-check workbook (Excel) below. Fill in what your team expects the week after a deal to sell, as a share of a normal week, how far off you think you could be, and how far one deal's result can miss on its own. Then add one result per deal, oldest first: sales in the full-price week right after the deal, divided by the average of up to four full-price weeks before it. The workbook moves your belief one deal at a time and shows the range around it. Larkfield's 36 simulated deals come filled in, so it opens on the chart's numbers.

belief-check.xlsx

Download XLSX • 19KB

If your belief moves, plan the weeks after your next deals on the new number, and add each deal's result as it comes in.

What No Tool Here Could Do

Jev gave a clear answer for one slot and scored how much it trusted that answer, low enough that the team's rule sent the call to a person. It didn't say how that choice changes the next slot, or what the team would learn from the result. Nina moved the belief herself, one deal at a time, and nothing carried the new number into the next plan.

Tom Becker from finance read the chart over Nina's shoulder and asked a question it couldn't answer yet. Every Sunny Oats deal at Kestrel Market lifted sales in the week it ran, so why did the team keep needing another one?

If your team holds a belief about promotions that nobody has checked, tell me in the comments. I'd like to hear it.

Where the Numbers Come From

Every number is a named fact computed from the simulated market. Each deal's result comes from the scan data and the trade plan: sales in the full-price week after the deal, divided by the average of up to four full-price weeks before it. The belief starts at normal, with "pretty sure" set as a typical miss of about five points, and each result moves it by the standard rule for combining a belief with new evidence. That rule assumes one deal's result can miss by about 10 points. The 36 results scatter more, about 15. With a 15-point miss, the belief ends near 96%, and the range runs from about 92% to 100%, just reaching normal. Jev's answer is one real recorded call. It has no track record on this market yet, so read its confidence as a signal. The market falls within published ranges, such as the dip after a deal, and its model card lists the sources.