Where we left off#

• Week 7: placement decides how far a shopper walks
• the short-distance position is the most valuable in the shop
• so something is built there, deliberately, to be looked at
• this week: what that thing is arguing

Learning objectives#

By the end of this lecture you should be able to:

• audit an endcap against price per standard drink
• state the direction and strength of the correlation found
• name the framing devices that shift a comparison
• treat a “special” as a claim rather than a discount

The audit#

Take an endcap. Take the aisle directly behind it. Compute both.

The reading did this across four chains and found the aggressively merchandised stock returned worse value on the standard-drink metric than unremarkable mid-shelf neighbours.

Not equal. Worse.

Why “negative” is the stronger claim#

• no correlation would mean displays are uninformative
• negative correlation means they are systematically anti-informative
• the second cannot happen by accident at that sample size
• it implies the display is selected for something other than value

Three framing devices#

• was/now pricing — anchors against a figure you cannot verify
• multibuy — moves the unit of comparison from price-per-drink to price-per-transaction
• category adjacency — sets a poor ratio beside a good one so the pair averages in the shopper’s head

Each is legal, ordinary, and effective. None is a discount.

What the display is actually for#

On this evidence, an endcap functions less as a discount signal and more as a visual argument against doing the arithmetic — the arithmetic this course has spent seven weeks building the tools for.

The skill this week teaches is noticing the argument is being made.

The uncomfortable part#

Before you compute anything, you will expect the endcap to be better value.

• almost everyone does
• that expectation is the product being sold
• being wrong here is evidence about the display, not about you
• record it honestly in your working — the lab asks you to

Common errors#

• auditing the endcap without auditing the aisle behind it, leaving nothing to compare against
• treating a was/now price as a verified prior price
• comparing a multibuy total against a single-unit price
• concluding retailers are lying; the reading claims a selection effect, which is a weaker and better-supported claim

Where this connects#

• week 9 stops doing this one display at a time
• the week 8 lab is the paired audit, endcap against aisle
• the Event Portfolio will be pitched to a room that has done this exercise and will ask where your inventory came from

Next week#

Doing it at scale: a hundred SKUs, and the failure modes that corrupt a dataset without announcing themselves.