Halo and pantry effects around a promotion
A promotion changes behaviour beyond the item and beyond the week. It can lift adjacent items in the same brand, depress them, pull purchases forward from future weeks, and leave a trough behind it. Measuring only the promoted item in the promoted period captures one corner of that and reports it as the whole effect.
What is a halo effect?
Additional sales of items other than the promoted one, attributable to the promotion. A brand's promoted flavour can lift its unpromoted flavours if the promotion brings shoppers to the shelf; it can also suppress them if shoppers simply switch.
Both directions occur and neither can be assumed. A brand that reports halo without measuring the suppression case is reporting the half it likes.
How is halo measured?
The same way as any incremental effect: the other items' actual sales against an estimated base, in stores running the promotion, compared against stores that did not. The arithmetic is identical to the main effect and the signal is smaller, which means noise matters more.
Smaller effects need more stores or longer windows. A halo estimate from a handful of stores over one week is usually indistinguishable from noise, and should be reported as such.
What is pantry loading?
Shoppers buying more than they need because the price is attractive, then not buying again for longer than usual. The promoted week's volume is genuine; part of it was borrowed from the weeks after.
The effect is larger in storable categories and smaller in perishables and in small-format stores where shoppers buy for immediate consumption. That channel difference is significant and frequently overlooked when a large-format read is applied to independent retail.
How long does the dip last?
As long as the loaded quantity takes to consume, which varies by category and by how deep the promotion was. A deeper promotion loads more and produces a longer trough.
The practical implication is the measurement window. If the window ends when the promotion ends, the trough sits outside it and the net effect is overstated by whatever was borrowed.
What window should be used?
Long enough to include the recovery. That means the promoted period plus a post period of comparable or greater length, with the base estimated across both.
Agreeing the window before the promotion runs is what stops the window being chosen afterwards. Any figure chosen after the numbers arrive is a rationalisation, however reasonable it sounds.
Does the channel change the answer?
Substantially. Storage space at home and shopping frequency both differ by shopper and by trip type, and the independent channel skews toward immediate-consumption trips. A promotion in that channel loads less and dips less, which is a genuine structural difference rather than a measurement artefact.
That is also why a read from one channel does not transfer to another, and why channel-specific measurement exists.
What should be reported?
Promoted-item incremental, halo across the brand, the post-period effect, and the net across the full window. Four numbers, and only the fourth answers the question the promotion was funded to answer.
Frequently asked questions
Is halo always positive?
No. Switching within a brand is common and produces negative halo, which is a form of cannibalisation. Reporting only positive halo is selective rather than wrong.
How do I separate halo from a general category lift?
With a comparison group. Stores without the promotion show whatever the category was doing anyway, and the difference is the attributable part.
Does pantry loading apply to perishables?
Much less, by definition, though it applies to anything with meaningful shelf life. Frozen and ambient goods load; fresh goods largely do not.
What if the post period contains another promotion?
Then the window is contaminated and the net effect cannot be cleanly measured. That is a planning problem as much as an analytical one, and it argues for spacing activity deliberately.
Should the dip be counted against the promotion?
If the aim is to measure net effect, yes. If the aim is to measure in-period volume for a supply plan, the gross figure is the relevant one. Say which question is being answered.
How does this interact with a competitor's promotion?
A competitor promoting in your post period will deepen your trough for reasons unrelated to your activity. Observing competitor timing is part of a credible read.
Can halo be planned for?
It can be tested. Running the same promotion with and without adjacent support and comparing is a cleaner way to learn than modelling a single execution.
What is the most common reporting error?
A window that ends with the promotion. It is the default in most reporting tools and it systematically flatters every promotion that loads.
### What should a brand agree before the promotion runs?
The window including the post period, the base method, the comparison group and the threshold that would count as failure. Agreeing all four in advance is the difference between a measurement and a justification.
Write them down and circulate them. An agreement that exists only in a conversation will be remembered differently by the people whose budget it affects. Industry context on promotional practice is collected by FMI.
Does halo apply across brands in the same portfolio?
It can, where the items sit in the same shelf set and the shopper treats them as related. It is a harder measurement because the comparison group has to hold both brands' activity constant, and it is worth attempting only where the portfolio is genuinely adjacent on shelf.
Working with retail data? See NRS Insights data solutions or contact NRS Insights.