Promotion Baseline: How to Choose a Period for a Trade Review

Choose a baseline that differs from the promotion only by the promotion. Covers matching factors, a suitability checklist, a prior-week comparison example, and observed change versus causal uplift.

Direct answer: choose a promotion baseline period that differs from the promotion period only by the promotion itself. Match both windows on seasonality, holidays, availability, distribution, price and concurrent activity, then report the result as an observed change rather than proven uplift.

Promotion baseline diagram comparing a promotion period with a matched comparison window
Match the baseline window to the promotion window on the factors that move sales.

A brand ran a two-week price promotion on a single SKU. Sales in the promotion weeks came in 18% above the week before, and the result was circulated as evidence that the mechanic worked. The comparison was weak: three new stores had opened, the SKU had been out of stock in nine stores during the baseline week, and the retailer ran its own anniversary leaflet during the promotion. The 18% was arithmetic, not causation.

Match the comparison period on the factors that move sales

A baseline is the reference period, or reference group of stores, that represents what would plausibly have happened without the promotion. Before accepting any comparison window, check six factors:

  • Seasonality. Payday cycles, school terms, festive peaks, weather and harvest patterns shift demand independently of any offer. Two adjacent weeks can sit on opposite sides of a payday.
  • Holidays and trading calendar. Public holidays, store closure days and religious calendars change the number of shopping days and the size of the basket.
  • Availability. If the promoted SKU or its close substitutes were out of stock during the comparison weeks, the baseline is depressed and the result looks stronger than it is. Check stock before blaming demand.
  • Distribution changes. New or closed stores, a new distributor, or a changed must-stock list alter the store base. A larger store base produces more volume with no change in execution quality.
  • Price and pack changes. A regular price move, pack resize or changed trade terms in either window breaks comparability, even if the mechanics look identical.
  • Concurrent activity. Overlapping offers for the same store and SKU, retailer campaigns, competitor price cuts or a launch nearby all contaminate the comparison.

Where possible, agree the baseline rule before the promotion goes live. A rule fixed in advance is harder to select after the results are known, which is the most common way a baseline quietly becomes a justification rather than a test.

Comparison-period suitability checklist

Use this table to screen a candidate baseline before you present any number. A period that fails two or more checks is usually not usable without adjustment or a written caveat.

Screening a candidate baseline period for a trade review
Check What to confirm If it fails
Store base Same target stores, eligible stores and reporting period in both windows Recalculate on a like-for-like store list, or exclude openings and closures explicitly
Availability Promoted SKU and key substitutes in stock throughout both windows Exclude stockout days, shorten the window, or choose a different period
Seasonality Same position in payday, festive and weather cycles Use several matched weeks or the equivalent week last year instead of one adjacent week
Price and pack Same regular price, pack size and trade terms Record the change and adjust or discard the period
Concurrent activity No other offer, leaflet or competitor event affecting the same SKU and stores Choose another period, or state the overlap in the review
Data integrity Sales and stock figures reconcile with store audits or distributor records Fix the data before comparing; name the source used
Length and granularity Enough weeks to reflect normal variation at store level Extend the window; avoid single-week comparisons for low-volume SKUs

Hypothetical example: why a prior-week comparison misleads

The scenario below is illustrative, not a measured result. A distributor compares promotion week 12 with week 11 for one promoted SKU.

Illustrative comparison of a promotion week against the immediately preceding week
Factor Week 11 (baseline candidate) Week 12 (promotion) Effect on the comparison
Store base 42 stores 45 stores, three newly opened Part of the volume difference comes from stores absent in the baseline
Availability SKU unavailable Monday to Wednesday in nine stores Full availability Baseline is depressed, flattering the promotion
Calendar Post-payday lull Payday week Seasonal swing unrelated to the mechanic
Price Unchanged Unchanged regular price plus promotion No distortion from regular price
Concurrent activity None Retailer anniversary leaflet Boosted by activity outside the plan

The arithmetic change was 18%, but the review cannot attribute that figure to the promotion. A defensible next step is to rebuild the comparison on a like-for-like store base, exclude stockout days, and select non-promotion weeks with a similar calendar position, or compare the promoted stores with matched control stores that ran no promotion. Whatever remains is an estimate under stated assumptions and should be labelled that way.

Keep observed sales change separate from causal uplift

These are two different numbers and should never share a label:

  • Observed sales change is arithmetic: promotion-period sales minus baseline sales. It is reproducible and requires no assumption about causation.
  • Causal uplift is the portion attributable to the promotion. It depends on the credibility of the counterfactual, the treatment of the factors above, and the method used to isolate them.

Completed execution evidence, such as a submitted task, a store audit photo or a confirmed visit, shows that activity was carried out. It does not by itself establish sales uplift. Record the baseline definition, exclusions, data source and reviewer alongside the result so a later reader can reproduce the comparison. Textbooks on promotion evaluation emphasise the same discipline: an agreed baseline, an agreed period, a defined cost scope and explicit treatment of other influences.

Airpalm’s public product description covers mobile visits, route planning, store audits, sales orders, inventory and promotion management (airpalm.com, checked 2026-10-09). Those capabilities help teams collect store-level availability and execution evidence as it happens, which makes a better baseline possible. They do not decide the counterfactual; that remains an analytical and editorial choice.

Limitations to state in the review

Non-experimental comparisons cannot remove every alternative explanation. Say so plainly:

  • Unobserved factors, including competitor moves, local events and assortment changes, remain in the difference.
  • A chosen baseline can be cherry-picked after the fact, which is why pre-agreed rules matter.
  • Promotions can pull forward future sales, cannibalise other SKUs, or cause trade partners to build stock, so the measured window may overstate or understate the net effect.
  • Small store samples and low-volume SKUs produce noisy comparisons that a single percentage conceals.

If the comparison fails the checklist, report the observed change with the caveats attached rather than converting it into an uplift claim. For a closer look at where evaluation language tends to overreach, see How Not to Overstate a Trade Promotion Evaluation, and for the data you need before the comparison begins, see the pilot data checklist for trade promotion management.

Frequently asked questions about promotion baselines

Should the baseline be the same week last year?

Only if distribution, price, pack, availability and concurrent activity also match. Same-week-last-year captures annual seasonality well but can still fail on store base and assortment, so screen it against the checklist before using it.

How many weeks should a baseline cover?

There is no universal number. Use enough weeks to reflect normal variation for that SKU and store set, exclude distorted periods openly, and record the rule you applied before the promotion runs. A single week is rarely sufficient for low-volume products.

Does a control group replace the need for a matched period?

Not entirely. Control stores address shared market shocks, but the comparison period still needs to be free of stockouts, price changes and overlapping activity for the control group to be meaningful. Use both where the data allows.

Does completed execution prove uplift?

No. A submitted task, store audit photo or confirmed visit shows the activity happened, not that sales rose because of it. Report the observed sales change and the execution evidence separately from any causal claim.

Put this guide into practice

This general-purpose worksheet is a starting point. Adapt its fields, responsibilities and rules to your business process.