What it means
A retailer sells 15,000 units during a promotion, and without a credible baseline it cannot tell whether the campaign added demand or whether 13,000 units would have sold anyway. Baseline sales provide the counterfactual estimate.
Bedrock Analytics explains using non-promoted velocity, seasonality and distribution adjustments to construct a promotional baseline, and Daasity documents base sales metrics, but both are vendor sources, so their particular methods and performance claims are not universal standards. Set the question first: are you measuring a price discount, advertising campaign, new shelf placement or a new store?
The baseline must represent the absence of that intervention, and the unit must be chosen, since sales can be measured in units, revenue or gross profit and a discount may lift units while reducing margin. Match the time, comparing the promotional week with an expected figure for that same week, not a random quiet period, and use relevant history, because recent non-promoted sales can provide a starting point while very old data may miss new competitors or changed distribution.
Adjust for seasonality, as a holiday week can have normal demand above a typical month and failing to account for it overstates promotion lift. Check distribution, since if the product gained twice as many stores, higher sales may reflect wider availability rather than the campaign alone, and watch stockouts, because low historical sales caused by empty shelves should not define normal demand.
Consider price trend as well, as a baseline from a period with a different everyday price may not be comparable, and separate one-off events such as weather, local events or competitor outages by recording material factors without inventing a perfect model. Use a control group when possible, since comparable stores without the campaign can help estimate what would have happened, provided they are actually comparable.
Avoid contaminated weeks, because a supposedly non-promoted period with a coupon or display is not clean history, and document the lookback by stating the weeks used, exclusions, model method and data source, as different methods can produce different answers. Calculate uplift clearly: promoted sales minus baseline gives estimated incremental units, and dividing by baseline gives an illustrative percentage.
Check pull-forward, since some customers buy during a discount instead of next week, so review post-promotion demand before calling all the uplift new demand. Look across channels too, because an online promotion can shift purchases away from stores so total brand sales may move less than the promoted channel, and a customer who would have bought in-store may shift online during a digital campaign, making a channel-level uplift larger than the business-level gain.
Check the margin as well, because incremental units can be unprofitable after discount, advertising and trade spend, and baseline sales alone do not answer return on investment. Use uncertainty ranges, since if demand is volatile a single precise baseline may overstate confidence, and review product groups, because a campaign may increase one item while cannibalising another.
Beware self-fulfilling models, as a system that uses promoted weeks as normal history lets future baselines drift upward, so mark events in the data, keep period definitions stable, and never compare a per-week baseline with a two-week sales total without scaling. Compare forecast baseline, observed promotion and later sales to refine the model, and do not claim causation automatically, because a sales gap can coincide with a promotion without being caused solely by it; for owners, baseline sales are the quiet line under the event, useful only when its assumptions are visible and credible.
In practice
Real-world examples.
Example
Three comparable non-promotion weeks average 10,000 units. The analyst uses them as the starting baseline and notes that none contained a coupon or display. The figure is then compared with the promotional week.
Example
A holiday adjustment raises expected sales before calculating lift. The prior year shows that the holiday week ran above a typical week even without a promotion. Without the adjustment, the campaign would appear more successful than it was.
Example
Post-promotion sales fall, suggesting some customers bought early. The two weeks after the event sell below baseline, so the analyst subtracts the dip from the uplift. The campaign's net effect is smaller than the headline.
Formula
Calculation
Illustrative baseline = average comparable non-promoted sales. For weeks of 9,800, 10,200 and 10,000 units, baseline = (9,800 + 10,200 + 10,000) / 3 = 30,000 / 3 = 10,000 units per week before needed adjustments.
Uplift = promoted sales - baseline, and uplift percentage = uplift / baseline. If the promotional week sells 15,000 units, the uplift is 15,000 - 10,000 = 5,000 units, or 5,000 / 10,000 = 50%.
Adjustments change the answer. If the product gained 20% more stores, the adjusted baseline is 10,000 x 1.20 = 12,000, so uplift falls to 15,000 - 12,000 = 3,000 units, or 25%. If sales then run 1,000 units below baseline in each of the next two weeks because customers bought early, net incremental units are 3,000 - 2,000 = 1,000.Case study
Seen in the real world.
This entirely fictional example follows Cedar Snacks. A campaign week sold 15,000 units against an initial 10,000-unit baseline, but more stores had gained distribution. The team adjusted the comparison and checked later weeks before judging the campaign. The example does not claim that all excess sales were caused by promotion.
Using the adjustments in the worked example, Cedar's analyst concludes that the campaign added about 3,000 units in the week and that two soft weeks afterwards took back about 2,000. The marketing director had expected a 50% uplift and sees about 8% on a net basis. The team decides to test a smaller discount in a control group of stores before repeating the full campaign.
Watch out
Common mistakes.
- Using a holiday sales spike as a normal-week baseline.
- Ignoring distribution or stockouts when comparing periods.
- Treating every unit above baseline as proven incremental profit.
Questions
People also ask.
What are baseline sales?
An estimate of sales expected without the tested intervention.
Why do they matter?
They give a comparison for estimating incremental effects.
How are they set?
Use comparable non-promoted evidence and adjust for known demand changes.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%