4 ways to test if your CPG sales baselines are accurate

Without an accurate baseline, your trade promotion optimization analysis will be inaccurate, and more importantly many of the decisions made from the analysis will be wrong.

There are many ways to calculate a baseline, and some are better than others. Here are four ways you can measure whether your baselines are accurate.

1. No correlation with promotional activity

A valid baseline model should eliminate any correlation between the presence of promotional activity and the baseline estimate. The term for this condition is “phantom spike,” and its existence is intuitively and theoretically impossible. The theory of “steady-state equilibrium” means that there should be no structural difference in baseline between promoted and non-promoted weeks. You can see from the example below that sometimes this bias in industry models is huge. TABS AccuBase® eliminates these phantom spikes.

2. Minimal week-to-week volatility

Again due to steady-state equilibrium, expect the baseline of any given week to be close to the baseline estimates in the weeks immediately before and after. If there is a change or drift in the baseline it should be a gradual drift or a sudden, short-term shift due to some outside factor, such as an increase in distribution.

3. No structural bias in the deviations

This means that baselines for non-promoted weeks should not consistently over- or under-estimate baseline sales over continuous periods. Misses should be more-or-less random.

4. Minimal error in non-promo weeks

Comparing to weeks in which promotions did not happen, baseline estimates should be close to the actual sales, and again, per Point #3, any misses should be randomly distributed between over and under.

Each one of these tests has a mathematical formula that can be applied to measure concepts such as “No Correlation,” “Minimal Volatility,” and “No Structural Bias.”

Once the measures are gathered and transformed to a common scale, weigh each of the measures based on importance to create a weighted average score. More exotic models can accentuate the penalty for deviations from the perfect score, like the development of a Quadratic Loss Function.

Analysis disproves the notion that it is impossible to evaluate the “best” model for baseline sales. A best baseline model can be measured, and companies that do the due diligence required to make sure they get the best model available will benefit from better trade promotion optimization.