CPG Data 911: What to Do in an Incremental Volume Emergency

CPG Data 911If you’ve worked with CPG data for a while, you probably know that incremental volume estimates can fluctuate and look a little funky (especially in weeks with little or no trade promotion). I wrote a post about this phenomena a few weeks ago. As I stated then, usually these fluctuations are small relative to total volume and can be safely ignored. 

But occasionally you can have more serious data issues: either way too much incremental volume (symptom: merchandising efficiencies over 100%) or not nearly enough incremental volume (symptom: negative incremental volume even when your time period includes a lot of promotion).  If this is happening to you, here are some treatments which should restore your analysis to health:

1)    Move up a level in the product hierarchy

Are you looking at promotions for individual UPCs? That analysis level might be too low to get reasonable results. Instead, try aggregating all UPCs that are priced and promoted together and look at incremental volume at that level.  This level is often called a “promoted product group” or “PPG” and usually contains items of the same or very similar size.

If you still have data problems at the PPG level, try moving up to the brand level.  Cannibalization between promoted product groups (e.g. different sizes of the same brand) may lead to negative incremental volume at the PPG level that washes out at the brand level.

2)    Look at individual retailers rather than a larger geographic aggregate

If only a few accounts are generating the majority of your incremental volume, you may not get a useful analysis at a higher geography (e.g. total US).  You need to be down at the account or banner level to get actionable results.

3)    Look at one promotion at a time 

If you are looking at long periods of time and aggregating across multiple promotion events, focus instead on each individual promotion.  Conduct an event by event analysis, narrowing in on the promotion weeks and excluding the non-promotion weeks.  You’ll quickly see which events had strong lifts and which did not.   From there, you should be able to see which tactics worked and which factors hurt the weaker promotions.

4)    Aggregate over a longer time period

In a way, this is the opposite of the previous approach (#3). Instead of narrowing your time period, expand it. Although you might have counter-intuitive results over four or eight weeks, things could look perfectly sensible over 13 or 52 weeks. This strategy can be especially effective for products with only small amounts of promotion.

5)    Get some perspective on your overall trade promotion presence

Sometimes, when your incremental volume numbers aren’t making sense, you need to step back and consider your overall promotion presence. If your brand only sells 10% of volume on promotion or its merchandising is limited to small shelf price discounts, the Nielsen/IRI incremental volume methodology may not be sensitivity enough to provide accurate results.  In other words, if you don’t have much going on, the “noise” of other activities overwhelms the impact of your trade promotions. Before drawing this conclusion, however, be sure to try approach #3 and zoom in on individual promotions (especially if you believe you receive quality support but just have one or two deals a year).  You may be able to hone in on a few events even if you can’t look at trade promotion in aggregate over a long period of time.

Need help diagnosing problems with your trade promotion analysis? Contact us for a 30-minute free consultation.  Our consulting work and training programs help CPG companies improve trade promotion ROI.

Subscribe to CPG Data Tip Sheet to get future posts delivered to your email in-box. We publish articles twice a month. We will not share your email address with anyone.

PrintFriendly and PDF

Comments

  1. 6) If you see negative incrementality, it could be competitive activity. Do you have access to “COMP” markets in Nielsen or IRI? It very well could be happening, since the baseline adjustment has to come through the “addition” of a negative incremental.

Have a comment or question?

*