Elasticity is a key concept when determining how much a change in a business driver results in a change in volume. For simplicity, you multiply the % change in the driver times the elasticity to get the % change in volume. (The formula for applying elasticity can be more complex, especially for pricing, but this simple multiplication will get a you a good answer if the change in the driver is within +/- 20%. Some examples: If the elasticity is 1.0, then a 5% increase in the driver results in a 5% increase in volume. If the elasticity is 0.8, then a 5% increase in the driver results in a 4% increase in volume (0.8 * 5%). If the elasticity is 1.2, then a 5% increase in the driver results in a 6% increase in volume (1.2 * 5%). The sign of the elasticity should make sense in real life – positive if the driver and volume move together or negative if they move in opposite directions. For example, if distribution goes up, volume also goes up so the distribution elasticity will be positive. Price elasticity, on the other hand, is negative because if price goes up we expect volume to go down. Elasticities can be determined in different ways, some simple calculations using data from your Nielsen/IRI database and some more complex from things like marketing mix or price-promotion studies.
I would love to know or see an example of how to effectively calculate price elasticity using Nielsen/IRI data. Thanks in advance!
Generally speaking, price elasticity cannot be accurately estimated from the market level, aggregate data available in typical Nielsen/IRI database. There is usually a lot more than price that is changing at the same time plus the prices you see are averages across a bunch of stores, not necessarily the prices that shoppers are actually responding to. Price elasticity work is typically done thru Nielsen or IRI’s advanced analytics custom projects group (or another consultant with data access and modelling expertise).
I second what Sally said! To get a true price elasticity it needs to be modeled using weekly, store-level data which is not typically available in regular client databases from Nielsen/IRI.
Although there are some “quick and dirty” ways to estimate price elasticity from market-level data, those would not work using data from the last year or so. That’s because during 2020, retail prices on almost all items increased but so did volume! That goes against the intuitive phenomenon that volume decreases when price increases. I’m sure Nielsen, IRI and the larger CPG manufacturers are doing extensive work around this – not an easy analysis.
Hi, I am a Data Scientist/Software Engineer, and enjoy building Machine Learning (ML) models to extract “actionable insights” from data. Your posts have been *extremely helpful* for me to understand the terminology used in the CPG universe (which sometimes seems opaque to me), and your advice on how to present results to stakeholders so they are useful and effective are *excellent*. Thank you for generously sharing your knowledge with others! Now to my question 🙂 Do you have any example that shows how to use elasticity values from models built using weekly data in order to build the “due-to” analyses you discuss in some of your posts? If you do not have examples, do you know of places that might have such examples? My goal is to combine my ML knowledge with the recommendations you have shared to present the “insights” to stakeholders. Thank you for considering my question!
Thanks for the kind words! We’re glad that you are finding the blog so helpful. One of the biggest opportunities for data scientists to make an impact is to identify insights and present them in a way that the business stakeholders understand. Regarding your question…I am literally writing a post all about how to apply price elasticity and that will be posted by the end of the week. I’ll edit this response to include the link when the post is published.
Thank you so much for the quick reply, Robin!
I can’t wait to read it! But take your time 🙂
Thanks again!
How rare is it that total value sales decline despite a big distribution increase? And what factors might be causing that? Something I’ve seen in Nielsen report for a competitor
Thanks for such a great resource!
We’re glad you’re finding the CPG Data Tip Sheet useful! You are correct that distribution and sales usually move at least in the same direction. Remember that the other component of sales is velocity – how fast is the product moving where it is in distribution. If you are losing distribution in the slower-moving stores then the velocity will increase since the product is still in the fast-moving stores. And that increase in velocity could be enough to more than compensate for the distribution loss to result in increased value sales. It would be useful to also look at velocity for value and physical volume. An increase in value sales could be totally due to higher pricing with shoppers actually buying less product. (See this post on the 3 different measures of Sales.)
Hope that helps!