Beyond Meat

The brand

Beyond Meat, founded in 2009, is a leading plant-based meat company with products in more than 133,000 retail and food service outlets across 65 countries. It went public in 2019, the first plant-based meat alternative company listed on a US stock exchange.

Challenge

Like most CPG brands with brick-and-mortar distribution, Beyond Meat wanted a more sophisticated attribution and marketing accounting framework to understand how media investment was actually driving sales. The mix was broad, spanning online and offline channels alongside traditional price promotions.

Retail Media Networks were the sharper question. Investment there had been climbing across the industry, but it wasn't clear how much RMNs were contributing. There was also a prior MMM from another provider whose outputs the team wanted verified. The underlying question, as always: what is actually working?

Approach

The datasets were large and varied. Media investment and delivery data covered TV, social, audio, display, video, and RMNs. Retail sales data spanned big box (Costco), traditional grocery (Walmart), and specialty natural and health food retailers. All of it needed processing and transformation before an analytics plan could be built.

• Data harmonization. Two years of history on sales, media, and events, reviewed and processed for modeling.
• Preliminary analysis. Trend analysis, correlations, and a basic MMM to understand variable fit against sales, then a review of retail store categories to form a hypothesis on model count.
• Marketing mix modeling. Thousands of iterations and hundreds of tranches across retailer groupings, landing on best-fit models for four primary retail sales channels. Three were standalone major chains, the fourth an aggregate of specialty retailers.
• Triangulation. Using the MMM decompositions to understand media's contribution to retail sales.

The Marketing Accounting Framework was oriented around incremental retail sales dollars driven by media, distilled into incremental ROAS by channel and platform. Four models meant four P&Ls. Grouping the smaller specialty retailers made the models feasible, while the three major chains were distinct enough to each warrant their own.

Results

What it meant

Two opportunities showed up. The first was inside the existing budget, shifting investment toward tactics with stronger incremental sales contribution. The RMN spread alone (3.9 down to 0.3) made that case without much argument. The second was a genuine case for more spend, with the forecast pointing to 26% growth in media-driven retail sales if the budget roughly doubled.

Media also behaved differently depending on the sales channel and retail network, which is the kind of finding that a single blended model would have hidden entirely.

The outputs produced recommendations and a set of hypotheses that needed testing. Those tests validate and refine the models, and update iROAS figures so confidence is high before scaling any large investment shift. Data-driven decision making of this kind never becomes set-and-forget. Media, consumers, and business dynamics keep moving, so advanced attribution has to be treated as evolutionary rather than static.