PepsiCo uses data science to decide its next crisp flavour, now it could inform its marketing

“Historically we’ve been very slow to respond to some of these, we needed to switch that around, so we’ve been approaching [trends] as tailwinds, rather than headwinds to fight against.”

A data marketing firm Black Swan has a suite of tools which use everything from weather to search, government to social data, to help brands design products and glean value from their media and retail buys.

To get to this point, PepsiCo has been using its social prediction tools Trendscope – which looks to give brands a first mover advantage by aggregating publicly available consumer conversations and insights across billions of touchpoints – with its own figures to decide which products to bring to market. Crucially, it’s an approach PepsiCo hinted it could emulate when it comes to assessing its media and marketing investments.

The product strategy has already led to the development of snacks like salted rice and pea chips or humus, garlic, basil and tomato crisps, with concept to development taking “half the time” it usually would.

These are the kind of nibbles and refreshments you’d expect to pick up in a hipster coffee shop in Shoreditch or Brooklyn, but instead, they’re products launched under the FMCG giant’s Off the Eaten Path vegetable snacks brand in Sainsbury's stores across the UK using a shrewd combination of predictive analytics and internal data.

PepsiCo’s Howarth explained that the strategy is helping the food behemoth turn data that would otherwise be a “maelstrom of chaos” into something that’s a lot more practical for the brand to act against. The taste of its next range of Sensations crisps – a brand that’s already experimenting with flavours like wasabi and ginger – will be informed by Trendscope.

“It allows us to then have a conversation with our sales and marketing teams, where we can say here are the two, three or four trends that we think have really big potential that we think should as PepsiCo act against,” he added.

Predictive marketing?

The same insights could in time be used to inform marketing placement, with Howarth explaining it can define strategy and help with wider comms planning.

The feature allows advertisers to pinpoint where conversations are happening at any given time, including what marketing trends are on the up. When it comes to sales, Howarth noted that there was also scope for Black Swan’s tech to help identify which retailers are hot right now.

Many brands have already bought into the data startup’s tools to aid media and digital teams. For Disney, Black Swan proved the correlation between weather and cinema visits, before making sure brand’s ad servers were automatically set up in a way that would drive people into theatres on cloudy or rainy days.

Pharmaceutical giant GlaxoSmithKline, meanwhile, tapped the agency to run a hyper-local UK campaign which combined sales data with external data around search behaviour, social media and weather to forecast pollen count rises or flu outbreaks.

The model was then integrated with GSK's digital media and retail partners to automatically serve drug ads online and in-store depending on where an outbreak was happening. The activation reached 9 million people, with an uplift of 200% on previous click-through-rates and a 12% rise in sales; showcasing the potential value of the forensic approach for marketers.

Driving change

Back on the product side, the Doritos and Tropicana owner is also in the early stages of adding its human trend network into the predictive mix, working with “around 300 to 400” consumers across the UK and Russia to glean their insights via an owned “always on” digital platform which has been helping PepsiCo get a better grip on where exactly the trends Black Swan is highlighting are occurring.

Next, Howarth said the brand wants to start thinking about how it can then pull in its traditional sales data, adding yet another layer of insight.

For now, PepsiCo is focused on the product development capabilities its plugged in digital forecasting tools allow for, so it can use its successes to ensure it's "less of a battle" when it comes to selling the concept to the wider business.

Since the start, Howarth's team have been on a journey to make sure the the wider group sees the material impact of the investment.

Howarth continued: "You need to make sure you get senior engagement quite early and bring people along..." he said. "So, if we know if we start doing that, we’ve not just gone off on our own as an insights team and created something, we’re already starting to answer [senior execs questions] with something that’s got a business impact.

"You have to put [something like this] on to the existing business processes, if you just create a new process in today’s world and in FMCG we’re all working harder than we have done before. It’s a challenging environment... people don’t have that much time so if you just say 'we’re going to add another capability' on they’re going to say 'oh god what do we do with that?'"