Forget cookies, location data is secret ingredient in KFC’s media strategy

According to execs at the UK media house, an approach combining footfall, location-specific sales and census data with a machine-learning model helped KFC roll out its delivery service in recent months, delivering 39% more responses than the next best campaign for the same cost.

Using a dashboard that resembles a “glorified Google Maps,“ the agency was able to use media investment data to show which KFC outlets were underperforming – before KFC’s own team even knew about it. Now, the agency is beginning to use the same strategy on Pizza Hut.

The fried chicken chain wanted to follow in the footsteps of rivals such as McDonald’s, which established a profitable delivery service in recent years, and absorb some of the value flowing into the coffers of delivery middlemen UberEats and Deliveroo.

According to Morys Ireland, Mindshare’s head of data and technology services in the UK: “The whole point was to dis-intermediate Deliveroo and UberEats. The campaign’s initial message was that if you got KFC’s app, you’d get your first bucket for free. That was quite a decent and motivating message, so the first piece was all around that.“

But the fast food joint lacked rich data on its customers’ habits. “It didn’t know much about its customers other than that they like chicken and how often they want to buy it,” says Ireland. Geographic first-party data – census data for demographic makeup, footfall and sales from specific outlets – helped bridge that gap.

Mindshare limited KFC’s advertising activity to within 3km of every store that was offering delivery in the UK. Then, combining geographic and demographic data with sales figures and Mindshare’s own audience insights, it altered its media buying mid-campaign in order to make the most of KFC’s budget.

Being able to test and learn new approaches midway through a major push meant that KFC’s delivery awareness campaign performed 39% better than the next best campaign, according to Mindshare.

“We targeted audiences in places where that deal could actually be fulfilled and started off with an omnichannel piece including out-of-home, digital audio, online and offline channels.”

Location, location, location

After the initial burst of the campaign, Ireland says Mindshare used a strategy developed by GroupM Nexus called ‘Unmissable’ to combine different datasets – none of them including personal data – and understand the impact of KFC’s marketing in real-time.

“We could start to see patterns emerging. One of those was that suburbs performed better than city center locations. It makes sense – people live in the suburbs, that’s where they’re going to order KFC.”

As such, the team altered the campaign’s media mix to appeal more to customers living away from big cities while using out-of-home and large mass-media channels in urban centers. “In city centers, people go in and eat or they take away. In the suburbs, people get it delivered.”

What began as a “small scale test” as KFC rolled out its delivery program has become a nationwide scheme operated by Mindshare. The media agency now provides KFC with a dashboard, overlaying media investments and sales figures with a map of the chicken provider’s UK outlets.

“It’s basically a glorified Google Maps and you can overlay loads of different datasets on there – it could be panel datasets, the client’s own data, third-party data, census data, anything.”

Media data helping operational decisions

And the insights generated by the system haven’t just been useful for tweaking KFC’s media mix. Ireland says it was able to feed back data to KFC’s operations team and help them identify – and begin to fix – outlets that had poor sales. “If the restaurant in, say, Watford had no delivery data, we could shut down media there because they’d obviously had a problem fulfilling deliveries. We could find that out for KFC before the franchisees came back to them to discuss problems.”

That function also helped Mindshare cut back media spending for outlets that were overperforming, reducing waste. “It transpired that some of the top-performing locations didn’t quite need that support, so we took some of that budget and pushed it towards the ones that were underperforming.”

According to Suzanne Perry, head of media at KFC UK&I, the approach has helped strengthen the parent business’ relationship with franchisees across the country.

“We don’t want to put too much pressure on our restaurant staff. Having the ability to turn media on and off around an individual store – if they’re struggling a bit or they don’t need the extra push because deliveries are already flying out of the window – you can’t monetize that but it is so important. It makes our restaurant teams feel confident in a new service that we're offering. And we couldn’t get that easily with a national buy.

“It’s really difficult to talk about media to franchisees… people don’t necessarily understand what we do, and how the nuances of it can affect the business positively or negatively. But this is something that we rolled out in front of some of our most important franchisees, and they were all hugely engaged. To have their support is so important. Even our leadership team that have nothing to do with advertising and media are acknowledging that it’s making a real difference.“

The sheer volume of data generated by Unmissable soon became an issue. “With the best will in the world, we’re not really going to be able to analyze 900 stores,” says Ireland.

The team developed a machine learning model that combines geographic and demographic data to highlight problem locations within KFC’s poultry empire. “It tries to find causation between different factors that might be causing areas to be under or overperforming,” says Ireland. “It could be population density and area, or there might be a specific demographic makeup in the area which leans towards KFC – student towns versus family-oriented suburbs. That’s led us to be able to build a better model to see what’s driving performance and then run nuanced creative in some of those places.”

Key to the success of KFC’s delivery push was app downloads and sessions. Without users adopting its app directly, they’d end up going back to UberEats. Mindshare began taking into account the number of times its app was opened – and for how long – as a means of measuring media impact. “Think of it like econometrics on steroids,” says Ireland.

By monitoring app sessions rather than sales figures or ‘last click,’ Mindshare found it was able to monitor eight times as many conversions. Spotify and digital radio were particularly useful channels, says Ireland. “That enabled KFC to change its media mix. It invested in higher impact performance and digital display and got rid of the sort of more standard formats. And it shifted more budget towards digital out-of-home and digital audio.” In some regions, video-on-demand (VOD) was particularly effective with this approach, too.

“In one particular ‘burst’ we found VOD was driving five times more app sessions. So we got some specific assets developed for VOD and shifted some of the linear TV budget on to online video.” Those tweaks, he says, increased app sessions by 12% for KFC. “Those channels are becoming increasingly addressable. And if you’re able to reach people in the right place at the right time, it makes a real difference.”

Ireland says the approach has turned out to be at least as useful as strategies that relied on first-, second- or third-party personal data. “We still want to be able to address audiences in a way that is accurate enough to drive results in media, but that’s also not so invasive that it bumps up against consumer privacy concerns.“

You probably don’t get as accurate a picture of an individual as you did with a cookie, he notes, but opting out of cookies also means opting out of the onerous legal and contractual compliance maze surrounding its use. And it’s less likely to give users that ‘followed around the internet’ sensation.

For Perry, the strategy has proven more cost-effective than more traditional means. And it avoided the blowback that might have been generated, had the brand run a nationwide campaign and ended up enticing customers in areas that didn’t offer delivery (approximately 10% of KFC’s stores).

”This is a much more efficient buy for what we’re trying to do,” she says. ”It allows us to be so much more agile and flexible around each of our individual stores. And that is invaluable to us.”

Pizza party

Mindshare’s KFC experiment has been successful enough that it has begun to expand it to other accounts. Pizza Hut, also owned by Yum Brands (KFC’s parent company), was a straightforward transfer, says Ireland, because it also had a delivery business based on physical locations across the country.

For another client that Ireland declines to name, Mindshare has been incorporating meteorological data into its mapping. The company worked with another data provider to create a geographic dataset that scored every UK postcode for potential market share. “We got some data from the client on how weather impacts its sales. And then we got the actual weather forecast. And then simple things like the population per postcode.

“We now have a model running that crunches all of that and spits out an index that enables us, working with the Trade Desk, to decide which areas to operate in each day.” The campaign runs different forms of creative depending on the forecast in a given area, he says.

Not every client will find the system so useful, he adds, saying that its focus on geography means it is of immediate application to businesses with a big brick-and-mortar presence (or which rely on retail vendors) and less so for online-only brands. “Where they don’t have that physical presence, we have to be more creative.”

In some ways, this strategy is one that’s been used by advertisers working with local media or local OOH inventory for decades. But heavier computing firepower and machine learning have enabled Mindshare to model and make decisions far faster than previously possible.

“What we’re saying to clients is that this is pretty much what we’ve always done. But hopefully, it’s a more future-facing way of doing it.”