The research industry will in future be dominated by a very few platforms able to leverage scale in the same way that Amazon and Uber do in their industries, according to Stephen Phillips of ZappiStore.
The future of market research and insight lies in a limited number of platforms that can successfully take advantage of the network effect, Phillips told the IIeX conference in Amsterdam this week, the network effect being what enables the quasi-monopolistic status enjoyed by the leading digital players in many categories.
“The value [of a platform] to an individual user increases dramatically with the number of users of that platform,” he explained. It’s not a new concept - years ago it was as applicable to phone networks as it is today with the explosion of digital platforms like Facebook. And just as retailers gravitate to Amazon as it has more consumers than other platforms, taxi drivers lean towards Uber as it has more potential fares.
He outlined the three key elements of platforms have, starting with tech infrastructure standardisation, which then allows applications to be built on top, and finally the creation of a marketplace of buyers and sellers; are made up of the combination of components, applications and the buyers and sellers within those applications. “The network effect pushes more commerce in a particular area onto one or two dominant platforms.”
And while that isn’t currently the case in market research, Phillips thinks it will be within the next five years - because not only can the above elements of a platform be built, “on top of these we believe there’s the natural power of automation in data science to make things faster, cheaper and better”.
So, for example, a survey asking consumers about how much they liked two different brands would be stored as “Q1, Brand One and a series of data points; Q1, Brand Two and a series of data points”. But, said Philips, such an approach “is completely useless if you want a platform where you understand and can compare the liking of data and the brand itself - you need that semantic understanding”.
But if you have that, then it becomes possible to build algorithms and data science, he stated. “We believe you can add human expertise built into a scaled model - that gives you the infrastructure to create a winning platform.” Then you can start building products and applications such as pack testing and brand tracking. “That allows you to potentially create a market research platform.”
The emphasis indicates that this progression is not inevitable. Phillips charted a route for how the industry can develop. A standard project involving an ad test can establish how good or bad that one ad is and maybe how it can be improved. But if you’ve done lots of ad testing and can compare this ad to others, “then you’re beginning to build up a landscape of advertising and your understanding of your brand’s advertising.” And you can ask more interesting questions, he said, like ‘what makes a good ad?’.
That changes the way insight interacts with marketing, he suggested, as it puts insights at the start of the process in a way that can help define the creative brief.
And if you add other elements, like a concept test or a brand tracker, you add more data and further enhance the landscape. Bring in social media data, sales data, media data and you can “start to draw inferences across all of those things and build much more interesting questions that you can start answering” - such as which ad should you place with which audience, or if you want to increase brand share or brand equity, what campaign or price positioning should be used, or even what do millennials want.
“You can start asking about a new competitor you’ve noticed on social media; you can look at your brand tracker to see if it’s come up at all, what image attributes does it have, does it have a particular proposition that’s of interest to certain people.”
The more data and projects that are put into a platform - and if the data can talk to each other - “you get an exponential increase in the power of your data”, Phillips argued. And not only can you digitise the landscape of one brand, you can build other brand maps on top and then use machine learning to identify patterns in multiple brand landscapes and to predict the future.
So, for example, if a brand noticed its awareness declining and that some image attributes were weak, even while sales remained OK, it could interrogate a platform with enough data to see where this had happened before with other brands and where they’d ended up three or four months later. “Then you can start to predict what is going to happen to your brand .. and you can start gamifying your marketing function: if that, then this.
“You get another level of understanding and this is why I think we’ll end up with just a few platforms,” said Phillips, “because the power of that data will mean any individual client will always want to go to the platform with the most data because that’s what gives the power of forecasting, the power of prediction.”
And ZappiStore is trying to create that platform, he added, so insight can lead the marketing function by “being able to tell the organisation what to do, when to do it and how to do it to maximise the effectiveness and profitability of that organisation.”