Ten big challenges for market research in cross-media measurement
Emily Barley
Warc
"The whole future of the media industry depends on one question: what children do now – are they doing that because they're children and are they going to have 'normal' media habits when they grow up?" Richard Marks, owner of Research The Media, told the audience of an ESOMAR multi-platform media seminar held in Amsterdam in June 2015.
"Clearly they are obsessed with YouTubers and new media", said Marks, but "the trouble is, the hardest thing in market research is to predict what people are going to do".
"We've been through that in the [2015] UK election" – where asking people what they were going to do meant pollsters got it massively wrong. In the end "the only accurate poll was asking people what they did".
Advertisers want to understand how their campaigns work across platforms, and are fed up with media 'silos'. "They want to be demonstrating that the internet has turned them from caterpillars into butterflies – their cross-platform reach has grown", said Marks.
This media landscape, and advances in data collection and analysis, throws up new challenges – and opportunities – for researchers. Marks shared the ten biggest ones he has identified.
1. Big data is old data
In the quest for certainty, advertisers need to remember one key thing: "media planning datasets can be points in time", which do not hold true for all time. There is, of course, huge benefit in understanding past behaviour, and some projections can be made about future behaviour, but advertisers need to avoid the mistake of assuming what has passed is what will be.
This is "the challenge of planning against trading", said Marks.
2. Measuring content versus measuring advertising
It used to be that measuring the content meant measuring the advertising, but "in a digital world advertising is increasingly decoupling from the content", argued Marks. Agencies and researchers need to find a way to track all of the content – including advertising, but no longer limited to it – that feeds the brand image.
3. Fighting for demographics
"A lot of people are arguing that demographics aren't needed any more", said Marks, "[but they] are largely the people who don't have access to them in their systems". As big data is able to report more of what people do, researchers and marketers need to recognise that data is often detached from understanding about who the consumer actually is. "A lot of stuff can be automated, but demographics are often the biggest challenge and the biggest contribution that research can make," said Marks.
4. The way data works may have to change
"We may have to be pragmatic to get cross-media research", argued Marks. Researchers should "move away from respondent level databases". Instead, a probabilistic approach may be the best way to analyse data.
5. Defining media metrics
"You could argue that each media has a definition that favours it – that shows it in its best light", said Marks. And whilst owners of each type of media will naturally favour the measurement it looks best in, advertisers really need something that works across every platform. Marks proposed 'time dwelt' or 'attention' as possible solutions, but admitted it's an uphill battle as some media suffer under any cross-media system.
6. Multitasking and divided attention
"There is increasing awareness that as more media is added in, the media day remains 24 hours" Marks observed. "Large amounts of people are multitasking – and if they're sharing their time with other media, should that be borne in mind when measuring that media?".
It might be that cross-platform measurement means "spreading that exposure … [moving] into the area of engagement" said Marks. A minute spent watching TV and using a smartphone might mean measuring half a minute for each.
7. Measurement of tablets and mobile
"These are a major challenge – far more than the internet and computers ever were," said Marks. This is partly because of the multiple operating systems being used at any single time, and partly because of hardware developers' own agendas. "Apple doesn't like advertisers", explained Marks. "Devices are almost designed to thwart advertiser measurement".
8. The big data challenge
"I'm a big fan of big data, but there are limitations", said Marks. "There is a real danger that those silos that the advertisers don't like just get replaced by another kind of silo which are source-centric".
As advertisers rejoice at the advent of cross-platform measurement, they will also begin to see multiple problems with the new types of data they are collecting. Big data usually comes with no demographics, with errors that can be as bad as survey data, and with data about devices rather than people. This is before considering issues around access, privacy and "whose data is it anyway?", Marks added.
To get a full understanding of what's going on, advertisers are going to need to work with research companies. And in turn, research companies are going to need to work with data owners. And "that introduces an additional layer of complexity", said Marks.
9. The weaponisation of media data
"We may be moving from an era of shared industry data to one in which it's used as a weapon," Marks suggested. This is because "media owners may be seeing their data as something that gives them a competitive edge".
Researchers want a consumer-centric approach – "where we're mapping data on to people", rather than the other way around – but media owners' primary concern is to make their inventory look more attractive.
10. Research may be becoming less representative
It is becoming increasingly difficult to persuade people to take part in research, and particularly panel studies. This is especially the case in more complex and demanding projects, and that's not just a problem for researchers in terms of recruitment. Quite simply: "The more you ask people to do … the less likely they're representative," Marks asserted.
To collect the amount of data they need, researchers are looking towards single source and data fusion methods. But neither of these are ideal, said Marks: "can anyone afford single source measurement? But equally, can anyone understand fusion methods?"
Market research needs new skills
The challenges and opportunities presented by big data means that "market research is expanding into three potentially distinct skillsets", Marks suggested. These are:
- Market research: the survey design, focus groups, etc, of traditional market research;
- Statistical modelling: here the industry faces a huge challenge and needs to attract "best and brightest" who can do advanced analysis;
- Big data methods: these require a more mathematical brain than market researchers – who are usually social sciences graduates – tend to have.
Emily Barley is Editorial Assistant at Warc.
