The Future of Media Research: big data, measuring social and dumbing down
Yesterday MediaTel hosted the hugely popular Future of Media Research event in central London. Chaired by BBC News media correspondent Torin Douglas and run in partnership with Nielsen, themes this year included big data, NRS PADD and measuring social media. Find out what our expert panellists had to say.
"It's as if the words 'Big' and 'Data' have never existed side-by-side before," says Andrew Bradford, vice president of client consulting at Nielsen.
"People are going around talking about it as if it's just materialised out of thin air; but it's really because there are now lots of companies producing it - and that is the first problem."
The first of many problems, it seems. Speaking at MediaTel's Future of Media Research event earlier this month in central London, Bradford began a lively conversation about big data - those two short words that seem to be on everyone's lips these days. But then you will face problems when you're dealing with data sets so large and complex that it becomes difficult to process by any traditional means.
Each of the five expert panellists at the event gave at least some cautionary criticism to what many see as an industry-wide panacea and a route to holding a competitive edge over rivals.
Yet, as Torin Douglas, the BBC News media correspondent who chaired the event, probed for explanations as to why it's so problematic, Bradford told him, simply, that no one seems to care how big data is put together.
"We need discipline and transparency, and I'm sceptical we'll get it," he said, knocking the wind out of a subject that everyone seems to be talking about at the moment.
"No one is questioning where the data comes from either," added Lynne Robinson, Research Director at the IPA. "The algorithms are what I can only describe as very 'glossy' and that is masking their quality."
The discussion on big data generated the biggest response from the 80 strong audience, and it seemed unlikely that anyone was going to stick up for it, but Richard Marks, Director of Research the Media, did see the "huge potential" it offered, but said that its name is a poor description for what it really is.
"We say it's big, but really it is deep," he said, and it's this depth that he finds some merit in. But the key to this potential is really about ensuring that the global research community improves its relationship with the likes of Google, Facebook and Apple and not to risk one research company getting into bed with any one source of big data. If that happens, argues Marks, then it would be bad for everyone.
"Big data also doesn't tell you who's doing it or what they're really saying."
It's this that feels like the key point; simply because much smaller, yet expertly controlled samples, can provide much more useful detail.
So who's driving this big data gold rush? For the panel, it appears to largely be CEOs.
Looking for a competitive edge for their businesses, it's where many of them believe it will come from and they certainly like what it tells them - quite often because they tend to own it. But, Marks says, where is the fight back from the research community? Does anyone have the guts to tell them what the faults are?
Big data is certainly vast, and people are clearly impressed by that scale - but, essentially, it ignores what consumers are actually thinking and feeling.
Marks says that too many people have the attitude that "just because it's huge it must be reliable - but that is certainly not true."
We must also remember, came one comment from the floor, that clients see value in what someone actually says, not rows and rows of data tables.
"So maybe we just need to make mistakes," offered Bradford. "Treat this as a test and learn environment."
When asked about social media - can it be measured, and if so, what can be done with it? - Richard Jacobs, Head of Commercial Strategy at Real and Smooth Radio, argued that yes, it can to some extent; largely helping programme makers to tweak shows live and responding to audience views as they happen. This, he says, is hugely useful for Radio - allowing listeners to open a two-way channel with broadcasters.
However, he did urge some caution and argued that we're not, on the whole, great at measuring social media.
Andrew Bradford agrees. "There is a very clear, linear correlation between Twitter buzz and TV ratings," he added. "But working out what people are thinking is harder, and we're not there yet."
Justin Sampson, chief executive, BARB, told the audience that peak-time TV was driving around 40% of tweets - but if you drill down to the numbers of the full TV audience the numbers are tiny: for instance, only 0.7% of the entire X-Factor audience was using Twitter whilst watching the programme - so it's certainly got some way to go.
More importantly, Sampson added, "can it detect sarcasm?"
In short, no it can't.
"It's about what these comments mean - and that is very, very hard to measure."
Death to cluster groups
Changing the subject to look at 'genuine research' versus 'PR-led research' - designed solely to grab headlines - there was some honesty from Real and Smooth's Jacobs.
"Hands up," he said with a laugh, "We're a business and we use research for headlines."
But, asked Torin Douglas, is there a danger of it dumbing research down?
"Yes. Absolutely." Which is why it must be done properly, Jacobs says.
Bradford agrees and says this type of research can be safe and valuable if done right. "It is fine to make research useful for editorial and marketing purposes; and research should always have intrinsic, internal value," he says.
But Marks offered more caution. "If someone says they want research to show them something - will they pay the bill if doesn't tell them what they want it to?"
Someone then suggested that all cluster groups should then be banned and the entire room shuffled and laughed in agreement.
Lynne Robinson could only offer further caution, conceding that it was a commercial world and that people should check who is supplying data. "If you, as an advertiser or a brand, are given research that says only nice things, then you should be very weary of it," she said.
Jacobs ended the conversation with an anecdote to neatly sum this up: when a colleague once presented some research to an automotive client, the client was thrilled with the results...but a terrible mistake had been made and the data was actually for the pharmaceutical industry.
Realising the error, the researcher called to apologise and tell the client that data was worthless to them. But the automotive client stuck with it...and only because they liked what it said.
Ending the session on a much more positive note, the panel looked at NRS PADD - the fusion of NRS print data and UKOM (Nielsen) digital data - that gives a much more sophisticated and broader picture of newsbrand readership.
"It's been interesting to see what you can and can't do to fuse different data sets," said Robinson, "and I think it's been very well received. There is good agency feedback and although publishers are split, they are getting used to it."
It is good news though - NRS has been around for 55 years, whereas NRS PADD has been around for less than 6 months. If the initial reception from the research community is this strong, then with future iterations the data should have a long, useful and healthy life merging the print and digital worlds.
To find out about future MediaTel events, please visit our website.
To see photos from this event, click here.