Commentary
Could Twitter Replace Nielsen?
- by Dave Morgan , Featured Contributor, June 4, 2009
Relative to the latter, my team has been spending a lot of time lately analyzing Twitter data to try to see if it could be useful in better understanding how people view television, since so many people regularly tweet about their TV viewing activities. The results have been interesting, but what was most eye-opening was the amount of interest and discussion that was generated in comments when the results were released on a popular technology finance blog, Fred Wilson's AVC (Disclosure: Fred is an investor in my company, Simulmedia).
advertisement
advertisement
What came through the strongest among the 100+ comments was whether Twitter might be able to replace Nielsen and other audience measurement services, or whether Twitter data might even be able to replace set-top-box data as a source for census-based television viewership. Could Twitter someday replace Nielsen ratings, or set-top-box data? In my view, it's too early to tell, but here are some of the advantages that it might offer:
Real-time results. What Twitter lacks in precision, it certainly makes up for in real time. Tweets tell you what people are doing in the moment. You don't have to wait hours or days or weeks for results. As more and more marketing becomes real-time, so must the tools that enable it.
Intention data. Twitter can not only tell you what people are doing, but why. Understanding intentions and motivations can take a lot of guess work out of marketing and media.
it's a focus group/survey tool on steroids. Twitter lets you watch, interact and survey lots of different types of people very quickly and very efficiently. The survey and focus group business will never be the same once companies learn how to leverage Twitter here.
Authenticity. Twitter today is as wide open and uncontrolled as TV panels are closed and controlled. Neither is ideal, but having both means that we're all more likely to find out the truth about viewers over time.
It's free. Yes. Hard to beat this one. Companies can tap into Twitter for free. I suspect that Twitter will find ways to charge for premium services and uses over time; now, however, it is free.
Is Twitter the new black when it comes to consumer marketing research? Should Nielsen and others be worried? What do you think?
Twitter may have some utility BUT it does not gauge the opinions, viewpoints, TV watching habits or buying habits of those that say nothing....countless zillions of consumers buy things - be it cereal, soda or a movie ticket - and never speak up and they too need to be counted..often times, it's the silent majority that's bigger and maybe more important than the vocal minority.
If 10% of Twitter users account for 90% of the posts, what do the other 90% have to say?
That's what smart research qualifies.
Jaan ... agreed. But the "vocal minority" of Twitter is much, much larger than the samples used for TV panels or focus groups. Today, Twitter's data is raw, but there is no reason that it couldn't be qualified and projected, just like other research tools.
Given the limited usage of Twitter by the general population (my estimates is it is about 15%) and radical skewness of the demographic associated with their users, the answer today is a resounding no.
Basic statisical principles recommend never using data that may present bias. Twitter does exactly that. When Nielsen formulates a panel for research, they do so by seating a panel that elminates potential bias and provides a sample that is reflective of the population the are seeking to forecast.
Unfortunately, I do not see Twitter being able to do this today or anytime soon.
I like the fact that we are seeing how to use social media for different purposes. Yes, it might actually provide a different perspective into why a consumer views the program and what makes them continue to view the same show. Their comments on Twitter, and other social networks will help offer these insights. Will it replace Nielsen? Maybe so, if Nielsen doesn't realize there are other resources to use to keep its customers informed of TV programming. But we are only seeing how it applies today. By next year this discussion will probably be about another new form of communication.
Tim, very good points, though I do think that the Twitter population will normalize with time. It doesn't seem that long ago that the same objections were raised about the use of the Internet for research. And, in fact, one could raise the same objection to telephone surveys since homes with telephone landlines not registered on the "do-not-call" list certainly represent a skewed consumer sample.
AN interesting premise, however I agree with Tim's point about the statistical bias involved with the sample. Further, Twitter lacks structure, and like it or not, industry standards become standards because they incorporate structure and minimize variance. Nielsen does this, its simple, probably so simple it loses some value, but a replacement needs will require universal agreement and simplicity.
It's unfortunate, but true.
The comments thus far have some disagreement as would be expected, but they all point to one interesting facet of this picture. For advertisers who are looking to reach those who are early adopters of new products and technology, the idea of tweet data on advertising could at the very least be an intriguing experiment.
That 10% of the "vocal" Twitter users that Jaan references, and the 15% of the general population that Tim references, may be precisely the demographic that some marketers are trying to measure.
I actually had a spirited debate about this via Facebook a couple of weeks ago! Aside from the sample issues that people have mentioned already, using Twitter this way makes a few big assumptions about how people actually use the service. First, that they tweet when they watch a TV show at all (which is one of the things TV ratings measure). Second, that they tweet in "real time" while or soon after they watch a show. Third, that tweeting about a show means they watched it in broadcast or with a DVR , not online, on a DVD, or on a mobile device (potentially long after broadcast or broadcast +3).
While I think Twitter might be a good gauge for figuring out how people are reacting to some things - particularly news, "event TV" like the Super Bowl, and some reality shows like American Idol - I don't think it'll replace ratings any time soon.
Maybe there will eventually be a real use for twitter and maybe some real money will be collected from/for the "service" (profits are another story). Although there are supposed to be more twits that Nielsen's pitiable samples, replacing twits for somewhat measureable boxes does not seem viable. The fickleness of the tech media world will create another source to spout shoe lace tying episodes before an effectual and profitable audience measurement system. (Note: I have purposely left out many adjectives.)
I agree with Eleanor on what it will, a gauge of reaction. They will have huge issue at Twitter ever being able to remove skewness of data. Dave's article hits on what everyone is asking, what will Twitter become. I don't foresee it being a Nielsen competitor.
I agree with all those that make the point of having representative research, not data that is skewed to vocal minorities and services like Twitter which have no provisions for demographics. It's just not practical for this kind of data to supplant services like Nielsen. We need properly selected quantitative samples for measuring viewing/usage and equally proper samples for measuring the qualitative feedback necessary to fully understand why consumers like what they see and do -- such as the emotional bonding data provided by Q Scores. Tweeting, etc. is unrepresentative and not even as credible as a few small focus groups contiaining the intended consumer target as respondents. Tweeters or whatever we call them are representative of tweeters and nothing else. I agree this will always be the case, unless significant adjustments are made in the way tweeters are measured -- which in its current form is probably unlikely. While every research method has its own biases, the biases associated with Twitter are far more limiting than phone, mail or even internet methods. The digital community's concern with speed and quick reponse is commendable, but it often gets in the way of doing research the right way.
Right now, we are comparing apples to oranges: Nielsen is a form of measurement. Twitter is a vehicle of expression.
That being said, I was under the opinion until recently that consumers (twitterers) don't microblog about inane things like consumer packaged goods until I had to go under the hood of Twitter and several social media sites for a client presentation, confident I would find little of value. I was absolutely wrong & have the calluses on my fingertips from typing all weekend to prove it. Consumers may not blog about the attributes of the product, but they do blog about the best place to get their favorite X at the best price. They also tweet about offers by the company (and those may be company representatives doing the tweeting – remember the old New Yorker comic adage: on the internet, no one knows you’re a dog!)
So, to the points of many – we don’t know WHO is tweeting about X, which marketers care about. If it isn’t their target market, a company and its ad agency won’t care if they’re tweeting or not. If it is their target market, they want to know everything they’re tweeting about.
But let’s think about this a little differently for a moment. There are certain layers of Market Segmentation that don’t change: race and gender. Others change more slowly: Age, geography, household income. And then there are those layers that can be influenced directly by marketers: psychographic and behavioral. This final, more translucent layer, is the one we need to pay special attention to in the social media space. Consumers are becoming increasingly savvy about getting the lowdown on a new product by consulting the opinions of others before they make a purchase. Identifying like-minded individuals and categorizing these opinions to learn about product perception is something we are just beginning to learn how to do. Sentiment toward these products may not yet be measurable in the traditional sense (i.e. Nielsen). But for some products and categories, (especially for politicians) I argue that tapping into Twitter and other social media sites will become MORE valuable than existing measurement tools. Especially when we can tell from the subtle nuances of word choice, whether the person who tweets is obviously a Republican or a Democrat. Conservative or Liberal. If they fit into a certain psychographic “tribe”. In short, we WILL be able to know by your tweets if you are a consumer (or voter) worth targeting… or if you are simply a dog.
Dave...and friends,
This discussion is an example of why Internet research and audience measuremen is so poor. Please go back and learn good research tecniques. Sample representativeness, projection, balancing panels, etc.
Yes, Twitter is a good focus group. But that is about it. Test your hypotheses with statistical rigor.
Wow, the delusions of uninformed originality!!
Research conducted through Twitter and all other forms of social media can be made measurable and reliable when using using ORM software. This enables both qualitative and quantitative research through analysing the conversations in terms of relevancy, credibility and sentiment so both the specifics and the overall trends can emerge. We have created a tool, BrandsEye, which enables customisable, real-time analysis, enabling the researcher to glean valuable insight from the proliferation of socially generated media mushrooming all over the Web.
Simply put, you can qualify and quantify what people are saying about your brand or specific campaigns via Twitter (and over 30 billion other sources) using BrandsEye.
I think the recent Harvard Business Review says it all:
http://blogs.harvardbusiness.org/cs/2009/06/new_twitter_research_men_follo.html. Based on their research, 90% of the "tweets" are coming from 10% of the users. The research showed that over half of Twitter users tweet 1x every 74 days. This essentially nullifies using it for research until it is embraced globally and people start tweeting. While Twitter is great for culling for feedback, we should not confuse feedback with research.
I'm glad I waited to comment - I actually thought the article was a joke.
Re: intention data and authenticity - since when is it guaranteed that people are telling the truth online?
Re: it's free - yes, and you get what you pay for.
Re: real time results - what serious marketer would accept giving up precision? Precision is what it's all about. We all have seen how a small error can result in catastrophically huge mistakes.
When the "sample" is not random nor representative of a population, there are no data.
I too thought this was a joke. The researchers and statisticians amongst us have queried Twitter's representativeness. Well, I'm here to tell you that it is representative ... of the hyperbole and total lack of distinction between data and research that abounds in the online world particularly among the start-ups. They see a data stream and say "wow ... with this I could be the next Nielsen". Never let rigour and robustness get in the way of some good PR spin that might suck in a VC partner!
A couple of comments on threads among the responses. First, set-top-box data will not supplant Nielsen either for one simple reason. STB data is not viewership it is Household TV tuning. I know of no marketer whose target is "households". STB data will be part of the new TV ratings metric I have no doubt, but it will be a hybrid of STB household tuning data and panel-based demographic viewing data. (The same will apply online with server data supplying the quantum and a panel supplying the demographics).
Also, people seem to be hung up about the Nielsen sample size. There is a point in every measurement system, when adding more sample adds no more precision - all it is doing is adding more cost. You just keep adding more and more meter households, you get "the same ratings" (within the sampling tolerance of course), you send yourself broke and make Nielsen richer! A nice analogy is when you're sick and you go to the doctor who is unsure what ails you, so her orders a blood test. This blood test is generally 5ml of blood out of the 5 litres of blood in your body - you don't need to take all 5 litres and do a thousand tests!
Lots of great comments here. One of the reasons that I wrote the headline as I did was to get people to focus on another data set as a possible replacement to incumbent audience measurement and consumer research panels. What went unsaid, but is certainly suggested here is that while Twitter may not be ready for "prime time" as a research product, television set-top box data is. No amount of "scientific projection" of thousand person samples will be able to match the data from millions and tens of millions of second-by-second measurements recorded by set-top box. Twitter may not be the next Nielson, but census-based set-top box data is quite likely to supplant panel ratings soon.
I actually think the premise is sound - frankly, we've been "replacing" (or correcting, amplifying) traditional forms of consumer research with listening thru social media channels and platforms like Twitter for the past few years.
But Twitter is just one input feed into this emerging, more authentic (more accurate!) way to gauge what people really think, are doing/saying, ready to recommend/trash etc when it comes our brands, products and services.
Like we do, you can use a range of tools to gather these converational feeds into robust data sets for modeling, slicing, dicing and - yes - measuring the hearts, minds and habits of consumers. Try it and see how much closer it gets you to what the customer wants and needs more than the old school tools.
So yes, the Nielsens and even the Forresters and maybe even the IRIs of this world should certainly be looking worriedly over heir shoulders. Morgan's right to suggest their days are numbered when it comes to giving marketers quality, actionable intelligence from which to build our winning insights and campaigns.
But it won't be a single channel platform, like Twitter, that kills these giants. It will be smart marketers listening, really listening, to the unfiltered, on-provoked voices from the marketplace --- applying a rich mix of new tools, channels and feeds that no one ever dreamed of when these big old dinosaurs of consumer research were born at the dawn of the previous age of advertising.