User/Site-centric measurement — Audience measurement, ratings, most visited, most views…(1)
First and foremost, why we do even need those Audience measurements? Let’s just take a moment, and forget about all those announcements where TV station(s) proclaim themselves as the most watched, or even more often lately, all those articles on domestic websites where they proclaim how they are most visited/read… second/first/third on the entire market, by some paramether.
Audience measurement plays an extremely important role, directly connected to the advertising market. Data from audience measurement is the base for any media planning, on TV it is even a currency. Most TV stations sells some kind of rating point (GRP/TRP) — not seconds, which are coming directly from Audience measurement. So it is more than clear how that can affect TV station incomes, because the price of every single TVC broadcasting is directly calculated based on how many people (not real people) — but number of people estimated by the audience research who supposedly watched that TVC broadcast.
Regarding Internet advertising, Audience measurment don’t directly affect the values on invoices, but still media agencies use this data to do media planning. Based on affinity data for some specific target audience, number of visits or number of people who visit some website or some specific category of some website — and again, as for TV, not the real people but estimated number of people by the audience measurement.
CPD (cost-per-day) or CPM (cost-per-mille) are still dominating methods for buying Internet adverts, so audience measurment will not directly affect numbers on the invoices. Nevertheless, I can say from my experience in 2008 that most of large international clients, had set existence of audience measurement as a prerequisite before they even consider spending a dime on local Internet advertising market.
On smaller markets, as such as each of our regional markets individually, it is quite possible to cover entire market via site-centric audience research, and to get quite precise data about website visit, users demography, overlapping users between websites, overall nice and detailed landscape of the entire market.
“Site-Centric” is standing for a type of research based on measurement on website side. Often done via javascript code snippet, embedded on every page of the website participating. Those Js snippets allows us to “track every individual user” and their behaviour (via cookies or some similar method) as user browse different Internet websites. But again, only those which have installed JS code — and took part in the research.
Besides Site-centric research type, there user-centric or panel method — in example AGB Nielsen uses this kind of research to conduct TV ratings measurement / TV audience research (in Serbia). In short: they have a gather representative panel sample (880 households) that represents entire Serbian population (over 2.5 million households). All those users have installed a device on their TV sets, allowing a company to track what program is on at any precise moment in time.
Both methods for audience research conduct has benefits and disadvantages. Main disadvantages for site-centric method is mentioned earlier, we can track only those websites who are participating in the research, or had implemented specific JS snippet. And implication is that we will never have data for most popular global platforms such as Facebook, Google, YouTube — cause they are just not too eager to implement some third-party codes for each market where they operate.
Primary disadvantage of user-centric research is before all gathering and maintenance price for that user-panel. For our total TV audience, measured 880 households can accurately enough represent entire 7.5 million TV population. But even that had specific accuracy on relatively small number of TV stations with national coverage. After digitalization of terrestrial TV signal, and rapid cable providers expansion, which resulted in much much wider number of channels available nationally, accuracy of this research fell in great amount — some statistician among you can calculate how much exactly.
By the sole nature of Internet, where number of websites where user can end up is close to the infinity, to have accurate audience research you will have to have much larger user-panel — and more people equals more money.
There is also a third, hybrid model which combines previously mentioned methods, from site-centric measurement we take traffic data and combines it with user-centric data to estimate traffic and audience of non-measured web-sites. But than again this hybrid method succeed disadvantages of the user-centric (panel) method — price of the panel.
Accuracy and reliability of Internet Audience measurement via site-centric method, greatly depends on ability to recognize same users on multiple websites, and current ways to do so is heavy dependent on web cookies — but there are some advanced methods for recognizing and tracking users that are in the research phase.
Some of available audience researches in Serbia are:
AGB Nielsen — people-centric panel research for TV
Gemius Audience — site-centric audience research for Internet
DotMetriks Audience — site-centric audience research for Internet
In next article we will go in detail in Internet Audience measurement, and cookies and new technological challenges that usage patterns create for companies who conduct these researches.
