OMMA Video Panel: Metrics
Monday, June 16th, 2008Moderator: Dan Ackerman, Sr. Editor, CNET
Andrew Budkofsky, Break Media
Lynn Bolger, comScore
James Kiernan, MediaVest
Maniak Mazumdar, Nielsen
Even Silverman, Lifetime Networks
There is a lot of discussion about ISP data and how it can be applied to the web. comScore and Nielsen may not be as accurate down the long tail.
Long tail is important, but there are challenges to that. If you are on the internet for a long time, Nielson would like to be able to track you.
What are companies doing to bring stability to a tumultuous situation?
It is more about looking at 3rd party ad serving metrics, rich media partners, and fusing all the data together so that it is actionable, and we can apply the learning to our next campaigns. Lots of success using dashboards to pull together data from disparate stories.
Is there a danger of having data that can’t be validated or even recreated by other parties?
Yes, there is some, but it isn’t like people are pulling from obscure data sources. Agencies do need to be more transparent about the data they are aggregating and using.
Time spent is an idea metric in the gaming sections. In other sections, it might be page views.
How do you choose what to cherry pick for data? Do you strive for accuracy and consistency first?
It is most important to be accurate, no matter what the methodology. You need to find out what is important to the client and the clients goals. We need to be honest about the ad experience.
Advertisers are all looking for different things, so it is difficult to have a uniform methodology.
Maniak, Nielsen: We’re hoping that people don’t have to cherry pick. We can provide enough data and consensus around that data.
Dan: Lets focus on time spent.
James Kiernan: time spent is a nice proxy for engagement. But until we can measure time spent’s impact on offline sales, it is sort of meaningless.
Lynn, comScore: Which metrics are you going to use to build a business model? Transactions are based on site side data. Demographic data comes from panels. How do those pieces fit together to create a marketplace? How does a non-linear video experience compare? And most importantly, which piece do you want to negotiate on?
Dan: Who is being undercounted under different methodologies?
Maniak: We see both sides. When it comes to volume metrics, server side is more accurate. But there is a big “but.” There are lots of bots and spiders out there, so you need to clean the data. If you are using server side blindly, you are basically targeting machines. Panels are great place for audience metrics.
Dan: What about demographic groups being under represented?
Maniak: Young people are under reported. But if you believe in statistical sampling, there are ways to measure and account for that. People say that you can’t measure one audience or another. There ways to adjust for the “at work” audience, for example.
Lynn: There are challenges with kids, not just online. The at-work audience is also difficult. The practical reality is that there are firewalls.
Evan: What about shared computers?
Lynn: We identify the individual user on the machine at the time.
Maniak: There are log ins for each person. We can leverage that to distinguish individuals from each other.
Lynn: people give up a lot of personal information that we can use.
Dan: What about other companies, like Compete.com, that put out data that you can quote, even if you don’t know where it comes from.
Lynn: Use it at your own risk.
Dan: What is the ideal set of numbers?
Andrew: it would have to be a direct measurement of some kind. There are just too many discrepencies right now. It would be mostly server based, but can also be cookie based. But we need to use what the agencies are using.
Lynn: It will still be panel based.
James: Hybrid models are very intriguing. They bring the best of both worlds. I’m waiting for someone to bring it to market, but it will be more on the server side.
Maniak: We measure people, no matter what the devise or platform. There are always gaps in measurement with new technologies.
Evan: As someone who works for a cable company, merging measurement for online and offline is the most intriguing to me.
Lynn: the question of accuracy is critical. Measurement of the transaction is critical. When it comes to looking at what will happen in the future, we need data that is predictive. The role of panel data brings an awful lot into the planning process and will be required today and moving forward.
Q&A: With distribution models changing so much, how are you tackling those issues?
Maniak: If it is TV content, there is a watermark. The distribution issue isn’t a problem, at least not for TV content.
Lynn: there are business questions that need to be considered. How the inventory is packaged and sold. There is a mapping system that tracks content and where it is played, so we can bring it all back together.
Q: With people surfing the web on multiple devices, are impressions all of equal value?
A: Yes. Doesn’t matter if you watch it on an iPhone or the web.
James K: The TV networks have it backwards. They have a model where they can’t really prove the value of an impression.
Q: Is there more value to watching something on TV or downloading it on your xbox 360 on your TV, is it the same experience?
James K: We’re setting up our agencies so that it doesn’t matter where you watch your content. There won’t be broadcast vs. broadband buyers.
Lynn: there isn’t going to be a universal solution. Some marketers will want it all in one place. Others will want it segmented. But the outcome is what is important. On mobile, you can determine if someone walks into a store with GPS. The outcome metrics need to be different than if you watch on TV.
Q: Since there are discrepancies, does comScore work with publishers to make sure that the panel based data is closer to the server side data?
Lynn: We will always work with anyone on the technology to make sure the data is accurate as possible.
Q: How do you track embeds of embeds and downloadables?
James: I’m very intrigued by using watermarks to track content.
