“Data Driven Thinking” is a column written by members of the media community and containing fresh ideas on the digital revolution in media.
Today’s column is written by Ken Rona, PhD, V.P. Data Strategy and Client Analytics at [x+1].
Over the last couple of years, I have been in the data business, selling and buying, respectively. My colleagues call me “the data guy.” But I don’t think about my work that way at all. As I see it, I am a user of data; I provide data-driven advice (and I think our clients would agree). But, when I am faced with a new business problem, the first thing I do is spend some time understanding where the data comes from. I analyze how the variables are defined; I ask for the data dictionary; I get some descriptive statistics. One other thing I do is ask “What rights do we have to the data?” That is, what can I contractually do with the data? This opens up an important issue – what’s known as data rights.
At a previous employer, my teams had access to the entire US household file. By that, I mean that we had data for every single household in the United States. That’s a lot of data. We would use the data for customer insight and database marketing. As far as I was concerned, if the data were in the database, we could use it for anything we wanted It was someone else’s job to make sure we could use the data. I never thought about it at the time, but in fact we did not own the data. We rented it. Data from third parties is not sold, it is licensed.
Then, when I worked for a company that provided data to others, I became aware of “data rights.” When we licensed the data, we specified what a customer could do with the data. I am not a lawyer, but the idea is that a user of data acquires certain rights to use data in ways that are specified by legal agreements. The more rights the user wants, the more the data provider gets to charge for the data.
You could think of the length of time of an agreement, or term, as the simplest right. The term, specifies how long you have access to the data — say for a year or a campaign.
Once again, I am not a lawyer, but from a business perspective, here is how I think of data rights. They break down into three classes: temporal, usage, and transferability. First, let’s discuss temporal rights.
Temporal rights are just the length of time you can do something with the data. The example above on the term of the agreement is a simple case of temporal rights. For use in a database marketing application, a finite term makes a lot of sense. Let’s say you use the data for a one-month campaign; so you assume you need temporal rights for a little over a month. At first glance it seems straight forward. Howeveer, what about the analysis you do using the data? Say the data was gender-based and you found that women were much more likely to respond than men for a campaign. You made a chart, showed it to your client and went on with your life. Who owns the analysis you did using the data you had rights to for a month? Data vendors would like you to destroy the analysis at the end of the term. They are thinking: “You paid for a month of access. Anything created with the data has to be destroyed after the term.” Meanwhile, clients want to hold on to the analysis, often for very practical reasons. Their reaction: “How am going to destroy every copy of an analysis? Besides, how am I going to forget I know something?” In general, the clients win this one, though it does not stop the data vendors from asking.
The lesson here is that when you are reviewing a contract, make sure that you don’t sign something that you can’t enforce. You are the buyer and if a data vendor wants something that you can’t do, talk to the vendor about it. I have found most have been very reasonable.
The next type of rights, usage rights, specifies what a buyer can actually do with the data. There are three broad classes of usage. First, a user can review the data to report and analyze something. In the ad business, agencies and ad networks can provide audience analyses that show who viewed a campaign or profile and who clicked on an ad. For this usage, you are using the data to understand something. It is sometimes in the data vendor’s best interest to make its data available for analysis at a very attractive rate, especially when the vendor or its clients are new to the market. Data vendors are interested in creating a demand for their data and it is difficult to know if a given data element is useful without some analysis. Also, there are analyses… and there are analyses. Sometimes a vendor may give you the right to conduct specific types of analyses with their data. For example, descriptive analyses may be fine, but regression is verboten. Just something to be aware of.
In addition, the data can be merged with other products and used to create “derivative works.” I have seen contracts that do not allow any derivative works, even for internal use. It is very hard to use third party data if you cannot do transformations (for example, you take the average of two sources of income). I won’t recommend signing a contract that forbids derivative works. The data providers are solving for a particular business risk; they don’t want you to transform the data, delete the original data, and use the transformed data for your business. Another possible reason for restricting the creation of derivative works is that they may want to preserve the right to combine data sources themselves and create new products. Typically, you can alleviate this concern if you just tie the term of derivative works with the contract term. Another possible reason for restricting the creation of derivative works is that they may want to preserve the right to combine data sources themselves and create new products.
The data can also be used “commercially.” That is, the data can be used to directly target an ad. At my company, this is the core of our business. We use data to find a valuable audience for a given advertiser. For this kind of usage, data vendors, rightly ask to be paid. We are making money directly from use of their data. In the offline world, how you use the data commercially is important. There is often one rate card for “univariate” use of the data vs. mulit-variate usage. By univariate use, I mean targeting a HH on the basis of filters. It’s something like demographic targeting where your target market say is men who are between age 18 to 25. Multivariate usage is when you use the data in a statistical model like regression and it is combined with other variables to find a target audience. Because multi-variate usage combines the data with many sources, it is typically cheaper to license than a univariate usage.
The last type of data rights pertains to transferability. Contracts often specify under what conditions the data is transferable. Sometimes the data and any analysis generated from the data are supposed to be used by the licensee and not shared with third parties. Other times, the contracts allow for a license to resell the data. Be sure to take a look at the kinds of uses you want for your data and ensure that you have the transferability rights that you need for your business need. Obviously, reseller agreements are going to look very different than data that is used for internal audience analysis.
Follow [x+1] (@xplusone) and AdExchanger.com (@adexchanger) on Twitter.
