RTD: What are the most important things that your clients are asking about data now?
Dimitri Vaynblat: They want brands’ customer relationship management (CRM) data and other data related to offline activity of consumers and brands’ Web site and app data (if the brand has an app). This data comes from online or mobile analytics solutions and/or brands’ own digital logs.
They also want brands’ social data—that’s data on user interactions with brand pages on social media. This is unique data that competitors don’t have access to.
RTD: What are the challenges to obtaining this type of data?
Vaynblat: Apart from data collection, the challenges are how to cleanse and normalize the data; how to connect data collected through various channels and devices to the same user; how to activate the data, or make the data work in marketing campaigns that run in various channels and devices; and how to measure impact of the data on marketing campaigns.
RTD: What do you say to customers spending hundreds of thousands of dollars on third-party data that doesn’t perform?
Vaynblat: When data of various third-party vendors are compared, there can be significant discrepancies. Even worse, data is often inconsistent within the same third-party data provider. For example, basic demographics of the same user (such as gender or age group) can be different on different days.
For brands to develop actionable data strategies and execution that drives efficiency on their advertising campaigns, they first have to realize that not all data is created equal. After this realization, brands need to understand the strengths and limitations of various types of data available to them, as well as required algorithms, tools and supporting infrastructure that make usage of these types of data most efficient. This applies to both first- and third-party data.
RTD: What’s your approach to understanding the value of data as it relates to campaign outcomes?
Vaynblat: We look at relevance, data quality (in particular, reliability, granularity, and recency) and uniqueness.
RTD: What should brand marketers look for in data partners?
Vaynblat: Brands need data and technology partners that can capture unique “differentiated” data—that’s data beyond pixel-based Web site visit data—such as social data and mobile in-app data. They need to connect disparate sources of data and extract campaign-relevant signals from less relevant data via modeling.
I think brands should share data only with a few “campaign delivery partners." For example, strong signals and high ease of extraction data (i.e., remarketing pixels for a campaign with misaligned goals) requires less technology and modeling. As a result, the barriers to entry are low, and a lot of undifferentiated players could get strong campaign-related signals. This situation is very susceptible to price wars, increasing user acquisition cost.
RTD: What should brands look for in third-party data?
Vaynblat: When brands use data bought from third-party data vendors and data exchanges they face certain challenges. Often, the data is lacking in quality, relevance and uniqueness. Third-party data may be less reliable.
When brands need to augment their own first-party data with outside data, brands could work with “campaign delivery partners” that collect their own proprietary data and use them in their own campaigns. The data isn’t sold so that brand competitors could use it.
RTD: What do you see as the biggest challenges for data?
Vaynblat: Our clients all want some kind of transparency. It’s an overused term—but for them, it’s about the cost. They want us to be transparent about cost.
There are also different types of targeting strategies. Clients want us to give them a breakdown based on the strategies. In most cases, agencies and brands use proxies when they measure campaign success. In most of the cases, it all boils down to how much did I spend on media and how much on data.
And if I’m a client and I’m working with a third-party DSP, the client wants us to tell them how to translate the results.
Also, companies collect third-party data and put pixels on advertisers’ sites, so the problem is that the data isn’t unique.