What it is: There are petabytes of data flowing every fraction of a second from social media and ecommerce, and most of those bytes hold some critical data that can feed a marketing strategy. The ad technology industry lives on the tracking and measurement of that data, but the value of those raw numbers to a marketer depends on how that data is analyzed. That’s where data-relationship analysis comes in.
How it Works: A marketer needs not only to understand that online sales went up after a display campaign was launched, but what the statistical relationship of those sales figures are to display click-throughs, social media conversations as well as the overall economy. That’s where relationship analytics comes in. Basic analytics puts data in a silo; 20,000 teens in Iowa mentioned Miley Cyrus’ jeans brand on Facebook after the brand’s ad ran a locally-targeted homepage in Des Moines, and the ad had a .02 CTR. Basic analytics won’t tell you deeper information that places the raw numbers in a larger statistical context, such as perhaps that Cyrus had a wardrobe malfunction and that was being tweeted at the same time or that there was a local sale going on at a chain store. Data relationship analytics looks at the complexity of multidimensional data and allows analysis to be based on established data relationships, so that faulty logic is less likely to creep in a skew insights one way or another.Relationship analytics often uses NoSql data methods, which places data and their reactive relationships to each other into graphs, not only into table format. This allows data miners and analysts to see the interaction of variables — a sale or a current event– with available data and see a more panoramic view of strategy effectiveness.
Who is Using it: Cloud-computing services that integrate analytics with content publishing and ad serving technology, Yahoo and Google ad networks and smaller companies like InfiniteGraph that provide analytics technology
Why it Matters: Seeing relationships in data can save a company millions in ad spend misdirected to a dubious or even non-existent audience segment. The display industry needs more than a little help in getting closer to being able to create a more measurable process for looking at branding success and engagement.
Assessment: Bad data is shorthand for strategy failure. The closer analytics can get to full-funnel attribution as well as more well-formed audience segments, the better off the industry, and the brands that depend on them will be.
https://digiday.com/?p=2240
