
By: Rob Young
There are two large Shell office complexes in London, located just across the Thames from Big Ben – the Shell Centre and, just downstream, the Downstream Building. Staffers in Shell’s intellectual property division, who occupied the building, insisted the downstream moniker came from its location relative to the flow of the Thames. This must characterize a peculiar kind of humour unique to engineer/lawyer Brits because the term “downstream,” in fact, is used in the petroleum business to represent refining, marketing and distribution of petro products. The “upstream” sector covers oil exploration and production.
Alex Ross, in his book The Industries of the Future, uses the terms upstream and downstream in relationship to what he calls the raw material of the 21st century: big data. Upstream big data is constantly being created by consumers, advertisers and media (amongst other producers like stock markets and meteorologists) and downstream refers to the refining and harnessing of big data for commercial purposes.
The designations of upstream and downstream big data is useful as we puzzle through the efforts being made by the legacy-based media divisions of Canada’s big telecommunication companies as they attempt to manage, commercialize and monetize their big data.
Google/YouTube, Facebook and other successful internet companies have, from the beginning of their commercial lives, been involved in the production (up) and application (down) of big data. When advertisers utilize these digital platforms, their big data becomes visible to the number crunchers residing within them. As a result, direct connections are drawn by the internet vehicles sales and research teams, between the generation of big data and the refining of the data leading to commercial application. The internet vehicles can hand the advertiser a handy dandy review of the campaign that says advertiser’s campaign “A” generated this data “B” which led to this finding “C” which can be used to improve the advertiser’s campaign “D.”
Turning to our petroleum metaphor once again, addressable media channels (internet) have integrated up and downstream big data sectors.
The big telecommunication companies here in Canada, (Bell, Rogers and Shaw), generate huge quantities of upstream big data but are disconnected from their advertiser clients’ campaign results. In fact, in many cases, their advertiser clients themselves are disconnected from campaign results. Non-addressable media (legacy media) are finding it difficult to connect their upstream big data resources to downstream commercial application.
This disconnect keeps media people, on both the buying and selling side of the business, awake at night.
Media agencies, marketers and DSP’s are all working to build data management platforms (DMPs) in an attempt to make the downstream data more refined and more valuable to the end user.
Some telecommunications companies are displaying silos filled with upstream data to advertisers and entreating these advertisers to partner with them in an attempt to develop their downstream commercial applications. The advertisers want to control the process and move the telecommunication data into their DMPs. The telecommunications companies want to control the process and move the advertiser data into their evaluation systems. In other words, carrying the petro-metaphor further, some advertisers are expressing concern that by participating in these partnerships, they are helping to pay for a pipeline that will distribute learning to their competition.
Other telecommunication companies have developed a few downstream solutions, built limited silos of upstream data to support the solutions, and are out searching for client problems to solve.
And so, in summary, our media big data environment is confounding. We see silos of data waiting for pipelines to be built, but few refineries around to accept the data once it gets flowing while media agencies try to build their own vertically integrated DMPs and in the meantime media researchers run with buckets to scoop up very particular bits of data and try to find very particular unrefined insights to very particular parts of their media campaigns.
It is all very exhausting. But the potential rewards of a connected up/downstream big data system are so attractive, no serious media player can simply sit on the sideline.
To demonstrate how a telecommunication’s company’s legacy upstream big data can be tied to advertiser downstream consumer’s transactions, consider the following story about Jane Doe’s consumer journey. The story is about an individual but in reality the connection between up and downstream big data involves non-personally identifiable information (non-PII). In other words, the data is about Jane Doe’s individual actions but her identity is hidden from the number tumblers.
Jane Doe got on her laptop at home in search of information about cars. She accessed an auto company’s build and price web page, created her own vehicle, painted it blue, made a note of the price and signed off. Her laptop was tagged by the site and display ads were sent her way featuring a pertinent nameplate in blue (retargeting). Judging from the nature of the sites the re-targeting ads occupied, Jane Doe was assumed to be a women in her 30s.
The code ID from her laptop was cross-referenced against data in the Bell/Rogers/Telus big data silos because one of these companies likely provided internet services to Jane’s home thus allowing the household ID to be established.
Jane’s household also subscribes to three separate cell phone data packages and since most households use one telecommunications company for both internet and cell phone, it is simple to associate one of those three cell phones to Jane. But which one? That would be the package for the smartphone that had a browser history best matching a female in her 30s.
Jane’s mobile was detected entering three auto dealerships including the dealership that offers the blue model build Jane created a week earlier.
And so here we have a testable proposition. Did re-targeting increase the likelihood of Jane Doe-like consumers visiting the dealer’s showroom?
But there’s more. The telecommunications company also owns broadcast distribution units (cable or IPTV) and so Jane Doe’s household’s set top box tuning big data can be screened.
Another testable proposition appears. Examine the auto company’s TV weight levels against Jane (and others) and determine what the relationship is between weight level and likelihood to visit the showroom.
Within this story resides several examples of how connections can be created between up and downstream legacy big data media sectors and how beneficial insights can be gleaned.
Big oil fuelled our 20th century economy and big data will fuel our 21st century as the big data pipelines/refineries get built and up/downstream sectors get connected.
Rob Young is SVP, director of insights and analytics at PHD Canada.