Here’s a question for senior marketing and sales executives who aren’t necessarily data scientists: Which of the following statements is true?
- Most companies don’t have enough data to perform optimal analytics.
- Most companies don’t own or control much of the data that’s rightfully theirs — and that could provide true business value.
- Most companies could fall out of compliance or suffer a data breach, not because of the data they have, but the data they don’t have.
Answer: They’re all true. To put it bluntly, companies that should own the data, and technically do own the data, often don’t have the data. Taking it back will drastically alter the data-driven business dynamic, better guard against non-compliance and asset theft, and benefit the bottom line.
This is a classic business transformation, enabled by technological innovation. In the past, keeping all data first-party and in its source form, and allowing custom apps from outside vendors to access it for functions such as attribution and segmentation, would have required considerable investments in technology infrastructure and related costs. Today, it’s all done with a SaaS model, and the same IT and data teams that oversee other disciplines can manage this work.
It’s reasonable to ask why this has taken so long. A little context helps: Some forms of data science are more scientific than they seem. In modern-day marketing, data is clearly the lifeblood of the ad stack, the raw material for audience segmentation and user linkage. However, the journey from data generation to collation to storage to analytics to intelligence to a full-fledged initiative is long and tortuous—and a lot of material gets jettisoned along the way.
We all assume that the sheer proliferation of digital devices, applications, channels, and more create direct connections between brand and customer, but that’s not how it works. Sure, this is a universe defined by technical sophistication and widespread user adoption. But largely because of the volumes of data involved, there are numerous third parties embedded in the process of data collation, aggregation, and syndication. These are effectively ‘middlemen’ who attach their tags — cookies, mobile IDs, etc. — to the advertisements and related content, then provide the segmented data they gather to the brands running (and paying for) the campaign.
There was a time when this was the only way to get it done, and there are still situations in which such arrangements add value. But overall, the process is inefficient, unproductive, costly, and thanks in part to mushrooming compliance mandates, potentially risky.
Imagine a large enterprise that’s not in the data business per se but understands its value. It runs sprawling marketing campaigns that encompass pop-up boxes, banner ads, rich media displays, how-to videos, social media meta-tags, search engine optimization, and more. The campaigns extend to every corner of the digital universe, making it a massive undertaking.
Of course, the digital content is not served up randomly; there’s some data-driven intelligence to guide placements. After that, the process is largely anonymous. A series of data management platforms (DMPs), also called data exchanges, essentially append their tags or cookie IDs to each segment and link that ID to the ad-serving environments. In the end, a tiny, tiny share of the total audience actually clicks through the ads. That’s about all the enterprise bankrolling the campaign actually has to go on.
But is that really the only data available?
What time of day did each individual consumer receive a particular ad? What particular piece of content were they consuming when the ad came up? Who were the consumers that watched part of a video but didn’t click through? How long did they watch it? What device did they get it on — PC, smart TV, set-top box, mobile gadget — and what was the ID or IP address? Can that data be correlated to other ID information in the company’s database? What other related ads or content did they respond to, and when, and how?
The list of questions is potentially endless, and so is the data generated from the multiple points of engagement. They constitute massive data streams of tremendous value that end up in disparate silos belonging to the DMPs. It’s invariably outside the reach of the brand that pays for the campaign and owns the data; the advertiser doesn’t really get to see it. Having it in-house — taking the data back from the third parties that collate and store it — will affect every business initiative.
Along with the obvious sales and marketing benefits, there are advantages to be gained from ownership. For example, when there’s a need for data attribution, segmentation, or orchestration, it’s done by having outside vendors dip into the company’s own data lake, not the other way around. This enhances efficiencies and cuts costs.
And to dispel another myth, there’s nothing proprietary about this data—the brands that own it can use it to guide business initiatives without running afoul of governance directives. In fact, keeping the data in-house is actually the best way to ensure ongoing compliance with evolving regulations.
Clearly, the digital ad market — in many ways a paragon of technical sophistication — has some very old-world challenges. For instance, the incessant revenue sharing can cost data publishers up to 90% of the overall budget. The ‘middleman’ arrangement means there’s little transparency — data companies don’t always know who’s buying the data. And of course, the third parties don’t reveal much about attribution — how the data is performing — because they’d lose control.
There was a time when middlemen were needed to manage voluminous data streams. But now, even as those data volumes continue to mount, solutions are available to tag, collect, and traffic all data sets, as well as store, manage and connect that data with multiple buying centers. The brands that own the data can actually have the data. And that will be the best way to make the data pay.