Register Now
Lotame Ignite APAC Join Us in Syndey on Feb 28th! Learn More

Bigger Data Isn’t Always Better

July 18, 2016

Marketers plan to double their spending on big data by 2018, a recent CMO survey found. However, companies “underutilize the marketing analytics that they’ve requested,” according to Christine Moorman, a professor at Duke’s Fuqua School of Business. Every day, marketers gather more data than they can use. Vendors sell marketers more data than is useful.

Why is there a gap between collection and utility? In large part, some of problem can be traced to ad tech’s decadelong arms race to overvalue quantity and undervalue quality.

Sell First, Ask About Performance Never

One of the byproducts of an arms race – or, in this case, a data race – is that buyers and sellers become irrational over time. Supply and demand leapfrog each other to dizzying heights of abundance. Eventually, both sides lose sight of their purpose and simply buying or selling more data becomes the goal.

But isn’t the real goal to run the most effective campaign possible?

Campaign efficacy should be the goal for all involved in ad delivery, but it isn’t for today’s data market. In reality, sellers help buyers zero out their budgets because if they don’t, another vendor will. If all data were equally effective and were fully commoditized on an open market, this wouldn’t be so problematic. But not all data is created equal, and so marketers end up spending their entire budget on data of dubious efficacy.

Marketers are in an unenviable position of having to select data with no information about its impact on the campaign at hand or to use anecdotal prior information. How can this be when today’s targeting technology is so advanced?

It’s relatively easy to construct thin, data-driven campaigns where the targeting is excellent. Search and retargeting are examples. But achieving the scale a marketer demands using these resources is a challenge. Where a thin campaign is defined by quality, a media buy at scale typically includes plenty of proxies for these intent signals. Data buyers struggle to select data points that may have been packaged for a variety of purposes and hope that they prove effective for their specific KPI.

Eventually, the data vendor’s incentive becomes clear: Keep supplying more data to make the campaign deliverable. Over time, that motivates vendors to create more data than their customers need. Worse, CPM data rates can reward selling data regardless of efficacy.

Incorporating Performance Into The Marketplace

Right now, vendors are paid on a CPM basis regardless of how well the data they sell performs. Many vendors make a handsome profit selling data that doesn’t perform any better than the most basic demographic matching. That is a broken model.

A better model is a market where data buyers can reward vendors that add value and punish those that don’t. If a vendor’s data performs on a particular campaign, it should likewise command a premium for that campaign. If another vendor sells data that is well packaged but ubiquitously available, then it should sell at a lower price that reflects its provenance.

Both buyers and sellers need a true market for valuating data, one where price is tied to performance. Such a model would benefit marketers because it would increase the performance of their ad spend. At the same time, vendors would also benefit from a more transparent, performance-driven market because the buyer feedback would incentivize them to collect and sell only the most valuable data, as opposed to simply selling the most data.

In the short run, shifting to a performance-driven data model will most benefit the vendors who currently provide specific data segments close to consumer intent. But soon enough, a performance-driven model that favors buying smart data over big data would benefit all market participants. Data vendors will be able to export more raw and unrefined data points and earn vastly higher returns. Data buyers will be able to purchase effectiveness rather than bundled CPM data.

Written by Omar Abdala, Chief Data Scientist – Lotame
Originally posted on AdExchanger // Wednesday, November 18th, 2015