Facing Facebook, Google and Amazon, brands pool their data

As the internet continues to blur many competitive differences like pricing and product selection, brands have increasingly seen their customer data as a key differentiator.

But, as it turns out, there’s a bigger competitive headache beyond whether a customer is just a click away from buying something like what you’re selling, but cheaper. It’s the planet-sized customer webs run by Facebook, Google and Amazon — online ecosystems so large that their centrality to advertising and retailing threatens even established smaller brands.

To gain the scale they need to compete, many brands are pooling their data and sharing it in some iteration of cooperation. The competitive angst of giving your treasured customer data to competitors and potential competitors — even when it is anonymized and grouped into segments — is exceeded by the need to know what your visitors/customers are doing elsewhere.

This data can help target and retarget customers, know what they are interested in buying and when they’re ready to buy, and know which devices are owned by individuals or households.

Ironically, pooling your proprietary data can help companies better compete. This is more than, say, the ad hoc exchanges that characterize second-party data sharing. Rather, they are organized efforts by publishers, retailers and others to learn more about their visitors and customers by helping other publishers and retailers do the same.

Late last year, for instance, UK-based affiliate network Skimlinks launched Audiences by Skimlinks, a data co-op that pulls anonymized customer behavior from participating publishers and retailers, employing user paths from content to click to purchase as guides for advertising and other targeting.

Skimlinks starts with an affiliate network that tracks clicks and sales for more than 1.3 billion unique users worldwide, more than 55,000 digital publishers and about 20,000 merchants. The company has not revealed the number of publishers/retailers participating in the co-op.

The co-op features high-quality information, because brands are sharing their first-party data and because users’ clicks on affiliate links indicate interest. Brands and agencies can target segments of users from Audiences through such data targeting platforms as MediaMath, BlueKai, Lotame and Krux.

“The more data, the more accurate”

At its launch, Skimlinks CEO and co-founder Alicia Navarro described the value of the user trail in the co-op:

… a user on Publisher A’s site could click on a link about a handbag in a story and be taken to a retailer’s site that sells handbags. If the user buys the handbag, there’s a page impression > click > conversion trail.

Even if there’s no purchase, there’s a pattern of behavior for that user when that action is added to other page impression > click > impression paths. That user can be targeted for, say, handbags on other publishers’ sites through the cookie that Skimlinks has dropped on her computer.

Each of about 200 product category segments will be divided into “wants to buy,” indicating interest but no immediate plan to buy, and “about to buy,” for users actively looking to make a purchase. These represent, Navarro noted, the top and bottom of the sales funnel.

Navarro noted that advertisers pay on a cost-per-thousand basis for the data. The income, she said, is shared among co-op members, based on the amount of data they provide and the closeness of their data to a sale. Data about a page impression, for instance, is worth less than data about a click.

“As far as we know,” she said, “no others [i.e., data co-ops] pay publishers and give data back.”

The key reason behind cooperating, in Navarro’s words: “the more data we have, the more accurate we can be.”

“Working with a co-op like us,” she said, it “results in better targeting,” since the data is all first-party, compared to a data-buying aggregator like Exelate.

Navarro added that another reason Audiences by Skimlink is unique is because the co-op sees “the whole content [path] to commerce” trail, while other data providers tend to see only retailer data or only intent-to-buy data.

A co-op of B2B intent data

One provider that has made its name by focusing on intent data for B2B is the New York City-based Bombora. It tracks search, white paper downloads, webinars, trade show sign-ups, article and blog reads, video consumption, social shares and other sources of activity by business personnel looking for products.

Publishers like Forbes, Aberdeen Group and about 2,500 other sites contribute data about more than a billion monthly interactions from their visitors, who are usually business buyers, and then advertisers and agencies employ that information for marketing and sales efforts toward groups within targeted businesses.

The company says it was “the first aggregated source of behavioral Intent Data for B2B, creating a ‘first of its kind’ data co-operative of premium media companies.”

All of its proprietary B2B intent data is part of the co-op, CEO Erik Matlick told me.

Suppose CNET wants to sell an ad campaign to a vendor of cloud storage, he suggested. It drops a cookie for visitors who visit its pages and download material about cloud storage. But the tech publication also wants to know other subjects — like backup storage or emergency recovery — that those visitors investigate elsewhere. Matching of the users’ IP addresses and other techniques indicate that those visitors came from, say, Target.

“What we’re selling,” Matlick told me, “is access to what users are doing on the other 2,500 sites” in the co-op. By knowing that most of the visitors to cloud storage info on CNET are also looking up “backup storage” or “emergency recovery” on other sites, CNET can get a better idea of those visitors’ primary motive.

But it’s not unlimited access to all the data, he pointed out, since publishers “only get access to what your [visitors] are doing elsewhere.” Other layers can be added onto this view, such as Bombora’s partnerships with data providers who know the kind of software the inquiring company has installed. If CNET knows that Target has recently purchased new backup software, for instance, the overall picture of that retailer’s interest becomes clearer.

While publishers only get access to what their visitors are doing, and there are a limited number of publishers contributing data about visitors to their sites, anyone can purchase the data in the open market. But data co-op members get a discount rate of as much as 50 percent off.

Adobe’s Cross-Device Co-op

Matlick compared his company’s model to the FICO credit score, where credit bureaus and other trackers of consumer borrowing and paying habits pool their data.

That model, he said, is particularly relevant to surge reports that Bombora has developed, where it can find trends about certain topics. There might be a surge of interest, for example, in anti-malware software, and Bombora says it can help an advertiser or marketer pinpoint the interest, in some cases, down to a specific company’s office.

But you can’t tell those interest trends, Matlick pointed out, unless publishers are willing to share their visitor data.

Another good example of a critical insight that is much more difficult to definitively determine without a massive sharing of data: all the devices that you own.

This past March, Adobe launched its Cross-Device Co-op. It employs logged-in data — contributed by publishers and retailers — to discover the common devices owned by given users or households, so that marketing can continue from laptop to smartphone to tablet, and eventually to other devices.

In its announcement, an Adobe executive indicated his company’s Cross-Device Co-op was a kind of counterweight to the Facebook and Google ecosystems and to the probabilistic device-matching schemes that infer the matches. By contrast, the Co-op uses a publisher’s logged-in data from the same user on several devices to definitively established the group of devices owned by that person or household.

As with other co-ops, Adobe has set some boundaries around brands’ proprietary data. Brand A contributes the group of devices from which each of its subscribers/members has logged on, allowing fellow co-op member Brand B to know that the anonymized user visiting its site on a laptop — assuming there is a match with Brand A’s users — also has this smartphone and this tablet. But site visit histories and other behavioral data are not shared.

Data sharing is obviously on Adobe’s mind, as this follows the November 2015 launch of its Audience Marketplace, designed to serve as a second-party data matchmaker.

A bigger footprint

Skimlinks’ Navarro pointed to some other data co-op models, such as the Co-operative Audience Exchange Platform (CoEx) from the Boston-based OwnerIQ. It is a self-service audience data exchange between hundreds of retailers and brand marketers.

Through CoEx, a brand employs user data about visits to retailers’ sites for retargeting online shoppers with offers about their products when they go elsewhere on the Web, while a retailer uses the info to target customers who have shown an interest in, say, Sunbeam toasters when they visit that retailer’s site. Navarro pointed out that this arrangement offers no information on the paths that consumers may have taken in their shopping quests, as Skimlinks does.

Programmatic marketing firm MediaMath’s Helix business unit, launched in January, includes a co-op of about 300 companies, of which more than 30 are among the 100 largest American retailers. The co-op members share their online transactional user data, overlaid with behavioral, browsing and other data, and in return they get access to the pool.

To prevent “poaching” of each others’ customers, the co-op requires that the anonymized 500 million profiles in the pool only be used if they are tracked in more locations than just one company’s. In other words, brands can only target customers who have multi-retailer buying patterns.

For its Measurement Service, Quantcast claims that more than 100 million web destinations worldwide are sharing their visitor data in order to build out the online interest profiles of visitors and to help target advertising.

Skimlinks’ Navarro noted that Quancast doesn’t pay its members for the data and mentioned that another model — the Pangaea Alliance — works as a kind of cooperative union for publishers who group together their data to protect floor pricing and to constitute a bigger negotiator partner. It builds on previous publisher co-op models, largely in Europe.

Data pooling in some form or the other has been around since the early days of digital marketing, including direct marketers’ co-op database choices like Abacus.

But now, brands and retailers not named Amazon, Google or Facebook have a new urgency: jointly creating a bigger footprint, so they can avoid getting crushed in the evolving digital jungle.