IPG Reorganizes Trading Desk Operations as Company Tries Automation

Change comes as clients across industry start to question model

Is it time to blow up the agency trading desk model? Maybe not. But changes are afoot as the industry grapples with the long-term prospects of owning and maintaining separate stand-alone businesses designed to deliver audience-based buying.

Following a recent leadership shake-up within its Mediabrands unit, Interpublic executives were said to be mulling over the future of Cadreon, the company's trading desk business. The discussions focused on whether the agency holding company should decentralize its trading desk practices as programmatic ad buying grows in importance, or continue to house them in a stand-alone entity with its own profit and loss structure.

As it turns out, a centralization strategy seems to have won out at IPG Mediabrands, at least in the U.S. The company is expected to announce the formation of Magna North America, a new entity that will house the holding company's various data and platform assets. The move is being driven by an overarching goal at IPG to move at least 50 percent of its buying to automated platforms over the next several years.

Kristi Argyilan, most recently chief transformation officer at UM, has been named president of Magna North America. She'll report directly to Magna's worldwide CEO, Tim Spengler.

Included within Magna North America are IPG's finance and technology divisions, the search agency Reprise, the tech platform AMP and Cadreon, which was rumored to be on its death bed. According to sources, some factions within IPG were looking to pull audience-based, automated buying back within individual agencies rather than within a centralized group.

But according to Matt Seiler, CEO of Mediabrands, centralization should reduce competition within the various IPG factions while providing more clarity.

"With Cadreon as a separate business with its own revenue targets and its own P&L, and with individual agencies saying, 'We can do it here,' we had an internal conflict. We had built artificial boundaries that may have limited its growth. This says, 'We only do it in one place. Through Cadreon within Magna.' It's kind of dumb to have your operating units competing with each other, so we're eliminating barriers," said Seiler.

While IPG's maneuvers are specific to that company's unique issues and experiences, the same sorts of conversations are being held across the industry as dollars swell, agency territorial battles flare up and savvier clients continue to ask hard questions. The agency trading desk is at a crossroads.

Here’s the essence of the debate: Should all of the company’s audience-based, real-time media buying be conducted through a centralized group, or should buyers at Initiative, UM and other IPG shops be able to handle that sort of thing going forward?

Underneath the debate may be an acknowledgement that the party may soon be over (or is at least running out of beer). Trading desks like Cadreon, VivaKi’s Audience On Demand, Omnicom’s Accuen, WPP’s Media Innovation Group and GroupM’s Xaxis were established to help agencies take back the hefty margins they were ceding to ad networks in the mid-2000s, while also looking to move the industry forward. Profit-starved agencies have long seen trading desks as a major shot in the arm.

But as more dollars are spent programmatically, more clients are starting to ask questions. And individual agencies are pushing to take more control of their own digital ad buying.

Plus, many insiders report that despite the high profit margins generated by trading desks (think 40 percent to 50 percent), not enough clients were using them. The thinking inside Mediabrands and other agencies is that individual digital ad teams may be more incentivized to spend client dollars in this fashion if they’re given more choice and control. The hope is that the cost of the technology employed by trading desks could be spread across agencies.

“Most agencies want the great profit margins,” said an insider. “But they’re afraid that clients will call BS. The money essentially goes over the wall, and they don’t know what happens to it.”

Clients are starting to ask about what happens over that wall. One tech vendor said this has started impacting pitches and reviews. He described a recent conference call during which a client grew exasperated with its agency, which was unable to provide even basic details about where its ads were being run—since they were being purchased via an agency trading desk.

For example, according to sources, Kimberly-Clark has insisted that its digital agency of record, Mindshare, handle all of its audience buying rather than Xaxis. AT&T has made the same request of its GroupM shop MEC. Bob Arnold, Kellogg's global digital strategy director, recently questioned the model at a Digiday conference.

For its part, Procter & Gamble has shifted such buying to Audience Science over its agency partners' own tech. And according to sources, Ford, Citibank and Unilever have also opted out of their agencies' trading desks.

“We are starting to hear from more and more clients that their ad spending is totally opaque because it's centralized via a trading desk,” said Emily Riley, vp product and marketing at Audience Science. “That’s not kosher for most clients. Big CPG brands are just not going to stand for it.”

Especially when the dollars get serious. Early on, trading desk budgets were mostly nabbing experimental budgets. Now, they can account for 10 percent of a brand's budget. At MediaBrands, the goal is to shift programmatic buying to 20 percent or 30 percent. It's going to be hard for many clients to stomach that much money flowing through trading desks, especially if they think they're paying some sort of markup.

“It’s straight up arbitrage in their faces,” said one former trading desk exec.

Naturally, many top agency executives disagree with that assessment and remain bullish on the model. “Our strategy for trading desks is clear,” said Rob Norman, CEO of GroupM. "Xaxis is our audience-buying company. It works on a business model that enables it to access and apply all available client and other data securely to all available inventory, including inventory not available in ad exchanges. We think that's important. We believe performance against benchmarks should be the sole judge of the success or failure of Xaxis."

In fact, if any trading desk is succeeding, insiders say it's Xaxis. Some estimate that the company is generating $300 million to $400 million a year—with a healthy cut sent to GroupM. On the flip side, Cadreon struggled to tally close to $30 million, sources said.

Seiler balked at that assessment. "Let me dispel the myth on Cadreon not doing as well as the other guys," he said. "Cadreon is not about arbitrage. That's different than other places. There is concern out there in the marketplace with companies like Xaxis because there is arbitrage going on, a 'clients served second' kind of thing. A lot of agency groups say they are structured around their clients' needs. That's not true. Kinda sorta isn’t enough. You have to be transparent, or you're not."

As for the Kimberly-Clark decision regarding Xaxis, Norman said, “Some clients prefer that we work with third parties they have selected, and we train our people to work with them in the most effective way for the advertiser. In the case of Kimberly-Clark, that partnership is operated inside Mindshare.”

Regardless, ad tech vendors are starting to build their products to work more nimbly in a future where they have more clients, not just a select few trading desk partners. “Brands aren’t dumb,” said one vendor. “There hasn’t been enough spend to matter. But with this shift to programmatic premium, the money gets real. Clients are going to expect you to act like an agent, not an ad network.”

The bigger question is, has the trading desk model ever worked for agencies? Proponents typically brag about the hundreds or thousands of campaigns they have for clients via their trading desks to date. But one former agency insider threw cold water on those claims.

According to some former agency executives, most global agencies have 500 to 700 clients. So even if a few dozen clients are using a trading desk platform, each running multiple campaigns, that’s a drop in the bucket. "It’s actually been quite hard to get more clients on board,” the insider said.

Why? Trading desks simply don’t pay off for certain clients, who either rely on low-cost cost per acquisition tactics or are very concerned about site environment.

And even for clients that do test trading desks, they’re not always immediately effective. “It takes a long time for an algorithm to learn and an enormous amount of volume needs to be captured," said an insider. "It may take eight to nine months and that’s hard for clients to stomach."

There’s also the cost of technology—never an agency sweet spot. One exec estimated that trading desks can cost $7 million to $10 million just to develop, and constantly need to be maintained. “It’s a sinkhole,” she said.

That cost is a great argument for stand-alone agencies to take over audience buying. Another is simple long-term relevancy. “To me, if agencies don’t start owning the process, they’ll start sending brand agencies to the nursing home,” said one agency exec.

Of course, many dismiss this way of thinking. Operating trading desks requires experts. "They are here to stay," said PubMatic co-founder, CEO Rajeev Goel. "Overall, the model works. And it takes a real expertise and skill."

"There's too much money to be made," said another insider. "Agencies aren't going to give up yet."