The Rise of the Client Trading Desk & its Knock-On Effect

With all the Wall Street fantasy IPOs happening at the minute, it’s often easy to forget the fundamentals of our business. Display budget is very much controlled by the agencies — even more so in Europe and the APAC region. There is a small percentage of ‘hands-on’ marketers managing their own display budgets, but with the rise of centralised trading desks we are seeing a lot more marketers becoming increasingly vocal about transparency around data and margin.

We have had three-to-four years of the agency trading desk model. It’s been reasonably successful, eating up traditional ad network spend and becoming a permanent fixture on the media plan — but the model is about to pivot once again. Having been mostly marginalised from the execution of data-driven campaigns, big agencies (and creative agencies) within holding groups are trying to join planning and execution back together around programmatic.

The Rise of the Client Trading Desk

At the recent ATS London event, Andy Mihalop, Head of Biddable and Programmatic Media for MediaCom, spoke about the Client Trading Desk and why clients are looking at this model. However, before we dive into what this all means, a little bit of clarification is required for the term. Effectively, the CTD is about control.

The relationship with the buying solution is owned by the client. So, whether it’s AppNexus, MediaMath, Turn, DataXu or Google the licence is managed by the client, not the agency. In this particular model, the data also sits on the client side. Again, the marketer will manage the relationship with the data warehouse solution provider to manage and store all of the marketer’s data.

What does the agency do here then? Basically, it will be asked to execute on a transparent margin. Of course, it can layer on services like analytics and data, as well as specialist trading expertise; but ultimately the model is about transparency. The marketers like the concept and efficiencies around programmatic buying — they just don’t like current models.

The Media Plan Goes Away

While the media plan is not going anywhere soon, the reality is that if the Client Trading Desk model takes hold, it will mean that the plan is not required. If all programmatic buys are being executed by the agency, why do you need third-party buyers? It is quite amazing that recently IPO’d ad nets are being valued at twenty times gross run-rate revenue, considering what is coming down the line.

This has serious knock-on effects for the entire space. No media plan means the end of the middleman buyer; but only in DR, admittedly. The client is requesting its agency to plug in data and inventory sources, not outsource buys to vendors making 45% margin. That is the reality of this new model: that programmatic will be used to cut out unnecessary layers in the buying chain.

The Agency Trading Desk Pivot

While spend continues to grow at the ATDs, these new agency ad networks might find it difficult to keep expanding if clients decide to in-house some of the processes. So, what’s next for the likes of Xaxis, for instance? Unlike the rest of the ATDs, Xaxis is a different proposition. It works and functions like a traditional ad network. It very much depends on the plan to be allocated budget.

If Mindshare, MediaCom, et al. are forced by clients to build this CTD model, where does Xaxis sit? The solution is pretty simple. Xaxis, like all good networks, must chose a side — buy or sell side. Given that the buy side will be managed at the agency level, Xaxis has but one choice: become an SSP. Now, the margins might be a lot smaller, but the scale of spend and sustainability of the model would make it very attractive.

Xaxis is in the unique position of being able to leverage powerful Group M trading agreements. Imagine the possibilities around programmatic guaranteed if Xaxis flipped its model. There is no reason why MediaCom and Mindshare couldn’t buy from the Xaxis supply; and it wouldn’t be just those two alone. Many of the creative agencies are now getting into media buying, and there is likely to be a huge opportunity for the Xaxis SSP as that spend increases.

There is also the opportunity of external demand too. WPP has a ridiculous number of data sources within its organisation. If Xaxis could get access to this data and layer it on to inventory bought in, it could well allow external buyers access too. This could pose a significant threat to those sell-side solutions working with established premium publishers.

If successful, you could imagine other trading desks following a similar route.

Agencies Look More Like Ad Networks, and Ad Nets Look More Like Agencies

The ad network is the great survivor, but its traditional role has radically changed. Instead of executing third-party buys, the ad net now has to look more like a consultancy. It will have to think about becoming an enabler, offering value to agencies. Whether it is analytics or tech, its role in the ecosystem has changed forever. It is for this reason that massive scale will become a major issue, particularly if it is dependent on media spend for its survival. This is why these ad tech IPOs are beyond comprehension. Algorithms won’t matter if there is no plan. Going direct is a dangerous option too. Only one company in Europe successfully managed it, but then retargeting is both sticky and highly addictive. So, the ad net survives, but now looks more like a SaaS solution. This transition will be interesting.

The client is now taking control, and the market feels it is about to shift again. Let’s see who is prepared for the fallout.



  • Harry

    lets not forget that the client wants the solution that provides the best return. agency trading desks are notorious for not doing this and blocking the competition that would make the client investment work harder.

  • Gavin Deadman

    Fantastic article.

    When brands have access to choose whichever DSP they like, they’ll choose the DSP which by taking into account QPS, fee, minimums, source etc makes sense to the business. This automatically rules out all agency trading desks because like you say they are more like ad networks with hidden margins in many cases of at least 150%. Also as brands take the adserver / tech contracts in house, the data which trading desks can access will decrease dramatically as on site pixels get replaced.

    Most digital planners, execs and directors will move over to brand side leaving ad agencies to deal with traditional offline non-programmatic buying as well as offering brands a consultancy service to help with integrations etc. Also perhaps a training centre for grads / 1st jobbers where they charge brands a recruitment fee once the grad has moved in house.

    Sorrell’s empire will crumble over the coming years and the signs are starting to show as he now wants people to refer to his companies as ‘data company’ rather than ‘ad agency’.

    Many tech companies like Google and Facebook are bringing out fantastic transparent and self-service media tools priced very competitively for brands. These tech companies also understand a marketers objectives in many instances better than agencies where spending might come before client objectives.

    As TV becomes programmatic, it will be a simple task for tech companies such as Google to connect their DSP to ITV and C4 which as we know still has the majority of TV spend. This will result in Sky begging to eat at the Google table.

    Consumer brands are keen to offer their products at a cheaper price to gain more market share, so cutting marketing wastage is one way of being able to afford this without impacting acquisition volumes negatively.

  • human row

    Until clients start to realise programmatic buying is as sexy as those F1 sponsorship deals and push back to agencies again…

  • James

    No offence but this whole article is based on the premise that a very small number of very large clients would actually buck the trend. And that simply won’t happen. It is true that there are few clients mostly in France that have started building their own trading desks and pose some threat to possibly the agency model and maybe algorithmic buying. The problem however is that that threat is entirely theoretical. Maintaining a trading desk is actually very expensive and the overhead is a lot higher than most advertisers are let to believe. In fact, most agencies no longer believe that their trading desks are a good idea. The trading desk was necessary in the early days of programmatic buying but today the overhead in both headcount and fiscal cost is actually a burden to most. Plus, you don’t really need a trading desk in order to get clint data.
    So believe what you like, but whoever the author of this article might be, had a very skewed view of the advertising world. Oh and also, if you think that Europe is dominated by agencies you should see the US. Yet not a single advertiser there has built a trading desk. Maybe that should tell you something …

  • cpokane

    Now James, if you actually read the piece properly you would have understood that it wasn’t about in-housing of programmatic buying – but was in actual fact about the client taking control of vendor and data relationships. This model allows buys to be executed at the agency level. If you’d managed to get down to ATS you would have seen a senior agency exec discuss it at length, and why ut us significant for the space.

    As far as I am aware there are very few clients executing buys on their own on this side of the pond. And it will probably be like that for some while yet. The idea of the “client trading desk” is about control over tech and transparency – not disintermediation of the agency layer.

    And a “skewed view of the advertising world”? Interesting to hear that from an agency bod (I am making the assumption here) that still can’t come to terms with the fact that the agency trading desk is effectively an ad network. The agency ad network (aka ATD) really is a tough model to make money out of especially when it’s almost next to impossible to get kicked off any plan. With the margins being so tight for the ATD (a shockingly low, 40% – 90% margin range), it really is ridiculously difficult to make any revenue from this expensive model. Truly, these are desperate times.

  • Gavin Deadman

    Hi James,

    The in house trading desk doesn’t have to be huge in order to do a good job and it can be done extremely efficiently. Agency heads think it’s quite expensive and will no doubt tell clients this because they’ve spent huge amounts on infrastructure rather than using third parties like Google / MediaMath. A significant amount of brands have been running SEO, paid search, Facebook marketplace and affiliates in house for years so it’s only pretty much a case of adding in display.

    Also most DSPs offer managed service for those who struggle with resource and charge an extra fee in a transparent way.

    In terms of cost, when you weigh up resource cost with contract, tech and media savings, the savings will outweigh resource cost in many cases from £300k to +£1m / PA.

    I can speak from experience and from one contract alone (adserving) I saved the brand over £40k a month just from negotiating the contract as the agency was adding in their rather large margin.

    When you take into account other hidden margins in buying platforms, kickbacks etc, there are plenty of savings one could make. How can WPP afford such a large empire and still make huge profits???

    If you look at large brands such as O2, Sky etc they’re wasting at least £800k a year so imagine how many agency people they could employ to sit in house with that money.

    Agencies will always be around especially for all non-programmatic activity such as offline but like the article says, they will be more in the form of consultancy rather than taking the clients money and spending it. Even Sorrell is admitting that they’re no longer an ad agency and more a data company.

    Many large agencies are actually piloting the vision which Mihalop has now with certain clients.

    Like Ciaran says in his reply, the programmatic buying transition won’t happen over night and it’ll take years as clients come to the end of their long agency contract.

  • Pingback: As Consolidation of Spend Comes at Agency Level — How Do You Play in the ‘Programmatic’ World? | siliconewsblog()

  • Pingback: Как клиентские Trading Desk (CTD) запускают цепную реакцию | RTBlog()