Embracing the Outcome Model to Deliver Measurable Value: Q&A with Nicolas Bidon, Xaxis

Trading programmatic media on outcomes is still a relatively new concept for advertisers; but as digital advertising starts to evolve beyond just the 'business of media' into the 'business of everything', the outcomes-based model is gaining traction. ExchangeWire speaks exclusively with Nicolas Bidon (pictured below), global CEO, Xaxis, about what trading based on outcomes means for the programmatic advertising of the future and why, despite its complexity, the model opens up a wealth of new opportunities for the advertiser, the agency trading desk, and the agency.

ExchangeWire: How has the outcomes-based model gained so much traction recently?

Nicolas Bidon: Trust has come into question a lot recently. From Xaxis’ perspective, we felt one way to address this is to evolve our model to deliver more measurable value to advertisers. The best way to do this is to understand the advertisers’ media outcomes that tie as closely as possible to the real business outcomes they are trying to drive. Advertisers are always trying to initiate some type of outcome: selling more units, improving awareness, or increasing brand recognition.

We asked ourselves how we can use all the assets we have. We are a pioneer in the programmatic space – we have experienced people, lots of data, and in the last two years we have spent a lot of money developing our AI platform, Co-Pilot. We wanted to embrace the outcome model to define with advertisers what the metrics/KPIs are that would give them confidence that we, as one of their media vendors, are delivering true and measurable value to them.

What does this look like in practice?

Well, just one example would be a recent project we ran with one of our big automotive clients. We entered into a dialogue with the agency and the client, and discovered they wanted to deliver an optimised Cost Per PI (purchase intent). This is an internal metric; and essentially they are defining it as a weighted average of different actions people take on their digital channels, for example, downloading a brochure, finding a dealer, or booking a test drive.

Based on their own econometric studies, and business realities, they determined which behaviours are the best predictor of someone walking into a dealership and, essentially, buying a car. They were trying to reverse engineer what we were delivering based on standard metrics, like CPV or CPM, to ultimately back it out to a cost per PI. We thought, why don’t we do it the other way around, building a custom algorithm using our AI platform, to drive the most ‘PI points’ for the cheapest price. Their response was: “You can do that?”

There’s a dialogue you think should be happening, but they didn’t know what we could do, and we didn’t know what they wanted. They knew from their first-party data the cookies that track the best – i.e. the people who did go into the dealership. This data was provided to our 30-strong data science team, based in New York, for them to calculate the variables – e.g. media, placement, time of day, cookies, targeting segments – that were the predictors for the maximum number of PI points achievable. Then, build that into an algorithm that will take all these variables and optimise them, so that every time we see a bid request coming in from the DSP we can score the likelihood of driving a specific PI point and modify the bidding automatically, based on that predictor.

This is at the leading edge of what people can do with programmatic. However, to put it into perspective, the whole conversation with the client, and the work that followed, took four months. That kind of customisation can’t happen for every advertiser – it’s not a turnkey solution – but a version of it probably can; and that whole conversation certainly should. For this client, the bidding algorithm has been live for a couple of months and it is producing great results. We have now started doing this with five advertisers globally and we are learning a lot from it. One key learning is: how do you change the dialogue with the advertiser? It’s a very different approach, more like consultative selling. We are trying to understand what you want, how we will measure it, and what data you have to help us to customise the algorithm.

This is a complete shift in dialogue between the agency trading desks and the advertisers – why wasn’t it happening before? Are advertisers keeping their third-party partners at arm’s length?

Nicolas Bidon, Global CEO, Xaxis

It’s not that the advertiser is trying to hold information from us. First off, the advertisers don’t know that there could be an alternative. From our side, we and our agency people aren’t necessarily trained or understanding enough of the real and the possible. We are too quick to go deep into the details, without peeling back, looking at the broader picture and asking: “Why? What are you trying to drive?” We aren’t necessarily asking the right questions. It takes time and trust, and needs both parties to want to invest in the relationship to get a better result.

How does the outcomes model impact budgeting and forecasting? Are you unlocking unlimited budgets, provided you are achieving the agreed KPI?

That’s the vision; that’s certainly where we want to go. Performance is the easiest thing to sell, but the hard part is delivering it. We need to understand the right benchmark. Today, the advertisers often don’t even know themselves, as they are reverse engineering it. We need to see what we feel comfortable with, from a risk perspective, as it is about taking a risk. We are learning and moving in that direction – but fast and, as far as we can tell, much quicker than most.

What are the risks to you?

We take a position in media. We do pre-buy inventory. We know from analytics, and our own experience, that we are good at driving complex outcomes. When we know what those placements are, we do try to enter into private marketplaces, or an overall upfront commitment with the publisher, pre-buying a certain number of impressions for a particular placement.

A bigger risk is that you never know what’s happening in the market. You may commit to a specific outcome and a specific volume, and agree to spend 20% of the advertiser’s budget on this cost on a specific week. During that same week, one of your client’s biggest competitors might have a massive promotion, which could completely influence your own performance. We can say we’re confident we will deliver to plan, based on our own historical predictions, but then we’re negatively impacted by this promotion. That’s where the dialogue is key – you need to become business partners.

The agency trading desk is shouldering a lot of risk with the outcomes model – is it equally risky for the advertiser?

For them it is less risky, as they constantly benchmark us. Sometimes people forget that agency trading desks never win 100% of the plan. We always compete with other channels or other competitors and, frankly, that is right, as it comes back to trust that we are delivering better outcomes. If we say we will take some risk, the advertisers have to ask themselves if they know what success looks like for them. Once we agree to trading on the outcome metric, the risk is all ours.

For the advertiser, the only risk is if we say we can deliver 20% of their budget on that particular cost per outcome, and we don’t deliver, then there is an element of risk on their side, as this campaign would potentially under deliver.

Is this dialogue being led by the media agency or the agency trading desk?

It’s a partnership. In my first role at Xaxis, as UK MD, of my top 10 clients, I probably met one. For all others, I was kept at arm’s length. Programmatic didn’t register highly on the radar. There’s also an element of fear and being protective from the media agency. From the planning perspective, we were the traders, and nothing more. Fast forward to today, of Xaxis’ top ten global clients, I, or somebody else in the team, have met with all of them. When we do, it’s always with the media agency. We have frequent QBRs (quarterly business reviews), and the reason is that programmatic and digital are becoming more important. As people understand the realm of the possible, they want to engage.

In addition, the advertiser is getting more of an understanding of what’s happening in the space and wants to talk to the expert. If a client tells its agency that it wants to understand more about Snapchat, the agency would be expected to bring Snapchat into the room. When it comes to programmatic and the outcomes models, we are the experts, so why wouldn’t we be in the room?

We need to be having these conversations jointly with the agency. We exist in our niche area of expertise, but we think it’s massive. Programmatic has grown tremendously, but it is still a portion of a portion. Most advertisers’ digital spend is between 10-30% of their total ad spend, and programmatic is a smaller share of that budget. In ad tech, we get a bit full of ourselves and think we should sit down with the CMO every other week, because what we’re doing is the future, but the reality is, if you’re the CMO of a big CPG and you screw up your next TV campaign, that probably has a more damaging effect than whether or not your programmatic strategy is right.

It the role of the media agency shifting to make it more of an integrator?

Yes. I see a world where the agency will continue to be the integrator of all the different specialists, who will each continue to bring their own specialties to the bigger picture. This is why all the talk in the market about consultancies coming into media is interesting. They’re trying to be more like media or creative agencies, but to succeed long-term, agencies will also need to be more like consultants. In the IT world, consultants are systems integrators. I see the agency being exactly the same for all things media. There is great potential to evolve skills and make sure they run the table. The middleware, the integral part of anything – it’s not the most glamorous part, but it is often what drives value.