There’s a huge buzz around media measurement and marketing effectiveness in the industry right now, and for good reason. The phase-out of the third-party cookie presents advertisers with attribution challenges that not only make it more difficult to make strategic decisions but also to achieve stakeholder buy-in for marketing investment.
To help you improve your knowledge of the modern measurement landscape and understand the best solutions to embrace user privacy without hindering marketing performance, Impression has launched a series of short videos.
Watch the full series here and gain insights into:
- The current technology and privacy headwinds challenging digital attribution
- How to respect user privacy online and implement compliant solutions
- How to more accurately measure the impact of your efforts through different solutions e.g. multi-touch attribution, incrementality testing and media mix modelling
- How to communicate the value of marketing measurement to key stakeholders
In part three, Technology Director, Aaron Dicks and Head of Paid Media Performance, Nick Handley introduce the trio of attribution and contribution: multi-touch attribution, incrementally testing and media-mix modelling.
See the transcript of the video below the recording.
Nick Handley: So we’re seeing a lot within the media landscape, and when we’re speaking to clients, things are changing. I think probably one of the big ones we are seeing is a lot of change to how we measure media. I think it would be actually quite nice to tell people what is our approach?
Aaron Dicks: Yeah, well, you’re right, there’s a lot going on. As we see it, there’s a trio of attribution and contribution work that can be done and the outputs of all of those are then insights really, that can be used in the strategy. The first of that trio is enriching your multi-touch attribution with privacy-safe, privacy-compliant information – that can come from a few different sources, particularly using the consent management platform properly, certainly something that people have got away with for a long time, but 2024 is going to ensure that that’s used properly. Enriching data, perhaps through server-side tagging or offline imports and using the enhanced match as well, where we can fill some of those attribution gaps between the standard click-based attribution we’re all used to but for users or perhaps transactions where they’re not tracked in that same way, later closing that conversion loop.
It makes our multi-touch, data-driven attribution a lot more effective and efficient. That’s the first part and day-to-day platform activation teams are going to benefit from that quite a lot. The second piece of that puzzle really is incrementality testing. What we do here, is measure the causal lift that a piece of activity has which allows us to circumnavigate digital attribution where perhaps you see two or three platforms getting credit for that sale. We can pull back the cover on that and understand what cause and effect has taken place so we can credit channels more appropriately.
The third piece in that puzzle is Media Mix Modeling. So with Media Mix Modeling, we take into account all of the inputs and all of the outputs, which are usually revenue and use statistical modelling, to tell the story of what happened in between. The benefit here is that we essentially don’t use any digital attribution whatsoever, and we let the data with some inputs and some modelling choices along the way, tell that story. That allows us to capture the long-term effects of different types of advertising. We can consider different contextual externalities as well, so the implications of pricing, competitor activity, and even holidays in the country where the modelling is taking place. Together, this gives us a really well-rounded view, a set of insights that can be used tactically on a daily basis.
We can still use multi-touch attribution for comparing the efficiency of one campaign versus another, they exist in the same sort of sphere. But then we can also use incrementality testing to reinforce our modelling process and off the back of our modelling process, we can use the outputs and the uncertainties that are produced for us to create new incrementality test as well so that it becomes a self-reinforcing, constantly improving model that explains what has happened. That’s really key as well, just to mention, is that this provides such an array of insight, but it does look backwards.
Nick Handley: So how do we make this a bit more tangible for people?
Aaron Dicks: Each of these pieces of activity are going to give us a different level of insight, some of those are going to carry forward into other pieces of activity. Some of the outputs from all of this work is that we can more accurately monitor and apply bids, basically to the lifetime
value of customers. That’s really important on the digital attribution side of things, because we want to find more customers who look like those who have a lifetime value, a higher purchase rate.
The incrementality tests are going to prove or disprove hypotheses we may have around activating new channels, activating different campaigns, or trialing different levels of spend, be that right down to zero or maybe increasing spend on a particular activity.
We’ll be able to understand some of the saturation that we’re getting from some of those channels and be able to look at that point estimate of the ROAS of that activity. The outputs from the Media Mix Modeling I think are perhaps some of the most exciting, certainly some mid-market companies may not have discovered this until perhaps in the last 12 months or so. Some of those key outputs, which I think are really interesting for teams, would relate to de-duplicating ROAS or being able to look at different media mixes. I guess that’s covered quite a few of the outputs and the insights that this sort of work can generate but iin terms of how you’re seeing this within activation teams and speaking with clients, how are some of these insights helping you?
Nick Handley: We’re seeing a lot of clients calling out for more understanding of where their conversions are coming from. We’re finding that people care a lot more about both their brand and performance activity and being able to tie it back to a single point of truth or one that is less ambiguous in nature.
We’re finding that clients want de-duped data and they want to understand where the conversion really does come from. When we are implementing things like MMM, we are seeing clients grow and have a better understanding of their media mix.
For us, it’s really important that we’re able to allocate a budget a lot more effectively. Seeing what channels are contributing is not just great for us, but great for the clients themselves – brands are able to accurately measure and put their budget where they can grow.
Lastly, probably one we get asked quite a lot is, ‘why can’t we just use an out the box solution?’ Ultimately, you can, but it wouldn’t be something that we would recommend. A lot of out-the-box solutions use AI to fill in blanks and that doesn’t give us much more meaningful data or insight into how we should look at budgets. It’s essentially creating a false artificial variant that we don’t really want to follow. Using our own MMMs allows us to manipulate that data, but also be able to give client output in a lot better and easier-to-read way. It also just helps us work closely with clients with a story that we tell which helps them educate internal stakeholders.
Aaron Dicks: That’s how, I think you touch on there, essentially why we do bespoke Media Mix Models. It’s because the transparent nature of that is that you can’t hide from the data. Any prior information that’s used to actually shape them, comes from tests. And you know, fundamentally those inputs and outputs are then entirely transparent. A black box solution may or may not give you the story but through a bespoke solution, we can take into account all of the externalities that might be affecting that business, that other solutions may not be nuanced enough to handle, and we do it in that transparent way as well. I think it really helps, as you say, with the storytelling as well. Ultimately, these things are just models. They’re a good explanation of what’s happened or might have happened, but they need to be taken as insight and the strategy needs to be built around them as well.
