Data Fragmentation Is the Biggest Challenge: Q&A with Tijs van Santen, Impact Radius

data fragmentation

The amount of data in and around advertising is staggering. But when that data comes from a multitude of sources, and is fragmented within various systems, how can a marketer properly use that data to make the best decisions for their campaigns? Tijs van Santen (pictured below), CRO, Impact Radius, talks about the importance of clean data, goal alignment, and the challenges that martech companies face around attribution.

ExchangeWire: Who is Impact Radius and what problems do you solve for advertisers and agencies?

Tijs van Santen: We help advertisers verify, understand, and optimise the ROI of all digital media across their media mix, while enabling new performance partnership opportunities. Our platform helps marketers remove nonhuman traffic, verify viewability, protect their brands, and eliminate attribution fraud. We also offer a system of record solutions, unlocking consumer journey insights across all devices and marketing channels with advanced attribution models. Last but not least, a large number of global and domestic brands use our platform to track, manage, and optimise all their performance-based partnerships in one place, attribute credit properly, find new partners, obtain real-time reporting so they know what’s working or not, and track the incremental value of individual partners.

How should marketers deal with data fragmentation? What steps do they need to take?

Data fragmentation is a big challenge for marketers. Everyone has so much data and it’s often coming from a wide variety of sources captured in multiple spreadsheets and not easy to reconcile, view, share, or take action on.

We believe that a system of record which captures the majority of data in real time, while ingesting and deduplicating internal and external data sources to build a single and clean data lake is extremely effective. It prevents data from being siloed in different teams and departments, allowing everyone to understand how their channel, campaign, or ad participated in a single conversion. Marketers need transparency in order to optimise, hit growth and revenue targets, to know exactly where their marketing dollars are going, and gain complete visibility into how each channel is performing relative to the others. This is imperative. A system of record is also key to attribution modeling and understanding the customer journey at a granular level.

What are marketers doing wrong when it comes to performance marketing? What are they doing right?

We see an increasing number of performance marketers doing the right thing: they strive for transparency in affiliate contribution in the conversion journey. They’re removing the network conflict of interest by separating technology and services, aligning crediting rules and payouts to value the partners who delivered and, most importantly, looking beyond traditional affiliates to seize on the growing opportunity to partner with a wide variety of partners in the mobile, influencer, content, and B2B ecosystem. It’s exciting to see how some of our clients are significantly growing their investments in performance channels by leveraging new insights and their ability to partner in smart ways with a wide variety of non-traditional partners.

The clock is ticking for marketers who stick to a siloed view of the traditional affiliate model – a legacy model that simply looks at top-line revenue based on last-touch crediting within the affiliate channel with a heavy emphasis on cash-back, loyalty, and coupon sites.

How will machine learning and other AI solutions help tackle ad fraud and transparency issues?

Like all methodological advances, machine learning is a double-edged sword for ad fraud prevention. AI, specifically machine-learning algorithms, can help us cluster together types of user behaviour and browsing that data scientists can then label as fraud or not. However, the key isn’t the model, but the feature engineering, as if you aren’t able to come up with good measures of what fraudulent activities look like, then you have no shot of detecting invalid traffic detection (IVT).

Impact Radius

Tijs van Santen, CRO, Impact Radius

Poorly built algorithms will then be a waste of both time and effort. Bots are also using increasingly advanced techniques to appear human and evade detection. Still, by using a combination of a rules-based and machine learning-based approaches, it’s incredibly difficult to make botnet traffic look like normal human browsing, and we’re confident we can stay ahead. Forensiq has incorporated machine learning into its latest algorithm improvements, which helped us discover some unexpected ad fraud patterns in our clients’ inventory, including widespread and sophisticated ad injection targeting premium sports domains.

What challenges do martech companies face around attribution?

The way I see it is two-fold: 1) Every attribution model relies on the quality of the underlying data. Bad data in, bad data out. Virtually every time we speak to an advertiser who has ventured into attribution, the single biggest challenge to get any model to work in the first place has been creating and maintaining a clean dataset. 2) Organisational governance models often create conflicting incentives between media buying teams, making it much harder to optimise campaign spend and performance across channels and campaigns.

We believe the biggest opportunity lies in emphasising a clean dataset, reviewing team structures, KPIs, and the alignment of personal, departmental, and organisational goals. In the end, attribution is about determining true value, and optimising activities based on what’s driving the desired business outcomes.

What do you see for the future of martech? How do you see the landscape changing?

A few things: mobile will continue to grow in importance, transparency across devices and marketing channels, and ‘true performance’ will continue to become more important. We’ll also see the continued convergence between ad tech and martech driven by AI, blended with an increasing conflict between consumer privacy demands and personalised experiences.

Comments