In this exclusive Q&A, Jorge Poyatos and Albert Nieto, co-CEOs and co-founders of Seedtag, discuss their recent acquisition of KMTX and how AI is being used to address challenges within contextual advertising.

What were the key drivers behind the decision to acquire KMTX? How will KMTX’s solution be integrated with Seedtag?

Over the past eight years, we have been building a privacy-first advertising solution, pioneering the use of AI and machine learning to create the best contextual product in the market. Now, we feel it’s time to evolve even further. When we had the opportunity to analyse KMTX’s technology, we knew they were the right target to help us do that.

The “death” of third-party cookies is a golden opportunity for our company, and the acquisition of KMTX puts us in the best possible position to take advantage of it. We were somewhat lacking low funnel capabilities, but with KMTX’s predictive models combined with Seedtag contextual data, we are able to offer brands that long awaited full-funnel contextual solution. This positions us as a one-stop shop for contextual advertising, reducing complexity while allowing clients to reach their campaign objectives.

What are the current challenges within contextual that can be addressed with machine learning and AI? How will AI solutions develop further to address these?

Jorge Poyatos (r) and Albert Nieto (l), Seedtag

The industry has traditionally adopted quite a blunt approach to contextual targeting, with strategies based almost exclusively on keywords or the domains people visit. However, in the last few years, AI has allowed advertisers to create a more evolved strategy, one based on the AI’s capabilities to read and understand text and in our case also images the same way you and I could. This allows us to target users' interest with a level of accuracy and brand safety that was unthinkable just a few years ago.

However, one of the biggest challenges advertisers are facing is understanding how to find their audiences through contextual targeting. For many years they have been analysing the demographic profile of their customers but they have had no clue about their contextual profile. Contextual AI can help them overcome this challenge by providing and analysing consumer interest data for them. For example we allow brands to see in real time what users are reading in the UK and then transform this data into client-specific targeting strategies.

These new capabilities are the result of innovating the uses of AI for contextual. Until now, contextual AI has mainly been based on supervised learning. This means the machine has been trained using data which is well “labelled”, it is to say some data is already tagged as being the “correct answer”. Our AI also used this approach for page level analysis (PLA). Every time a URL is fed into it, the technology is able to place it into a number of categories based on prior training.

However, we have pioneered the use of unsupervised learning for contextual advertising. Unsupervised learning means the machine is left to work on its own to discover patterns in the data. Through unsupervised learning, our AI allows us to group articles based on similarity and semantic closeness. All of this means we can not only perform page level analysis (PLA) but also incorporate the intelligence of the network level analysis (NLA) in order to offer brands recommendations regarding what content they should be targeting based on how close it is to their audience, similar to how Netflix recommends what movie you should watch next.

How can full-funnel contextual solutions benefit advertisers from a performance perspective? Similarly, how can contextual benefit marketers on branding objectives?

Over the last few years, we have seen a strong correlation between contextual signals and performance results, although we have not had the technology to predict post-click behaviours at scale. The acquisition of KMTX brings AI based predictive models into our stack that combined with our proprietary contextual data will constitute a leading solution for achieving performance results in a cookieless world.

As strategies based on third-party cookies die off, new solutions are cropping up. For example, Unified IDs and first-party data are both great alternatives but lack scale and reach. Through our contextual performance technology we are now able to offer a solution that combines both scale and effectiveness across the open web.

How are contextual advertising results and KPIs currently quantified? Are new and/or evolved metrics required here?

KPIs depend on the objectives of the client. Contextual branding campaigns are typically measured using commonly-used industry metrics: CTR, VTR, viewability, etc. However, we have found that these are not enough when determining the attention quality of a contextual media investment. We usually expand that information by partnering with independent research companies such as Lumen or Metrixlab to measure the real impact of attention on brand perception.

Our performance solution focuses much more on post-click KPIs such as time spent on page, qualified visits, leads, etc. This means Seedtag is now able to optimise campaigns across all areas of the funnel.

How does the adoption of contextual solutions vary across EMEA, LATAM, and North America? What are the specific challenges and opportunities within each of these markets?

We have seen an increase in interest in adopting contextual advertising solutions across all markets but we definitely notice differences across regions, with Europe being a pioneer in this respect. Europe was one of the first regions to tackle issues of user privacy, forcing countries to take regulatory measures to guarantee the privacy of their citizens. France has been at the forefront of this and is probably the market that we operate in with the strictest regulations. As a European company, these changes have pushed us towards being leaders in contextual advertising, providing businesses in these markets with solutions that respect user privacy and abide with all regulations.

One of the challenges posed by being an international company is the linguistic differences between the various countries we operate in. We have set out to deal with this challenge since our inception, and our contextual AI is currently capable of analysing content in over eight languages, with more to be added in the future.