Thanks to artificial intelligence, Sonata is capable of discarding all data of fraudulent origin

Álvaro Mayol, CTO of Taptap
Read in Spanish →
Read in English →


Although programmatics has been the clear bet in the industry in recent years, the latest controversies about traffic fraud and bad practices of some agencies, along with the impossibility of agreeing on a common measurement system in the digital world, have deflated the bubble somewhat. Nevertheless, the future is inevitably technological, and the purchase of inventory should automatically be one of the priorities in any company’s media strategy. Having a system that offers a guarantee on the verification of data, offers a qualified audience and the possibility of finding the consumer along the funnel, it’s essential to increase advertising effectiveness.  Platforms such as Sonata, developed by TAPTAP Networks, have already achieved an advantage in the market thanks to some particularities that we wanted to find out about from the hand of Álvaro Mayol, Partner & Chief Technology Officer of TAPTAP Networks.

1. What is Sonata?

Sonata is a DSP (Demand Side Platform) and DMP (Data Management Platform) of a global scale, and multiplatform (mobile and desktop) with a high focus and specialization in mobile-centric activations. We’ve specialized in audience profiling and cross-channel interaction at both physical and digital levels. It’s a platform and technology 100% designed and implemented by Taptap.

2. What benefits does this tool offer marketers?

Sonata allows you to design strategies based on the analysis and application of data of a different nature (Data-Driven Branding) to be able to apply them in a precise and different way, according to the point of the funnel the consumer is in. It also allows profiling and activating users in a secure and transparent way, in a contextual and relevant manner, regardless of the digital or physical channel they’re interacting with. It becomes a cross-media and cross-device tool that takes advantage of the great focus on mobiles, to learn and get to know the users, thanks to the digital interactions and the physical places where they’re located. Additionally, thanks to the physical attribution engine, we can measure the effectiveness of digital campaigns in terms of visits to the point of sale they refer to, adding new models of physical attribution to those already known digitally.

3. Where is the differentiation of this platform with respect to the competition?

Sonata was born with a very clear and differentiated approach for the mobile. This has led to a very specific specialization with respect to the differential value that mobile devices can offer us in terms of audience profiling, activation and attribution (especially at the physical level related to store visits). Sonata allows you to connect the physical and digital world, not only to learn and better profile users, but also to activate in a precise and contextual way depending on the user's environment and interaction with any other offline medium. This has allowed us to adapt our platform and specialize the technology to adapt to different channels such as retail, auto, travel, etc., offering different capabilities to maximize the effectiveness of the results according to the customer or the product in question.

Sonata works actively on the rigorous verification of its inventory, the evaluation "pre-bid" of each of the impressions and, of course, the subsequent analysis of the campaign through the design and implementation of proprietary technologies. Therefore, to offer the best possible result, our platform applies machine learning techniques, analysis and implementation of predictive models and carries out human verification processes, in addition to having additional third-party certification. As owners of our technology, we have greater control and knowledge of the measurement systems and all the applied engineering to carry out the best implementation of these measurement and verification methodologies. All this allows us to audit the sources of inventory, and thus offer mobile advertising of high quality and guarantee for our customers.

4. One of the biggest concerns for professionals is the lack of transparency in the purchase of media. How do you, from TAPTAP, solve this problem and ensure the supply of qualified audiences to your customers?

A key variable of great value, both in carrying out the segmentation, activation and attribution of the audience at digital and offline level, and implementing mobile-centric strategies, is the location. For this, Sonata has an audit and classification system for the location signals available in the advertising ecosystem called LQI (Location Quality Index). Thanks to artificial intelligence and the analysis of billions of pieces of data daily, and on a global scale, Sonata’s expert system is able to discard all data of fraudulent origin or low quality, taking into account not only the location of the user, but also dozens of additional variables, such as advertising interactions (clicks, videos, ...), viewability, creative placement, origin and quality of the medium, etc. This entire data ecosystem allows us to know the moment in which the user is in and generate a behavioral profile, both digitally and physically (offline).  Additionally, I reiterate that Sonata works actively on the rigorous verification of its inventory, the "pre-bid" evaluation of each of the impressions and, of course, the subsequent campaign analysis through the design and implementation of proprietary technologies.

Therefore, to offer the best possible result, our platform applies machine learning techniques, analysis and implementation of predictive models and carries out human verification processes, in addition to having additional third-party certification. In the case of brand safety or viewability, we have IAS as an additional certifier of our analysis and quality assurance processes.  Our reporting can break down the campaign data and identify which are the segmentation combinations that maximize the quality of the impacts and advertising interactions, at the level of the medium, creativity, operating system, etc.  Regarding the qualification of the audience, we mix audience deterministic data provided by certified partners with our intelligent profiling processes based on the location, history of digital navigation and interaction with the different brand typologies. In this way, we build profiles with great value and specialization for each specific strategy.

5. The huge amount of data available today can help marketers lose focus. How do you help them convert that mass of data into valuable information?

It must be understood that for each phase of the consumer funnel, different data sources with different specializations are working together. As we move forward in the funnel, the data is treated according to the affinity or the profile of the consumer. On the other hand, we believe in specializing our data processes according to the vertical or the typology of the consumer journey. Buying a car isn’t the same as buying tickets to the cinema. Understanding the context and how to contribute value is key. That’s why we profile the audience based on four highly relevant variables:    

  1. The location and the correlation with points of interest. We call this geo-profile.
  3. Digital navigation according to the category of the medium. This better defines the user's interest.
  5. The interaction with the advertising stimuli (clicks, videos completed, conversions, etc.) of the different campaigns depending on the vertical. This allows us to better know the affinity to the brand or vertical.
  7. The consumer profile based on visits to the point of sale. This information best describes the user at the consumer level. These profiling processes are mixed with external data sources of a deterministic nature, mainly at a socio-demographic and behavioral level, where the data has a non-statistical certainty.

In Sonata (DSP & DMP), thanks to our own technology, we’ve managed to evolve our processes of analysis and data processing, in order to build audience profiles and pre-bid systems with a greater degree of specialization, depending on the specific moment of the purchase process each user is in. We start with unsupervised models of machine learning (clustering) to later refine algorithms based on supervised machine learning models, where we include samples of real data, as well as real advertising interactions for each vertical or category and, of course, all the research and knowledge of the user by the brand. Of course it’s important to know how to incorporate and weigh other external data sources, such as the DMPs of the brands or agencies, or data providers of another nature (telco, social media, consumption/affinity...).

6. What are the biggest challenges that marketers face when executing a mobile-centric cross-media strategy?

The first thing is that the mobile gives us a distinguishing factor from the rest of the media, which is the proximity to the user and the ability to offer us their location. Knowing how to qualify and audit the user's location data is key in being able to subsequently build audience profiles, apply segmentations or even make more complex attribution models of foot-traffic or store-traffic. Having our own DMP that allows us to build and efficiently combine different audience taxonomies allows us to combine this aspect with highly qualified localization, thanks to our LQI.  This allows us to know very well, where and when the audience moves based on the socio-demographic or behavioral profile. Once we have this information, the creative adaptation at a dynamic level and in real time allows us to serve display, video or audio, depending on the offline medium we’re dealing with. We’ve already interacted efficiently with the external environment, not only extending the audience with impacts on their mobile when the users are close to the OOH point, but dynamically adapting the ads in DOOH circuits thanks to the knowledge of the audience and their profile in each exterior location. This is an example of what we’re doing at the level of mobile interaction with other media, and on distinct platforms.

7. Although the adoption of the smartphone is, at this point, massive on the part of the users, it seems that as far as advertising investment is concerned, the marketers are still a little behind. What do you think is slowing investment in this field?

I think the main problem lies in designing valid attribution models for all channels. We know that mobile devices represent a crucial point of contact (touch-point) in almost every step of the path to purchase (customer path-to-purchase journey).  Starting with the trigger or need that initiates this process, until reaching the final decision or purchase, mobile devices have a fundamental role in influence, interaction, momentum and relevance that we can’t stop addressing and analyzing. It’s necessary to have a good understanding the moment the user is in to adapt the content or the advertising activation in the best way. We continue working to make the mobile phone not seen as an independent media, but rather as an extension of the audience of any other non-digital media such as TV, OOH or radio. I think that, in this area, the mobile is going to have a very important role and keep growing, since it’ll be part of the offline strategies that have to be digitized.

8. Where do you think the trends in mobile advertising are heading?

The interaction of users with mobile devices is growing exponentially, both at the frequency and session levels. On the other hand, the consumption of multimedia sources such as video or audio while also opting for on-demand consumption, independent of the location or consumption status. We have to adapt segmentation and activation capabilities so that advertising is useful and relevant to the user, while we manage to transfer the needs of brands or advertisers.  Mobile will continue growing and so will its interaction with traditional media (TV, radio, OOH), providing them with greater interactivity, personalization and efficiency. We’ve moved to a mobile-centric world where the mobile device has greater relevance and interconnectivity with everything that surrounds us. New ways to exploit it are being opened, especially at a creative level, as we have more and more data and information to personalize and adapt the message at any time and in any place.

9. What are the future plans of TAPTAP and Sonata?

We’re constantly working and adapting to new trends at the technological level and at the operational level in order to continue offering new, more efficient ways of interacting with users. We’re on the way to being able to buy and interact with any platform and any screen (programmatic TV, digital radio, DOOH), thus maximizing the relevance that brands have in any phase of the funnel interaction with users. We keep on growing internationally and readapting to each market and each client. Our goal has always been to maximize the value of advertising, making it more relevant and personal, while maximizing the results of advertisers who rely on our technology and services in any medium and on any platform.

January 9, 2018

Get in Touch

We’d love to hear from you!