Data collection can be categorized into four distinct categories:
- Zero-party data, which refers to explicitly collected data, is provided directly by users through forms, chatbots, and similar channels. Such data is willingly and consciously shared by users.
- First-party data is collected implicitly and involves user-shared information retrieved from platforms they are on, such as websites, apps, social media platforms.
- Second-party data, involves data that is directly acquired or exchanged with other companies (it is their first-party data). There is no middle-man in such a transaction.
- Third-party data is acquired from data aggregators; they are not the original collectors of such data.
Reliable and Qualitative Recommendations & Personalization
Collecting data as close to the end-user as possible, cutting out second-hand information, is key to offering on-point personalization, as the likelihood that there be mistakes, misunderstandings and incoherences, is thereby reduced. Staying away from second- and third-party data providers and harnessing a user’s interactions such as their clicks, plays, hovers, and watch time, makes it possible to generate relevant taste-based recommendations. First-party data is effortlessly gathered, stored, and, when properly analyzed, significantly enhances customer satisfaction and retention rates.
For us, resorting to first-party means helping consumers find content they are likely to enjoy based on what they have been previously drawn to – rather than what they would be “expected” to appreciate if their profiles were to be based on personal data having to do with age, gender, occupation, etc. Besides, end users are growingly reluctant to share such information – rightfully so –, and are careful not to leave as much traceable digital footprint.
When it comes to data collection, we believe the sweet spot lies in this happy medium between the accumulation of unnecessary and overshared personal data gathered from end-users who have been stripped of any control over their own information, and the stepping away from any form of data use, so opaque and untrustworthy the whole process often looks, as it fatally leads to poor-quality – not to say unachievable – personalization.
Relevant Ad-Targeting & Addressable TV
Our product “Explore” aggregates all the information collected from the users’ interactions with the platform – and consequently, our semantic metadata –, thereby making up coherent segments of people who share similar tastes. Again, such segmented taste profiles are based on nothing but first-party data, and no personal data are known on our end nor inform the algorithm. Basing yourself on these segments, you can put in place targeted advertising in an easy and pertinent way. We match sets of ad metadata with sets of contents that will suit a particular segment best, and you will then be able to use them to push ads, notifications, cross-media marketing strategies, newsletters, etc.
Contextual advertising is easy to launch thanks to the bridges opened through Spideo’s semantic metadata between audiovisual content and any other related products. In this way, the ads shown to each viewer may be in direct correlation with the program they are viewing on their video platform. The way is thus paved for the promotion of tie-in products, transmedia adaptations, and other items the viewer is most likely to be interested in. For such content-based advertising to be efficient, Spideo has mapped its taxonomy to that of the Interactive Advertising Bureau (IAB). Indeed, the “common language” provided by IAB content taxonomy makes sure various data providers use a consistent labeling so that audience targeting is facilitated for contextual advertising and targeting. We have established bridges between our taxonomy and theirs so that relevant connections between users and products are made, and that no information is lost in the process. Here is how advertising is done most effectively!
Addressable TV can feed on Spideo’s semantic metadata when targeting specific household segments. Indeed, behavioral characteristics – among which the users’ tastes – are taken into account to push certain content, and the right ads may thus be shown to the right people and at the right time whenever they livestream content, watch their regular TV programming or on-demand TV. You can thus customize your TV ads in an easy and relevant way, and one which will engage your viewer a lot more. What’s more, any connected device will benefit from such targeted ads thereby created. Now, that’s the future of advertising!
A Matter of Ethics: Transparency, Trust & Controllability
Spideo algorithm is built on first-party data, but is progressively leaning towards zero-party data. When developing Spideo, we chose to refrain from using personal data and instead focused solely on user interactions, as not only are we convinced they provide the most valuable insights, but they also prevent confining people to predefined patterns, as though their behaviors were entirely determined by the categories they happen to fall into.
The only way around is to base recommendations and personalization on semantics, and this is why we qualify each piece of content with precise and varied keywords, so as to make subtle connections between the items on the catalog and the end-user. Our team of content data supervises semantic content indexation across multiple ever-evolving dimensions (moods, themes, settings, format, time period…), which are guaranteed to be tailored to your needs, fit any content type, and be available in your language. Part of this referencing is AI-generated, although always supervised by our content metadata experts.
We believe it really is in our human approach to AI and algorithms that our uniqueness lies, and our goal has indeed always been to create a transparent product that inspires trust among both our clients and end-users. This led us to introduce the “Semantic Profile”, which is the visual representation of the user’s preferences and tastes. As of today, our clients can play with the algorithm, tweaking it so it best suits their editorial needs and fully leverages their acute domain expertise. Overtime, our ambition is to give the upper-hand to end-users too, turning semantic first-party data into zero-party data. We would like them to have the ability to access and modify their Semantic Profile, so that they become active participants in the recommendation process, and that black-box algorithms spurring defiance and incomprehensiveness are but a vague remembrance of olden days.
Throughout the entire process, transparency, trust, and controllability have been our guiding principles. We built what we know is an ethical recommender system.