The data-based world of programmatic advertising adds new possibilities to ad-creation. Commercials can be less boring or disturbing when they fit users needs. This can be achieved by better targeting for example which shows only relevant ads. But how can these "relevant ads" made more interesting for user? How could we incluence the ad itself and the whole process thats wrapped arround the creation of audio-ads with the goal of serving an individual ad to every user.

What it does

Everytime an audio-stream (like online-radio) or an audio-plattform (like Spotify) is used, its gathering information on the user thru FB-login for example. When it comes to serving ads, these platforms make often use of adserver/ssp-systems. ProgrammagiQ integrates with these systems seemlessly thru established standards like VAST.

The ssp/adserver decides which campaign should be served. But instead of sending a single stored file for each campaign to the player, it pipes the request for an ad to ProgrammagiQ together with user-data. This could come from a local 1st party data store or DMP.

In ProgrammagiQ you can define an ad as a set of audiofiles/audio-slices and rules where the passed user-data and other internal data (like weather, football results) are determining the creation of the audio-ad. This process is done in real-time, means passing the request, evaluating the data, compiling the ad and handing back a VAST should be done in less than 100ms.

ProgrammagiQ offers an intuitive interface that gives agencies the possibility to create rules and choose data-sets to set up individual ads. The whole compiling-magic is than done "under the hood"

How we built it

We started with the idea about a week before hackday drawing some sketches and talking to audio-producers who are experienced with creating classical audio ads. He helped us to produce audio-slices for the case (Sky-ad). After having the audio-slices we thought about the process of compiling audio in realtime. As we solved this it was "just" about coding a nice intuitive frontend and glue it together with the backend process which extracts the data from the request, searches the internal data store for football-results and creates the audio.

Challenges we ran into

When we decided that we want to interact with adservers in realtime thru VAST, we had to master the challenge of compiling audiofiles in realtime. So there is very little time to do the whole process. On the other hand, setting up the rules for the ads could be very complex. So we needed an interface which allows agencies to fullfill the setup process without any knowledge of programming or regexp ;)

Third challenge was to convince audio-producers of this new way of advertisers. Our partners (an experienced copy-writer for audio-ads and an audio-producder) have been very skeptical about the production of just ad-slices instead of an audio-ad. But as we worked thru the process, the result gets better and better. They came up with the idea to use several "audio atmospheres" on the edges of each slice (1st slice: kitchenradio-atmosphere, 2nd slice studio-presenter, 3rd slice with atmospheric music). So the user would not hear that the final ad is produced from small parts in realtime. We'd like to avoid that it sounds like am automated train announcement.

What's next for ProgrammagiQ

Talk to the audio-ad-industry and find partners to roll out ProgrammagiQ?

Share this project: