How Quividi compares to mobile data for programmatic dooh

While there are many types of sources of mobile data, app/bid stream is the most used in DOOH today. Bid stream data consists of a repackaging and anonymization of mobile ad requests to act as a proxy metric for audience at a particular location.

This means that in order for it to be usable, the mobile device metric needs to be converted into audience impressions. This often involves correlation with another body of data such as foot traffic or venue circulation.

PROGRAMMATIC DOOH

The online programmatic world is based on standards that require a live client-side data protocol, however, out-of-home has historically been based on historic averages and server-side audience. Mobile bidstream data is currently only available server-side which means it is limited to historical averages as opposed to live measurement.

This is due to 3 data qualification factors:

  • Density
  • Fidelity
  • Recency

DATA DENSITY

When we use the metaphor of density, we mean the ratio of “hard” detection data to “soft” audience data. Hard data comes from the detection model. Soft data is extrapolated or otherwise engineered from the hard data.

  • Maximum Density: the detection model is perfect and can account for 100% of the audience.
  • Minimum Density: the minimum statistically representative sample size needed to make a predictive audience model.
  • Medium Density: somewhere on a spectrum between Maximum and Minimum Density.

Quividi’s model approaches maximum density on camera-equipped locations while mobile bidstream data approaches minimum density as we can see by reviewing the three factors driving data density. They are:

Mobile bidstream has a very low capture rate because it depends on a member of the audience using their phone in a way that an ad request goes out with their location attached as meta-data. Furthermore, the location-based platforms are fragmented, which means not all ad requests are available to the measurement system platform.

This low capture rate has consequences on statistical significance for many types of DOOH screens and it puts limits to data granularity (e.g. hourly audience is not possible) and coverage (e.g. some areas don’t have enough samples to be considered “measured”).

DATA FIDELITY

There is a real challenge converting mobile device data to human-valid, cross-platform impressions for a few reasons. First, if it uses bid stream data, then it is susceptible to the same ad fraud as online ads. Second, it is difficult to discern whether the owner of the mobile device is facing the screen or not. Third, it is often difficult to correctly geofence mobile data to ensure the owner is within the viewable zone. This means that bidstream data is not easily converted into standard audience impressions.

DATA RECENCY

Typically, mobile data such as bid stream or carrier data is gathered after the fact and used as historical data. This is related to the fact that it is difficult technically-speaking to get privacy-clean data to the point of playback in real-time.

CONCLUSION

Mobile bidstream data is a popular solution for estimating audience in DOOH because it can be used to further qualify legacy out-of-home measurement systems. Due to its limitations, however, it is only available as a server-side data source which makes it incompatible with online programmatic requirements.

Quividi is compatible with both a client-side (live actuals) and server-side (historical averages), which makes it compatible with both DOOH-oriented programmatic as well as its online programmatic counterparts.

EnglishChina