�ݺ�ߣ

Editor's Notes

• #6: Webcam can be any webcam, or more high end IP cameras (for instance with long zooms). Quividi is no eye tracking solution (can’t guarantee if viewers have seen a specific point); rather, it’ll count how many people and how long people were exposed to a certain Point of Interest (here: a screen). The tracking makes sure that a person is counted as one, as long as it stays in the field of view of the camera. The classification extracts image features from the head that are then hierarchized to cluster the face into predefined groups. Simultaneously, the system learns what’s fixed and what’s moving in the scene. It analyses the movement and compares it human silhouette movements. Data can be kept locally (it hardly takes more than 10Mb per month), or uploaded every 30 minutes to the Quividi Back-office
• #8: Video is sent to a local machine for real-time processing. No human sees the image (except at setup time) VidiReports turns image into audience data + saves some monitoring info and uploads this regularly into an online server Users access the back-office with their credentials and can’t see data from other clients. Rights make it possible to set who can admin/see what data One last use of VidiReports is to pass a live description of the audience to the local CMS player for it to adapt content
• #9: The default service is that the data is made available to you in an online dashboard, ie a collection of charts and tables which you can customize, edit, duplicate, etc. Data comes in within minutes of the measurement and is constantly refreshed. Quividi can also organize the publication of a report matching a format and branding defined together with you. This is generally set for companies interested in keeping a record of a period (e.g. every month) At last, Quividi can provide a consultant (or recommend local partners) who will perform some advanced datamining, create methodologies, build panels, etc. Typical, a large US retailer employed a consultant for a whole year to work on the Quividi data and make recommendations for the next phase of its roll out.
• #10: OTS = by-passers AUA: this is basically the number of persons, over a given period, which could have seen for sure a clip within a loop. It is computed by dividing the global dwell time of all viewers by the duration of an average loop. More info on http://www.quividi.com/news/090109/pressrelease. Real time data is to be used with applications needing to react to the audience (e.g. change content if a child is standing on a peculiar spot for more than X seconds) OVAB: Average Unit Audience, which it defines as “the number and type of people exposed to the media vehicle with an opportunity to see a unit of time equal to the typical advertising unit”.
• #11: Data coming with authorization from partner MediaReciprocity (a screen network in office business centers, in the south of France) The chart illustrates Combined metrics Rankings Notable differences between screen: the Atrium screen at the entrance of the building has more viewers but with much fewer dwell/attention times vs screen above elevators Data displayed By location and aggregated Could also be grouped (eg all screens above elevators vs all those in corridors)
• #12: Average periods: average day, average week, aver Monday, year, etc. That’s where you see trends going in OTS = nb of passersby, in front of a scren Conversion ratio = % of those who saw the screens There are variations across the day as expected. In that case (screens above elevators in business centers), people get attentive to new content when they come in the morning, then gradually turn away (no refresh in the day) Noon vs 1PM difference : people returning from lunch time (watching screens on their way back from lunch)
• #13: Tokyo on March 11th. Just after the tsunami stroke (3:46PM local time), live video was shown in the subway. People started looking longer, and at the same time many people returned home earlier London Eat Street (within Westfield Center) on Valentine Day 2013: after 6PM, boys realize they need to rush to grab some present for their girlfriend
• #14: There are clear differences in the behavior of demographic groups: some are more attentive to a specific content Some groups rather show up at certain moments in the day
• #15: Eventually, we want to know which content is the most seen. We can do this whatever the software, provided it can export playlogs The media in this table are ranked by nb of viewers. However, media viewer count doesn’t mean much without the context of its duration and number of times each media was played. In that case, Media 398 is the most efficient. This analysis could have been done on a specific demographic group. It could also serve to deliver automated reports for a specific ad campaign, just after it ended airing, much like on the Internet.
• #16: Data provides a strong asset to build your competitive advantage upon If you don’t measure, you can’t improve; Quividi provides information at many different level You can use the data to introduce new biz models You can use it to target specific messages, as we’ve seen with Ocean. Others are doing it on the simultaneous # of viewers, their position, their behavior (eg whether they are staying long, looking away, etc) At last, with data you can start comparing the various locations and sites within your network, but also your complete network with similar ones
• #17: The Quividi technology can serve other purposes: Measuring the shopper behavior in stores. This is achieved with cameras being set on the ceiling to count entries in stores and into specific categories (that’s what we call teh VidiGates mode), but also below points of contact for brands (shelve, end cap, window, animation podiums, etc) Quividi supplies the VidiCube for this Other sensors will soon be proposed to get a complete picture of the customer journey and interactions with brands Measuring specific situations Audiences at home. The goal is not here to replace the people meters used in most at home TV audience rating systems but perform some analysis of how dwellers consume TV at home, in an age of the multiplication of screens. Thinkbox, a research company, got the Grand Prix at The 2012 Media Research Group (MRG) Awards in Nov 2012 for their research ‘Screen Life: the view from the sofa’ and actively used Quividi in that research Quividi is also sometimes used in tradeshows, sports events, to assess the impact of the brands in terms of people seen, attentive, demographics, etc.
• #19: Let’s see now a few examples of DS networks who have been implementing the solution. Equidia is typical of a network who’s decided to go with a panel. To be certified, they appointed France’s leading audience rating company Study paid for itself in just a few month.
• #20: Amscreen is the leader in Europe, both in terms of # of screens, # of countries covered (8 and growing) and thought leadership. The key data points for them is nb of viewers, dwell time and conversion ratio. With good data on all fronts, Amscreen has been using that data to raise substantially it profile with big media planners and announcers. We could have mentioned PilotTV as well, which now has over 2000 screens measured in Taiwan in McDonalds, 7 Eleven, Family Mart stores, etc.
• #21: even large screens (60m²!) can be measured (ie we can use HD cams + track tens of people at once) the network operator here utilizes the technology to the most (triggering + campaign reports per clip) we integrate with well known names
• #22: Our recommendation is to start with a pilot. That’s where you can learn about the techno, compare vendors, experience their efficiency and challenge them to your very needs. 3 months and at least 15 screens is generally what we recommend. Once you have all that data, you should sit down with your vendor and analyze the first received metric At this stage you’ll be able to define a plan and allocate an organization to it. You could either decide to deploy the solution on your all of your future screens: by embedding cameras and going volume, the cost for this audience measurement plaform becomes a low percent of your total deployment cost and can reap in benefits rapidly. Alternatively, you can select a few representative screens to serve as a panel. To have your data recognized, you’ll want to work with research companies on that front.
• #23: Quividi has customers across 35 countries – in many different venues (malls, hypermarkets, superstores, boutiques, restaurants, transportation networks, gas stations, business centers, etc). making it the clear leader in automated audience measurement.
• #24: Measured by Quividi = the quality seal from the leading company