Introducing Abe Greenstein, Senior Director, Data Science

Abe is the head of data science at VIZIO with over 10 years of experience in the AdTech industry. He received his Ph.D. in Engineering from The Georgia Institute of Technology and has been working on the advancement and integrity of Inscape data for over 3 years. 

Q: How does Inscape continuously improve its smart TV data?

A: We have close relationships with our customers, so we understand what they need and how they use our data. Our product roadmaps have always reflected that, catering to the innovations and enhancements that unlock granularity and transparency for our clients. Plus, the human capital here is incredible. We have always been enthusiastic when it comes to technical innovation, and we continue to innovate to increase detection rates in streaming, further our transparency in differentiating inputs, and disambiguate and deduplicate similar commercials to provide granular, trusted data.

Q: How does smart TV data help the TV marketplace? 

A: The trend in media over the last 50 years has involved technology making the distribution of more content possible - first cable, now streaming. However, with more content available, fragmentation in the industry underscores the necessity for a unified perspective of TV consumption – that's where our device level data such as automatic content recognition (ACR) comes in. ACR provides scale that can measure viewership on opted-in devices across all inputs: streaming, cable, tuner, vMVPDs, etc. 

Q: Why should the TV marketplace have confidence in Inscape’s ad detection?

A: At Inscape, we are passionate about our products, and therefore we take integrity in being a source of truth very seriously. We have our own TV test lab and an amazing team of engineers dedicated to Q/A. They are hyper focused on rigorous testing methods that replicate real world scenarios, like changing channels and viewing on different inputs, as part of the validation process for our reporting data. For further validation and monitoring at scale we have a variety of non-ACR data sets we use to check the accuracy of our ACR detections – like asrun logs, which are records of which commercials were aired on a TV channel, and tuner data which provides signals of which content was aired directly from the tuner in the TV. Plus, we utilize robust tooling and playbooks for A/B testing. Overall, we know we are doing all we can to help ensure data accuracy, and that in turn helps us provide the market with a trusted source of TV viewing data.

Q: Why do IP addresses and the matching process matter to the TV economy?

A: We know many numbers are floating around in the industry on match rates. The reality is that the numbers are meaningless when shared outside of the context of the problem being solved, which they often are. Match rates are dependent on the quality of both datasets being matched, as well as the accuracy requirements of the match. Our data science team has tested our matching process for IPs, and we had a 95% match rate. Over my 10+ years in AdTech I’ve worked with a variety of IP address datasets, building identity graphs, and performing IP matches. Smart TV data is the best for two simple reasons 1) Unlike your phone or your laptop, your tv never leaves your home. Your friend never brings their TV to your home when they come over for dinner. You never take your TV to the office, etc. 2) TVs are habitually used devices, every time one of our opted-in TV’s turns on it pings us with its current IP address. Because of the quality of our TV-based dataset, customers feel they can trust us and can structure their identity graphs around our data. At Inscape we analyze the data to understand the reliability of IP address data. This allows us to identify the small number of IP addresses / tv combinations that are not reliable because of rotating IP addresses or carrier natting, which occurs when a single IP address represents a group of TVs.  

Q: What would you tell the TV industry about the impact of Inscape’s Tuner Data product?

A: Before tuner we used ACR to detect content and then attribute to the closest ingested network – now we have almost 100% certainty of the channel because we have the exact signal from TV. The tuner data is reliable because we are the TV. The TV knows when it's using the tuner, and which station it's tuned to. Moreover, when a station broadcasts its digital signal, it includes metadata. This metadata is decoded by the TV and includes station identification along with other information. In fact, tuner data has also become one of the validation data sets that we use to monitor and verify the veracity of our ACR data.

Q: What can you tell us about the Political Ad feed?

A: Panels have statistical challenges when measuring smaller markets, and advertisers need simple and effective ways to measure their impact on potential voters, with the ability to pivot strategies quickly. The Political Ad Feed is a tool for confronting these attribution barriers. This new TV data feed provides visibility into locally distributed smaller ads in smaller districts – solving for that fragmentation. The Political Ad Feed provides the trust of our core products, with the trust of AdImpact’s expansive political ad library, giving the market exactly what they need at the perfect time.