January 25, 2023
Comscore Pitches Social Media Deduplication Beta
- by Joe Mandese @mp_joemandese, May 10, 2023
Details on the rollout of the new product, dubbed "Social Incremental," were not disclosed, but in a promotional announcement, Comscore described it as providing "deduplicated digital reach across web and social media -- one number, one person, multiple platforms," adding it is capable of deduplicating the following scenario:
"The same individual could be watching a show on a streaming service, browsing social media, and shopping on a website... all at the same time."
The pitch does not disclose the explicit methodology, but describes it as using "machine learning and AI applications."
advertisement
advertisement
6 comments about "Comscore Pitches Social Media Deduplication Beta".
Read more comments >
Next story loading 
I'm not following - if one person is on three platforms simultaneously consuming different content, what is the deduping that is taking place? We typically want to dedupe for social channels when showing unique audience, or reach. So if the same person follows a brand on three social platforms, we want to count them once, not three times.
But if someone is using three platforms simultaneously to consume content, what is the deduping here and what is the impression measurement?
The other challenge is how does Comscore know that I'm one person? If I use three different methods to sign up for three different social channels - maybe my email for one, my phone for another, and a different Google account for another, how does Comscore know that it's the same person - especially if I have three different handles/usernames across those platforms?
Dan, stop worrying...! It's "AI" and "ML."
Adding to Dan's question, how can anyone "use" more than one platform at the exactly the same time? True, one might have two or three or more platforms providing content at the same time on different devices, but you can be attentive to only one at a time---unless you have more than one head or more than two sets of eyes and ears as well as a brain that could process all of the content received.
@Ed it's pretty common for people to use social media while watching a live TV event or a live Twitch stream. For example, maybe someone is watching a live Twitch stream, but tweeting about it at the same time. I'm not one of those people, but I know many who do this - real time engagement during live events (or appointment TV for something like Game of Thrones).
Considering how many people use ad blockers and can opt out of tracking on their mobile device PLUS having different login credentails for different platforms, I'd love to know what their methodology is.
If someone has a million followers on Facebook and 500,000 on Twitter and 30,000 YouTube subscribers - there is obviously going to be an overlap in people who follow the same person on several platforms. Most want to know what the unique audience is because many influencers will offer a cumulative following not taking into account duplicates.
Dan, I agree with you. In fact all of the observational studies---cameras, "spies" etc.--- have long confirmed that people often cease paying attention to various portions of a show they were watching and, in particular, the commercials. We first had hard evidence of this in the mid-1960s and currently from newer studies, mainly using webcams. And certainly, the availbility of smartphones and tablets encourages the activtion of second screens---no doubt about it.
My point about this is that the degree to which this kind of thing takes place is usually overstated to double, triple or quadruple the actual amount. For example people who have an agenda will conduct polls which ask respondents how often they turn to social media ---or their smartphones in general---- while "watching TV". Not surprisingly, many say that this "often" happens which is then interpreted as "it always happens". In reality, the research tells us that during commercials about a third leave the room, a third looks at the TV screen and a third remains but engages in other activities---or simply pays no attention to the TV. This last group is where most of the "other screen" activity is found with about two thirds of it---or 20-25 % of the program audience so involved at a given moment in time. Needless to say, a fifth is a lot less than 100%. And that's for commercials. Even though I have yet to see a similar tally for program content, it's logical to assume that viewers are more attentive and less prone to use other devces while "wtching".
Adding to the list of 'concurrency' is the 'device duplication'.
The issue relates more to 'time spent' than 'reach' with concurrency. Given that servers can monitor at the second level (as can TV meters) we are now accepting low thresholds and duplication.
As an example, in a minute 'session' someone could be watching TV and using FB at the same time. Yes, it is fair to say that both the TV channel and FB can claim they have reached that person.
The FB server will be 'open' all the time but it doesn't accurately know for how much of that minute was on FB (I assume FB attributes the whole minute as the user address is active even if the user isn't), and the TV meter is based on tuning rather than behaviour (e.g. nature calling) and is probably attributed to the minute (we use 15 second in AU).
In such a scenario that single minute would be attributed twice for a nett 2 minutes. Neat trick eh! No wonder usage is increasing.