The Marketer’s Guide To ACR Tech In Smart TVs

If you own a smart TV, chances are its integrated automated content recognition (ACR) technology tracks everything you watch.

What started out as a niche technology tracking viewership in high-end smart TVs has spread to the masses: Forty-seven percent of the United States’ 120 million homes own smart TVs, up from 39% a year ago, according to Nielsen National TV Panel data.

Smart TVs with ACR tech are finally reaching scale, which is unlocking vast possibilities for advertisers. The tech enables novel forms of targeting, like showing an ad on CTV to someone unexposed to the same ad on linear TV. The tech will power dynamic ad insertion on linear TV. Marketers can use ACR data to connect ad spend to business goals, like driving in-store traffic, and make smarter TV planning decisions.

But as ACR tech becomes more widely used, buyers need to understand the nuances of how ACR works, which also reveal its limitations.

The following is everything you ever wanted to know about what ACR data is, how it works, who supplies it and how it can be applied.

ACR tech has the potential to capture all types of TV viewing: linear, video on demand, OTT, commercials and video games. That’s made it especially attractive as overall viewership splinters.

ACR identifies content by matching snippets of content against a reference library of nearly everything that’s on TV. The technology must be able to make a correct match, and the content must be in the reference library in order to be matched.

To create the reference library, you need “listening posts,” explained Jane Clarke, CEO and managing director of CIMM, a coalition for media measurement. Computers in data centers “watch” TV and catalog what they see. Then, that reference library is matched against a schedule of what ran, so the computer can match the image and audio to a 9 p.m. prime-time episode, for example.

Companies use snippets of video, audio or both in order to match content. Video ACR captures full or partial images on a TV, anywhere from several times a second to every several seconds, and matches those images to existing content stored in a reference library.

Some companies, like Vizio-owned Inscape, also bring in audio ACR for language detection – necessary to identify Spanish-language content, for example. “We started out video only, then switched to a hybrid model,” said Inscape SVP of product, Zeev Neumeier.

The collection, analysis and sale of ACR data happens along a chain. Some companies specialize in one step while others perform two or three functions – which is why it’s first important to understand the convoluted vendor landscape.

Just a handful of companies have their chips or operating systems embedded in smart TVs and collect ACR data. Only two of those companies license their data to others: Inscape and Nielsen Gracenote. The rest run media sales and analytics businesses and keep the ACR data for themselves.

Striking deals with TV manufacturers is a winner-takes-all proposition. You’re either in the TV, or you’re not. So these companies are hyper-competitive with each other. They make confusing claims, trash-talk their competition and have taken their disputes with each other to court. (There have been three lawsuits so far.)

And the space is dynamic. Roku, a recent entrant, placed its ACR-equipped operating system in one in three smart TVs sold in the United States the first quarter of this year, giving it a rapid climb in market share.

Once the data has been collected, TV analytics companies ingest ACR data and combine it with other data sets to make it more accurate and usable.

ACR data is “dirty,” said Denise Colella, NBCUniversal's SVP of advanced advertising products and strategy. “You have to make sure it’s been cleaned and organized and processed in the proper way. It takes a lot of time to ingest that data and learn how to use it.”

TV analytics company iSpot licenses Inscape data for its commercial-focused product. “Most ACR data sets are really, really raw,” said Sean Muller, CEO and founder of iSpot. The data is “useless” without significant effort to make it consistent, accurate and scalable, with planning, optimization and dashboard features on top – like what his company has built.

In addition to iSpot, Data Plus Math (recently acquired by LiveRamp), TVadSync, 605 and VideoAmp license data from ACR data collectors, clean it up and infuse it with other data sources – such as from set-top boxes or panels – which makes the data usable for sophisticated measurement and attribution.

Each company has a different methodology around cleansing their ACR data, and many validate it by cross-referencing it with other data sets.

For example, Nielsen uses its panel as a truth set for its ACR data, said Kelly Abcarian, GM of advanced video advertising at Nielsen.

The final stage is activation. Companies such as Simulmedia and Cadent license ACR data to inform their media buying.

But many of the companies mentioned previously participate in more than one link of the chain. Alphonso, for example, recently stepped up its focus on analytics.

Samsung Ads sells media on its TVs based on the viewing behavior across the Samsung footprint.

Nielsen is getting into targeting with a dynamic ad insertion product that broadcasters can use. Samba TV offers retargeting on mobile phones. Roku uses its ACR data so buyers can show commercials to viewers that haven’t seen a linear spot – improving incremental reach.

But that double and occasionally triple dipping makes ad buyers suspicious. “The ones that sell data and advertising together aren’t honest about the limitations of their tools,” said Omnicom Chief Research Officer Jonathan Steuer.

Who are the vendors and how much scale do they have?

Each ACR company supplied AdExchanger with information about their opted-in US footprint:

Nielsen Gracenote: 18.9 million devices in the United States. Works with 12 TV manufacturers. In TVs since 2013. It licenses data to others on this list.

 Inscape: 11.2 million opted-in TV sets from Vizio in the United States. In TVs since 2014. Purely a data licensing business.

Samsung Ads: 33 million opted-in Samsung TVs in the United States. Does not sell or license its data.

Roku: More than 10 million US TVs have opted in over past two years. Over 10 manufacturers, including TCL, Hisense, Sharp and Hitachi, include the Roku OS. Does not sell or license data.

Alphonso: 15 million opted-in US smart TV households, including ones licensed from Gracenote. Four manufacturers including LG, Hisense (on Android and Linux operating systems) and Sharp. Additional footprint of 19 million from audio ACR, in which smartphone microphones listen to TV shows on select mobile apps.

Samba TV: 14.4 million opted-in TV sets in the United States, and 20 million globally. In market since 2013. 14 TV manufacturer deals. Imprint started in 2013. Does not license its data or use others’ ACR data.

How can marketers use ACR data?

ACR data can be applied at all stages of TV buying, from planning to activation to attribution.

Both Roku and Samsung Ads can serve CTV ads to viewers who haven’t seen linear ads, which can boost reach by double-digit percentages. This feature lets TV buyers compensate for the reach declines plaguing linear.

Still nascent, dynamic ad insertion could make linear TV addressable. Nielsen, using ACR tech from its Gracenote and Sorensen Media acquisitions, is piloting dynamic ad insertion with A&E and CBS. The ultimate goal: “unlock the ability to break up the ad pod in linear,” Abcarian said. Broadcasters continue to sell their own ads, but Nielsen helps keep measurement accurate even when they’re chopped up.

What are the limitations of ACR data?

ACR data has real advantages compared to other data sources, like the Nielsen panel and set-top box data, since it can see both linear and OTT viewing.

But because each ACR company covers only a fraction of US households, scale is a challenge. Sometimes, for instance, the data sets are too small for accurate attribution.

“Some of the attribution use cases are still challenging,” Steuer said. “You might not have enough weight to see what happened from a single smart TV provider.”

Even with data sets millions strong, ACR attribution works best for mass advertisers, like a fast-food chain wanting to measure the increase in foot traffic after a TV spot ran, Steuer said. Advertisers with niche products, like pharmaceutical companies, might not have enough statistically significant data to evaluate on a similar basis.

Another challenge is that even in an era of cheap cloud computing, ACR companies might not store the data long enough for marketers to use, Clarke explained. That makes it difficult to reference historical data.

“It’s expensive to maintain a library. When companies don’t want to pay storage costs beyond a week, that’s an issue for TV measurement and attribution,” Clarke said.

Marketers new to ACR who want to find out how their ads did years ago will often be out of luck. Their commercials may not be in the company’s reference library – one of the holes in ACR data they should be aware of.

For ACR to identify content, the content must be in its reference library. Some companies catalog more content than others.

“There is a lot of variability in terms of the effort ACR providers make to manage those databases, and how long they keep those fingerprints [of shows] around,” Steuer said. “They may only be able to look back a day, a week or a month.”

ACR companies often ignore long-tail content, which sometimes includes local market TV feeds, deep library video-on-demand viewing or tiny cable networks. But that uncatalogued content may matter to a marketer, especially an entertainment client.

Also, many companies struggle to catalog commercials. Even if commercials are in the library, ACRs can miss them or not understand creative variations – like auto ads that feature a local dealer. Nielsen said it developed image recognition in addition to video and audio matching to solve that problem, but most other ACR companies don’t specialize in identifying local content.

Commercials’ short length can foil ACR tech as well, by making it hard to identify the ad before it’s over.

Sometimes, ACR tech can’t match content. Shows that run in many places are tricky, because ACR might not know whose content was being played. That can happen with a show in syndication or if the same show is available on linear TV and a streaming platform.

ACR tech also has trouble identifying content with overlays or that’s been resized, like CNN’s tickers or tune-in promos during a show. Because cable and satellite feeds differ across the United States, ACR companies must track multiple streams (often just East Coast and West Coast) in order to match the shows to their reference libraries. 

Legal contracts can prevent ACR companies from capturing the content being viewed. Native smart TV apps like Netflix, Amazon Prime Video and Hulu often can’t be tracked.

Those contracts don’t necessarily mask all viewing behavior. While Netflix ensures it’s a blind spot across the board, Hulu data can be collected when viewing happens on a streaming stick, instead of the smart TV’s native app.

Most ACR companies wouldn’t share how much of their content ends up unidentified, though it can be a significant amount, according to Steuer.

Nielsen was the only ACR company that readily shared statistics around identification. Its technology identifies 70% of content, video games and ads a viewer watches. The remaining 30% happens places it isn’t allowed to see, like Netflix, or “rarely” can’t be matched. Nielsen also claims a 98% accuracy on its identification.

A TV that watches what you watch just feels creepy. In a privacy-focused climate, that creepiness is precarious. Consumer Reports even created a guide to removing ACR tech from your TV.

Consumer unease with ACR has already caught the attention of Federal Trade Commission (FTC).

In 2017, Vizio paid $2.2 million to the FTC to settle a claim that it misled consumers and collected ACR data without consent. Users who thought they were getting “Smart Interactivity” personalization features actually agreed to data collection.

All ACR data now requires a clear, separate opt-in in the United States.

Although Vizio lost millions of viewers from its footprint when it switched to an opt-in, Inscape now claims to do opt-ins better than the rest of the smart TV pack.

Other manufacturers don’t follow the FTC’s requirements, Inscape SVP of sales and marketing Jodie McAfee said. “We have been through the exercise of unboxing every smart TV in the business to see if they are FTC compliant, and they are not.”

So perhaps a bigger consumer backlash to using the data is yet to come.

“People don’t think of data collection and privacy issues with their TV yet. But it’s coming,” predicted Paige Bilins, chief product officer at Telaria.

Source: Adexchanger.com

Powered by NewsAPI.org

Keywords:

Marketing • Adobe Photoshop • Technology • Television • Smart TV • Automation • Speech recognition • Adobe Photoshop • Technology • Niche market • Technology • Smartphone • Television • United States • Television • Panel data • Television • Adobe Photoshop • Advertising • Advertising • CTV Television Network • Advertising • Linearity • Self-control • Advertising • Linearity • Marketing • Adobe Photoshop • Data • Advertising • Business • Retail • Adobe Photoshop • Technology • Adobe Photoshop • Adobe Photoshop • Adobe Photoshop • Technology • Television • Linearity • Video on demand • Over-the-top content • Television advertisement • Video game • Adobe Photoshop • Technology • Coalition for Innovative Media Measurement • Chief executive officer • Chief executive officer • Coalition for Innovative Media Measurement • Computer • Data center • Computer • Sound • Prime time • Video • Sound • Video • Adobe Photoshop • Image • Image • Vizio • Brigham Young University • Adobe Photoshop • Language identification • Spanish language • Brigham Young University • Vice president • Adobe Photoshop • Data • Integrated circuit • Operating system • Smartphone • Television • Adobe Photoshop • Data • Data • Brigham Young University • Gracenote • Mass media • Sales • Analytics • Business • Adobe Photoshop • Data • Manufacturing • Plurality voting system • Value proposition • Company • Trash-talk • Competition law • Lawsuit • Roku • Adobe Photoshop • Operating system • Smartphone • Television set • United States • Market share • Adobe Photoshop • Usability • Adobe Photoshop • NBCUniversal • Vice president • Advertising • ISpot • Brigham Young University • Advertising • Adobe Photoshop • Chief executive officer • ISpot • Mathematical optimization • ISpot • Adobe Photoshop • Usability • Media buying • Analytics • Samsung Electronics • Advertising • Mass media • Television • Samsung • Advertising • Product (business) • Broadcasting • Samba TV • Mobile phone • Roku • Adobe Photoshop • Data • Television advertisement • Linearity • Television advertisement • Advertising • Advertising • Omnicom Group • Chief research officer • Jonathan Steuer • Distribution (business) • Company • Information technology • Gracenote • Electronics • United States • Television • Brigham Young University • Television • Vizio • United States • Television • Brand licensing • Business • Samsung • Advertising • Samsung • Television • United States • Roku • United States dollar • Television set • TCL Corporation • Hisense • Sharp Corporation • Hitachi • Roku • Operating system • Data • United States dollar • Smart TV • Gracenote • LG Corporation • Hisense • Android (operating system) • Linux • Operating system • Sharp Corporation • Data compression • Adobe Photoshop • Smartphone • Microphone • Television • Mobile app • Samba TV • United States • Marketing • Adobe Photoshop • Marketing • Adobe Photoshop • Data • Adobe Photoshop • Roku • Samsung • Advertising • CTV Television Network • Advertising • Linearity • Advertising • Linearity • Linearity • Television • Gracenote • Doc Severinsen • Mass media • Advertising • A&E Networks • CBS • Advertising • Linearity • Broadcasting • Advertising • Adobe Photoshop • Adobe Photoshop • Set-top box • Linearity • Over-the-top content • Adobe Photoshop • Smart TV • Adobe Photoshop • Fast food • Food chain • Advertising • Pharmaceutical industry • Statistical significance • Cloud computing • Marketing • Adobe Photoshop • Advertising • Advertising • Adobe Photoshop • Data • Adobe Photoshop • Library • Adobe Photoshop • Adobe Photoshop • Marketing • Video on demand • Cable television • Content (media) • Marketing • Entertainment • Customer • Company • Mail order • Advertising • Advertising • Advertising • Computer vision • Video • Sound • Problem solving • Adobe Photoshop • Content (media) • Advertising • Adobe Photoshop • Advertising • Adobe Photoshop • Adobe Photoshop • Broadcast syndication • Linearity • Streaming media • Adobe Photoshop • CNN • Cable television • Satellite television • United States • Adobe Photoshop • Streaming media • East Coast of the United States • West Coast of the United States • Adobe Photoshop • Content (media) • Smart TV • Mobile app • Netflix • Amazon Video • Hulu • Netflix • Law & Order: Criminal Intent (season 6) • Hulu • Streaming media • Smart TV • Mobile app • Adobe Photoshop • Adobe Photoshop • Company • Technology • Video game • Advertising • Netflix • Privacy • Consumer Reports • Adobe Photoshop • Adobe Photoshop • Federal Trade Commission • Federal Trade Commission • Vizio • Federal Trade Commission • Adobe Photoshop • User (computing) • Smartphone • Interactivity • Personalization • Adobe Photoshop • United States • Vizio • Brigham Young University • Smart TV • Federal Trade Commission • Brigham Young University • Vice president • Sales • Marketing • Smart TV • Federal Trade Commission • Privacy • Bilin (biochemistry) • Chief product officer •