Now Trending

The 2018 NBA Finals have come and gone. In the end, the Warriors were victorious after meeting the Cavaliers in the NBA Finals for the fourth straight year, and YouTube TV elevated the experience with a real-time cool factor that we thought was worth highlighting. Their “Presented by YouTube TV” advertising campaign effectively blurred the lines between content and advertising in a truly contextually relevant way.

If you didn’t see the ad, here’s a quick run down. The 60-second spot highlighted the service’s key features, but what’s unique in this specific case is that the ad seemed to organically pull sports-related content into the ad to make those features even more relevant to an audience watching the NBA Finals. This is contextual targeting at its best.

Now, what made this spot truly unique is that, at the end, Warriors star, Kevin Durant gets comfy on the sofa in front of his TV, gives the announcer on the TV, Doris Burke a quick shoutout, and then presses play to watch the game – the irony here is that the game he’s watching on TV is the game in which he’s actually playing at that moment. (Timely, eh?) Then, in wrapping up the spot, Burke acknowledges the shout out and then the live game seamlessly resumes.

YouTube Love on Twitter – 

Beyond Cool: Why does this matter?

So, you may be asking yourself: why do we think an ad like this is so important to call out? The answer is simple. Given changing consumer behaviors, the recent introduction of GDPR, and the continuing evolution of the technologies that impact virtually every part of the day-to-day, the fact that YouTube TV’s ad is not only contextually personalized – based on whatever show the consumer is watching – but that it also blurs the lines between (advertising) content and the real world means the very notion of “advertising” today is rapidly changing. In fact, it’s already changed – and in a pretty big way. This is a good thing for consumers (and brands).

In fact, in a recent study that we conducted with IPG and Magna, we found that content targeting is a much less disruptive – and therefore, much more effective – video advertising solution (vs. demographic/audience targeting and channel targeting). In content targeting, for example, advertising complements the content and sometimes actually feels as though it’s a part of the content itself.

When there is such a strong contextually relevant link between content and advertising, it’s been shown that the perception of the ads associated with that video content receive a lift. Why? Because consumers like unified, consistent, and non-intrusive content experiences. When ads are mismatched with content, they feel like ads. When they complement content, they feel like content.

This is a major win for brands and explains in large part why we’re now seeing a massive shift towards content targeting across all brands and categories, especially in the face of GDPR. It’s simply a more effective strategy for video advertising. Data proves it to be true. Consumer sentiment proves it to be true. Most importantly, it creates a more pleasant – and potentially more trusted – overall viewing experience for consumers.

Learn more about how Zefr can help you master video content targeting on YouTube. 




Product & Tech

The popularity of YouTube, the world’s largest video platform, continues to soar and with that comes an increased appetite for online video content. This comes at a time when (US) online video viewers are expected to hit 236 million by 2020, up from 213.2 million in 2016. For advertisers, the situation is pretty straightforward: spend more advertising dollars where there are more eyeballs. However, achieving precise content targeting at scale remains a big hurdle.

Zefr harnesses the power of machine learning – with a touch of human curation – to wade through YouTube content and classify it at the individual video level. Why is this important? By doing so, we’re able to provide brands and advertisers with a content targeting solution that is relevant (at the campaign level), brand safe, and highly scalable. Only machine learning can do this – effectively – at the truly massive scale of YouTube. (We’re talking billions of videos!)

Machine Learning

Let’s start with the basics: what is machine learning? Machine learning is a computer’s ability to learn and improve from experience without being explicitly programmed to do so. It is another way of talking about artificial intelligence, which was specifically designed, in its early days, to improve the overall speed and efficiency of data-heavy processes. Today, this allows us to tackle otherwise costly and time-consuming tasks at a manageable scale.

As you would expect, machine learning isn’t a one-size-fits-all concept. There are actually three primary types of machine learning: 1) unsupervised, 2) supervised, and 3) semi-supervised. All three types require data input – either “labeled” (manually done by humans) or “unlabeled.”

  • Unsupervised Machine Learning
    How an “intelligent” system learns through unlabeled data inputs. Because this isn’t human-curated learning, there is no easy way to determine learning accuracy through the output (as it’s hard to identify the relationship between input and output). If the output of unsupervised machine learning eventually delivers against a pre-stated hypothesis, you know that “learning” is taking place and delivering the desired end result(s).
  • Supervised Machine Learning  
    How an “intelligent” system learns through a combination of labeled input and output examples. Because the inputs and outputs have been guided by humans upfront, this is the easiest way to ensure learning objectives and, once achieved, begin applying that learning to new functions.
  • Semi-supervised Machine Learning  
    As you may have guessed, this is a hybrid of the two learning models above. This process is guided primarily by unlabeled data, but influenced by a small amount of labeled data to guide and expedite the learning process to achieve the desired outcomes. Combining the two processes have not only been found to be more effective overall, but more cost-effective as well because it requires less time and attention from a human to curate that learning process.

How it Works

The team at Zefr primarily relies on supervised and semi-supervised machine learning models to make the magic happen. According to Zefr Principal Data Scientist, Ryan Deak:

“We have millions of videos to categorize in some way. If we had an infinite amount of time, we (humans) could go through them one-by-one and say, for example, ‘this video has X property, so we want to group it with other videos that also have X property.’”

Unfortunately, we don’t have an infinite amount of time nor would anyone in their right mind want to dedicate that much budget to tackle a process like this manually. Doing so is simply not scalable in any way. That’s precisely where the supervised machine learning comes into play.

“We can go through the ‘cherry-picked’ process for a select number of cases to label videos based on the properties we find. At that point, however, we pass those labeled videos over to the machine learning algorithm, and it analyzes that input to predict which unlabeled videos should be associated with a particular category. The ultimate goal of machine learning, therefore, is to be able to generalize what it’s learned, based on the labeled data provided, and accurately apply that learning to data it has never analyzed before. This is how Zefr scales a massive amount of online video content to deliver a precise content targeting solution to brands and advertisers.

Precise content targeting for online video content at scale is the challenge that Zefr's technology solves for.

One of our key differentiators are the proprietary tools we use to streamline this machine learning process. We use a framework called Aloha, which makes it easy to extract data (from content) and integrate it into our learning and prediction pipelines. This is huge! All too often, as much as 75 percent of a data engineer’s time is spent on manual tasks – like extracting and cleaning data – instead of focusing on more complex, problem-solving-oriented tasks.

“In the past 5 or 10 years, there has been an explosion of higher quality open source machine learning libraries, like Spark, Vowpal Wabbit,, and MXNet to support our deep learning needs. The fact that we operate on a single framework (Aloha) to integrate all these libraries means both our data scientists and data engineers can do their jobs much more efficiently without being bogged down by the typical troubleshooting that comes with model deployment.”

*More information about these libraries is outlined in “Hidden Technical Debt in Machine Learning Systems” and in our own SysML 2018 conference paper, focused specifically on Aloha.”

Human Review

The massive volume of videos on YouTube means that machine learning is table stakes when it comes to developing and deploying a precise content targeting solution. It is the only efficient way to navigate – and make sense of – the totality of the YouTube online video ecosystem. As Deak points out:

“It’s not like TV where you have maybe 1,000 TV shows at any given time in a year vs. a billion videos on YouTube. It’s both a problem and an opportunity for us as well as for advertisers. The only hitch: it’s an opportunity that’s about a million times the size of TV.”

The truth is, without some human review mixed into this process, it’s difficult to accurately assess the relevant alignment of a particular piece of video content to a particular brand message. That’s all about brand safety and, in spite of how advanced machine learning has become, we’re not ready to let artificially-intelligent algorithms take complete control of the driver’s seat here. There’s a certain ethos involved with human review and curation that rounds out this process – and all with the goal of driving the most value to brands and advertisers at the end of the day. Eliminating human review from the machine learning process raises the chances of “unsafe” content slipping through the cracks or, more generally speaking, content misalignment with brands. At this point in time, that’s not a gamble we’re willing to take.

Our machine learning process starts and ends with human review. We feed our algorithms useful data and then assess and evaluate the output from those algorithms. And it’s thanks in big part to this human-machine collaboration that we’ve been able to create highly accurate models that can blacklist unsafe content, whitelist safe content and deliver more precise content targeting.

The Zefr Touch

The team at Zefr is evolving alongside machine learning. We’re creating new ways of making machine learning more accurate, effective, and efficient vis-à-vis YouTube while simultaneously taking those learnings to make our (human) team more effective as well. The foundation of our system is based on modeling and scoring. This means that our algorithms are built specifically to provide high quality machine learning models that improve as they learn over time. Adding the human layer to this dynamic is our “secret sauce.” We know the ins and outs of the YouTube ecosystem better than anyone else (except maybe for YouTube). For this reason, Zefr is uniquely positioned to provide brands and advertisers with a precisely targeted opportunity to align their promotional messages with contextually relevant online video content – all at scale!

InterviewsThought Leadership

The Brave New World of Online Video Consumption

Consumers continue to catapult online video consumption into the ethers. For each of the past four months, YouTube video hourly consumption has risen 30%. By 2020, video is predicted to account for 79% of global internet traffic. Brands and advertisers are scrambling to stay at the top of their game and embrace this brave new world, while consumers have traded in their remote control for consumer control, aka choice.

Video Visionary, Rich Greenfield

Recently, Zefr caught up with Media Futurist, Media and Tech Analyst at BTIG, Rich Greenfield – a video visionary in his own right. Greenfield spoke of the dynamic between consumer and advertiser, the advantages of the data-rich environment in which we live, brand safety and the beneficial partnership between advertisers and “companies that truly understand the digital sphere.”

ZEFR: In this rapidly evolving video consumption landscape, what gives? Is the steady decline in traditional television viewership a result of uninterested consumers?

Rich Greenfield: Consumers just have so much choice. It’s not that they don’t like what’s on television, it’s just that they’ve got more choices. The number of scripted television shows has more than doubled over the last five years. So, there’s actually an explosion – there’s over a hundred digital series that are being created right now.

ZEFR: That’s a large playing field for brands to tackle. What do you see as the biggest ad innovation and opportunity being for advertisers right now?

RG: Putting the consumer in control. So, not only showing the right ad, but making sure to give the consumer control. Do they want to watch this ad? I think if you give the consumer the control over whether they want to “watch this ad” you end up with a much happier, satisfied consumer.

“I think if you give the consumer the control over whether they want to ‘watch this ad’ you end up with a much happier, satisfied consumer.”

ZEFR: Speaking of control, what are your thoughts on pre-roll?

RG: The pre-roll business on YouTube is well north of 10 billion dollars globally – it’s a huge business. Pre-roll works. I think when you put the right ad in front of the right person, it’s not upsetting. Whether we’re talking about YouTube or Facebook or lately, Instagram – the ads are becoming content.

ZEFR: Brand safety – resolved or do you seeing the challenge continuing this year?

RG: The challenge is, digital is scary. It’s really hard to police – there’s an infinite amount of content and you’re going to end up with the issues around brand safety that have come up.

ZEFR: What do you see as the solution for brands? How can they protect themselves without sacrificing online video advertising?

RG: Brands have to figure out how to partner with companies who really understand the digital sphere. I think the traditional media companies are really struggling to figure out digital. If they want to push product—move cars off lots, products off shelves—they’re going to have to reach consumers. The only way they’re going to do that is through the big digital platforms and they’re going to have to embrace them.

View full video –



Last week, Cannes Lions brought the global advertising industry together in the French Riviera. Conversations that had been taking place around the world came to a head as the most important minds in media and marketing joined together to discuss the convergence of technology and creativity to create and distribute the world’s best work. Here, we share our key takeaways from the event, rounding up what mattered most on and off the Croisette.

Innovation Matters

Communication and innovative thinking are key when it comes to maintaining relationships and seeing results in brand marketing, and both clients and agencies benefit when a commitment to both is applied. Our co-CEO Rich Raddon shared his thoughts alongside executives from FourSquare, OMD and more in a discussion onsite.

Brand Authenticity Matters

Brands now, more than ever, are seeking ways to relate to their fans and involve themselves in the conversations happening around them on all platforms. Social video allows them to so that in a way that feels honest and involved. Zefr’s Trygve Jensen joined client Kingsford Charcoal and top Snapchatter Shonduras to discuss how to integrate influencers into brand campaigns while maintaining authenticity in message, for both the influencer, and the brand. Read a recap in the Lions Daily News on pg. 16.

Context Matters

The content an ad runs against impacts the way it is perceived. As advertising investments shift from TV to digital, brands are looking for ways to maintain the traditional content alignment they’re used to via contextual targeting, like that offered by Zefr and its BrandID technology. For more on the way TV viewing behavior is changing,what brands need to know, and why context matters, download our original research on TV 3.0.

For more on Zefr and our contextual targeting solutions for brands, request a demo here.








Moms are a core audience on YouTube. The platform solves almost all of their digital needs – from entertainment to crafts for the kids to problem-solving and advice. In fact, a recent Google study found that 83% of mothers search for answers to their questions online – and 60% of them turn to online videos in particular.

YouTube plays an important role in Mom’s life, and to better understand her interests ahead of Mother’s Day, we looked into what she’s watching on YouTube.

356 Million Total Views of Mother’s Day Content on YouTube

Much of the Mother’s Day content on YouTube is about love, heartfelt messages, and praise for Moms. People share Mother’s Day moments, gift ideas, and surprises that can help others think of ways to make their moms feel special. We found 365 million views on Mother’s Day content on YouTube, presenting an opportunity for brands to reach moms around the day that’s all about them.

Mother’s Day Topics and Trends

Inspirational / Happy Moments

Moms do so much for their families, but Mother’s Day is the day to give back to them. Kids young and old turn to YouTube to share their love for Mom, whether they’re reading a poem, giving mom praise, or simply saying, “I love you.” For example, “Moms Are Magic” is a heartwarming video featuring young children sharing their favorite things about their moms.


Not all Mother’s Day content is sentimental. Entertaining videos, like music and kids content, makes up a large portion of what’s being viewed around Mother’s Day. Many musicians have created songs especially for their moms and children’s programs, like this clip from Angry Birds, educate kids about the meaning of Mother’s Day.


Of moms who watch videos on YouTube, 81% watch how-to content. In fact, moms are significantly more likely to watch how-to content than the average viewer. How-to and DIY videos that feature Mother’s Day are also popular on YouTube – from DIY Mother’s Day gift ideas to how to put together the best surprise party for mom.


Although practical, informational videos are popular amongst moms, that’s not all they’re watching. Moms also go to YouTube to watch funny content and have a good laugh!


Influencer Moms on YouTube

When families are searching for the perfect gift for mom on Mother’s Day, they turn to mom influencers on YouTube for ideas. 68% of consumers say recommendations influence their Mother’s Day gift purchases. Working with the right influencers can help a brand amplify its message to an audience that is seeking advice.

We used our technology to identify influencers creating content that resonates with moms on YouTube:

Five Fun Mom Influencers To Follow On YouTube

1. What’s Up Moms

2. Daily Bumps

3. Ellie and Jared

4. Rachel Talbott

5. Brittani Louise Taylor

Align with the Best Mother’s Day Content on YouTube

Mother’s Day videos on YouTube provide an opportunity for contextual alignment, allowing brands to reach consumers at a moment when their message is an extension of the content moms are watching.

Zefr provides brands with the opportunity to target the most relevant Mother’s Day topics and trends and amplify that messaging with custom influencer activations.

Sign up for a demo to find out how our technology can help you align your message where and when it matters most.

Register for a demo now!

Thought Leadership

Brands are increasing their investment in YouTube advertising, as the largest video platform in the world continues its rapid growth. Marketers know they need to advertise on YouTube to reach their audience, an insight that is well supported by audience viewership data. But today, most brands lack the control or visibility into the specific inventory they are buying, which raises critical issues regarding the brand safety of the content they are running against.

Recent news has thrown into focus the issue of brand safety as marketers are running ads against content that can be created by anyone. PewDiePie, arguably the world’s most successful influencer, lost his standing with Disney’s Maker Studios and with YouTube due to anti-Semetic stunts. Super Bowl ads have been seen running in front of terrorist recruitment videos. These stories are trending, but the issue isn’t new – and it’s only going to grow as more content is created and brands continue to increase their investment on the platform.

How does it happen?

Most advertisers target their intended consumers  on YouTube based on keywords, audience (demographic) and channel (typically via Google Preferred, the most popular influencer and creator channels). But each of these approaches assume that all of the content isolated by one of these targeting options is similar and therefore safe. ZEFR has found that this isn’t the case.

Let’s take Michelle Phan’s channel for example. She’s known as a beauty influencer, but only 52% of her content is related to beauty. The other 48% is better categorized as lifestyle content, including career advice and current events. A beauty brand may only want to align with that  52% of videos, but if they’re buying her channel in Google Preferred, they don’t know which videos within the channel they’re aligning with.

Michelle Phan channel audit conducted Summer 2016
Michelle Phan channel audit conducted Summer 2016

The same theory applies to brand safety. Not all content within a creator’s channel will be considered safe or on target for every brand.

There is another way

ZEFR’s technology delivers TV-like contextual relevance and brand safety on YouTube, allowing brands to leverage the incredible potential of the platform at the individual video-level. There’s a tremendous amount of great content on YouTube for brands, and video-level targeting allows for marketers to exclude specific videos from their YouTube buys, ensuring that they’re never running ads against content that is not relevant. It requires an understanding of each discrete video that only ZEFR is able to provide.

ZEFR’s solution

Here’s how it works. A video is uploaded to YouTube. ZEFR analyzes the video for:

  • Relevance – Is the video content aligned with the brand’s media strategy? What is this video actually about? If a video is of a father and son tossing a football and it’s called, “My Son is the next Tom Brady,” the video isn’t about Tom Brady. It’s about a father and son moment.
  • Brand safety – Is this video appropriate for a brand to align with? Does it have profanity, negative imagery, violence, or controversial opinions? Is it promoting negative activity, or hate speech?
  • Forecasting – Is this a video that’s generating views, or trending upwards? Will it deliver enough impressions so that an ad is seen by a real human audience at scale?
  • Performance – Is this a video that can carry an ad? Is it meaningful enough content to justify placing an ad before it? Will it help you reach your key performance KPIs?

If each of these indicators are met, the video is organized into a content segment with thousands of other videos that are all contextually related. Every day, ZEFR evaluates over 8 million videos, while only qualifying 250,000 of those videos for inclusion in premium, brand safe campaigns that deliver performance.

Video-level targeting is the future

So, should brands be afraid to align with influencer content on YouTube? The answer is no. Brand safety is an easily mitigated risk when video-level targeting is employed. “It’s so important to target ads at individual videos on YouTube, so brands can ensure that they’re aligned with the most relevant content at any given time,” said Rich Raddon, co-CEO, ZEFR. “The content within channels varies, and not every video is right for every brand. ZEFR’s technology and review process ensures that an ad will never run against a video that is not considered safe.”

For more on ZEFR’s video-targeting solutions, contact us at

Oded Noy is the Chief Technology Officer at ZEFR


The link between content alignment and success on YouTube

The kind of content an ad runs against has a huge impact on the way a person reacts to it. It’s true on television, in magazines, and now – more than ever – online.

With more than 400 hours of video added to YouTube every minute, there’s an incredible breadth of content for brands to advertise against. But how are they to know what to advertise against? With so many videos to choose from, how does anyone know who’s watching what? Historically, there’s been no way to determine what types of video content brands should align with – until now.

At ZEFR, we have unique video technology that allows us to understand the content of every video uploaded to YouTube. We organize YouTube by content segments – over 13,000. This allows brands to align their ads with the specific type of content a person is seeking out. Pairing ads with relevant content delivers a better experience for users and brands, and for the first time, we’ve mined the data to discover what content works best for various brand verticals to advertise against. The idea is to align ads with the right content, hence the name of the project – The Alignment Effect.

What we discovered was an interesting blend of results of what works. Some content area/brand matches you might expect, and some that were surprising. For example, segments like food challenges and 4th of July Parties worked well for CPG brands. That was expected.  But premium entertainment content – like Aziz Ansari and Swedish House Mafia – also delivered strong results. Within the entertainment vertical, we saw both traditional entertainment content and content unique to YouTube perform equally well. Traditional stars like George Clooney and Adam Sandler appear alongside personalities that were born on YouTube – Miranda Sings and Sam Tsui. The traditional definition of “entertainment” has broadened, and brands can (and should) take advantage of the content emerging in the category.

CPG example

Great content is everywhere on YouTube, and audiences go where the content is. By showcasing what types of content work best for brands, we can create better outcomes for advertisers and a better experience for consumers – helping to drive optimum success on the platform.

Click here to download The Alignment Effect for a deeper look at what kind of content works for brands.

Danielle DePalma, Lionsgate


Thought Leadership

This piece originally appeared in MediaPost.

The ad industry is facing a Hollywood-style potential apocalypse—and the only question is will it turn out like Armageddon or The Day After Tomorrow?

The growing and real threat is that software and hardware being developed and used today can cut out ads altogether. And like our machine nemesis from The Matrix—the machines don’t care if they are good ads or bad ads, helpful or annoying. Once they are in charge it’s simply Ad-mageddon.

It should come as no surprise that people are using technology to cut down on ads. According to various studies, we are exposed to between 3,000 to 6,000 advertising messages a day. Considering how many of these messages, paid to grab our attention, are completely irrelevant to us, it’s more surprising we aren’t more outraged.

According to Nielsen, we will spend two years of our lives watching TV commercials. Compare that to the mere 48 days, on average, we will spend kissing during our lifetimes and the issue becomes very clear: our lives are being wasted as we wait through completely irrelevant ads. This general disdain is not new—but the tools and technology to mobilize those feelings into action are.

By the way, it’s not just consumers who should be outraged. Advertisers should be outraged as well. It does them no good when they pay for ads that reach the wrong people.

It seems that as soon as there were mass media to advertise in, there were people complaining about the ads. “Many of the claims made for products were excessive and often mendacious, bringing advertising into disrepute well before the turn of the [19th] century,” wrote Jeremiah O’Sullivan Jr. in The Social And Cultural Effects of Advertising.

So why is today’s impending doom any more real than last year’s, or when the VHS became fashionable?

There are two reasons that this time the threats are not idle—they come in the form of hardware and software.

There are machines that let consumers skip TV ads. DVR penetration in the U.S. now stands at 63%. And if ads people see during their favorite shows remain mostly irrelevant, the number of people using their DVRs to skip ads will continue to climb. Once people have decided to skip ads it will be very difficult to convince them to go back to the old ad-filled ways.

The latest and most telling development is the release of software that simply helps people “turn off” ads on digital platforms. Two weeks ago, three of the top ten iPhone apps in the App Store were ad-blocking apps. Yes, people have been grumbling about ads forever, but the tech landscape has now provided an outlet to turn those feelings into actions that have a very real and very deep impact.

Back to our central question: is the industry facing an Armageddon or The Day After Tomorrow scenario? Otherwise stated, is this a story that can have a happy ending?

The good news is that just as technology has empowered a potentially cataclysmic industry event, it also has the power to fix the problem. Search advertising has proven that the information typed into the search bar provides a rich indication of what the perfect accompanying ads should be. Today, people, through things like search terms, are constantly sending out signals about what brand message they are most open to. On YouTube, that little search box has the power to make sure the ads are finding the right people at the right time.

Mobile opens new worlds of data that can use location to provide even more accuracy in the game of matchmaking the right ad messages to the right consumer at the right time. The digital world has the data to save the day. But the clock is running out. Many of the digital giants are holding onto the data that might be the cure for the entire industry.

Time is the factor. We have to improve the accuracy of connecting the right ads to the right people at the right time before the ad industry self-destructs. We have the tools, and the need—now all we have to do is realize the asteroid is hurtling towards us. Moving slowly is the same as doing nothing at all.

Modified image via Ryan Hyde


Get future posts delivered to your inbox