<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[This Week In AdTech - Medium]]></title>
        <description><![CDATA[This Week In AdTech helps mid-size &amp; enterprise companies that own a content platform, improve their ad platform’s click-through-rate, relevance, and overall revenue as it relates to Gen Z. | SiliconValleyConsulting.io - Medium]]></description>
        <link>https://medium.com/this-week-in-adtech?source=rss----2f232efba4af---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>This Week In AdTech - Medium</title>
            <link>https://medium.com/this-week-in-adtech?source=rss----2f232efba4af---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Mon, 16 Nov 2026 20:42:12 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/this-week-in-adtech" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Top Journalists in AdTech Mini-Series: Kathryn Lundstrom]]></title>
            <link>https://medium.com/this-week-in-adtech/top-journalists-in-adtech-mini-series-kathryn-lundstrom-29c6357b8239?source=rss----2f232efba4af---4</link>
            <guid isPermaLink="false">https://medium.com/p/29c6357b8239</guid>
            <category><![CDATA[branding]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[advertising]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[marketing]]></category>
            <dc:creator><![CDATA[Shane Austrie]]></dc:creator>
            <pubDate>Mon, 10 Aug 2020 17:13:43 GMT</pubDate>
            <atom:updated>2020-08-10T17:13:43.559Z</atom:updated>
            <content:encoded><![CDATA[<h4>Kathryn Lundstrom writer at Adweek…</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*956NqsjkKHPCSIJpzl_WDA.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@climatereality?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">The Climate Reality Project</a> on <a href="https://unsplash.com/s/photos/journalist?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><blockquote>The only Mini-Series doing quick, two-minute highlights of some of the best digital writers in Marketing, Branding, and AdTech!</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4aPRw5dKCH8fzs77uDLWtA.jpeg" /></figure><h4>Overview</h4><p>Katheryn Lundstrom is currently a writer at Adweek. She covers stories ranging from advertising privacy to brand marketing.</p><h4>Background</h4><p>As a Minnesota-native, Katherine is an enjoyer of all four seasons: from eating Thanksgiving squash to people watching during Halloween.</p><p>She spent a few years in Wisconsin, completing her Bachelor&#39;s in International Studies at the <strong><em>University of Wisconsin-Madison.</em></strong> A few years after that point, she decided to move to Austin in order to complete her dual Master’s in Journalism and Global Policy Studies at the <strong><em>University of Texas: Austin.</em></strong></p><p>During her time at <strong><em>UT Austin</em></strong>, she held two internships: one as a Research &amp; Writing Intern for the <strong><em>Nonviolent Peaceforce</em></strong>, and another as a Reporter Intern for <strong><em>Sports Illustrated</em></strong>.</p><p>After college, Katherine dedicated her time to writing for several important causes. She was a Junior Writer for the <strong><em>University of Texas</em></strong>, Reporting Fellow for the <strong><em>Texas Tribune</em></strong>, as well as a Policy Analyst for the <strong><em>Texas House of Representatives</em></strong>.</p><p>After her time with the <strong><em>Texas House of Representatives</em></strong>, she continued to strive for impact by writing for <strong><em>the DONUT</em></strong> — where her writing focused on politics and current events.</p><h4>Now</h4><p>Katherine is currently a writer for the NYC-based journal, <strong><em>Adweek</em></strong>. Some of the top articles she has covered at <strong><em>Adweek</em></strong>:</p><ul><li><a href="https://www.adweek.com/brand-marketing/diageo-pulpex-limited-johnnie-walker-whisky-paper-bottles/"><em>Johnnie Walker Whisky Will Come in Paper Bottles Next Year</em></a></li><li><a href="https://www.adweek.com/tv-video/bet-is-leading-the-conversation-on-racial-justice-with-new-programming/"><em>BET Is Leading the Conversation on Racial Justice With New Programming</em></a></li><li><a href="https://www.adweek.com/brand-marketing/is-hard-seltzer-here-to-stay-or-will-it-go-the-way-of-the-wine-cooler/"><em>Is Hard Seltzer Here to Stay — or Will It Go the Way of the Wine Cooler?</em></a></li></ul><p>Katherine is one of the more up-and-coming digital journalists on our list. Through her writing, she has made a name for herself and has shown that she has great aptitude in the sectors of Marketing, Branding, and AdTech. Her current writing mainly focuses on a range of topics around branding and consumer insights.</p><h4>Our Fan Favorite</h4><p>A recent fan-favorite from her is her take back in July on <a href="https://www.adweek.com/brand-marketing/heres-where-u-s-data-privacy-regulation-is-headed/">the direction of data privacy and regulation in the US</a>.</p><p>Though, most of the article is about Adweek’s CEO — Jeff Litvack — opinion, we do like a quote from the interview that she pointed out: “<em>A more comprehensive regime would give consumers the ability to opt out or in early on in the process.</em>”</p><p>Ever since GDPR and CCPA were enacted, we have been experiencing issues from all sides of the advertising industry.</p><ul><li>Small advertising platforms (<em>e.g. news publications</em>) can no longer use third-party cookies nor third-party ads. Additionally, a lot of them do not know how to create their own first-party advertising platform.</li><li>Large advertising platforms have to be extremely cautious of the information they collect or else risk being heavily fined (<em>something that has been happening heavily recently</em>). Additionally, they are losing the ability to put their advertisers’ ads on third-party websites (<em>e.g. the Audience Network feature in Facebook Ads Manager</em>).</li><li>Advertisers are experiencing a widespread problem: declining click-through-rates, ad fraud, and slowly losing the ability to access the wide-spread of data that advertising platforms used to give them.</li></ul><p>Follow her on Twitter: <a href="http://twitter.com/klundster">http://twitter.com/klundster</a></p><p>Reach out to her over email: <a href="/service/mailto:kathryn.lundstrom@adweek.com">kathryn.lundstrom@adweek.com</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=29c6357b8239" width="1" height="1" alt=""><hr><p><a href="https://medium.com/this-week-in-adtech/top-journalists-in-adtech-mini-series-kathryn-lundstrom-29c6357b8239">Top Journalists in AdTech Mini-Series: Kathryn Lundstrom</a> was originally published in <a href="https://medium.com/this-week-in-adtech">This Week In AdTech</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Top Journalists in AdTech Mini-Series: Tim Peterson]]></title>
            <link>https://medium.com/this-week-in-adtech/top-journalists-in-adtech-mini-series-tim-peterson-9236b35baeda?source=rss----2f232efba4af---4</link>
            <guid isPermaLink="false">https://medium.com/p/9236b35baeda</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[branding]]></category>
            <category><![CDATA[adverting]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[marketing]]></category>
            <dc:creator><![CDATA[Shane Austrie]]></dc:creator>
            <pubDate>Mon, 10 Aug 2020 17:13:36 GMT</pubDate>
            <atom:updated>2020-08-10T17:20:40.020Z</atom:updated>
            <content:encoded><![CDATA[<h4>Tim Peterson writer at Digiday…</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*956NqsjkKHPCSIJpzl_WDA.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@climatereality?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">The Climate Reality Project</a> on <a href="https://unsplash.com/s/photos/journalist?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><blockquote>The only Mini-Series doing quick, two-minute highlights of some of the best digital writers in Marketing, Branding, and AdTech!</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/1*QaKV80dJNNEG3_Rph4Us7Q.jpeg" /></figure><h4>Overview</h4><p>Tim Peterson is an LA-native and current Senior Media Reporter at Digiday. His stories range from discussing the Big Two AdTech Conglomerates to theorizing the future of Digital/Linear TV.</p><h4>Background</h4><p>Tim has both an interesting educational background and an interesting professional background. Before even attending school, he held a writing position at the <strong><em>Santa Monica Observer </em></strong>— a position that he held for a bit over a year.</p><p>After his time at the <strong><em>Santa Monica Observer</em></strong>, he attended <strong><em>New York University</em></strong> for undergrad. While at <strong><em>NYU</em></strong>, he specifically attended the Gallatin School of Individualized Study allowing him to uniquely craft and tailor his degree path. He held several positions while in college, one notably being an internship at <strong><em>Lagtons International Agency</em></strong>.</p><p>During his senior year at <strong><em>NYU</em></strong>, he joined the <strong><em>Meredith Corporation</em></strong> as an Associate Editor. His two years at the <strong><em>Meredith Corporation </em></strong>gave him exposure in the consumer interest and sports entertainment side of journalism.</p><p>After his time in that sector of journalism, Tim spent the next year as a Reporter for <strong><em>Direct Marketing News</em></strong>. Being a journalist at <strong><em>DMN</em></strong> is what initially got Tim into the digital marketing and advertising space. Some of his work covered exclusive interviews ranging from the CMO of <strong><em>Dell </em></strong>to the CMO of <strong><em>Clear Channel Digital</em></strong>.</p><p>Continuing his journey in the MarTech and AdTech space, Tim joined <strong><em>Adweek</em></strong> in 2012 as a <strong><em>Staff Writer</em></strong>. As an <strong><em>Adweek</em></strong> journalist, Tim got to witness several great events in the early, blossoming stages of AdTech:</p><blockquote>*Excuse the usual Adweek paywall</blockquote><ul><li><a href="https://www.adweek.com/digital/twitter-opens-ad-demand-floodgates-self-serve-expansion-149051/"><em>Twitter becoming a self-service advertising platform in 2013</em></a></li><li><a href="https://www.adweek.com/digital/ad-targeters-are-laying-groundwork-visual-advertising-148760/"><em>The early days of Instagram advertising (pre-influencer marketing</em></a><em>)</em></li><li><a href="https://www.adweek.com/digital/facebook-exchange-rolls-news-feed-148166/"><em>Facebook using third-party data for News Feed ads</em></a></li></ul><p>Considering that Facebook is currently dealing with privacy regulations, as well as Instagram ads are experiencing decreasing CTRs (<em>without influencers</em>), it’s easy to see Tim’s early thought leadership growth in the space!</p><p>After spending a year at <strong><em>Adweek</em></strong>, Tim joined <strong><em>Ad Age</em></strong> mid-2013. At <strong><em>Ad Age</em></strong>, Tim got to cover the early days of some of popularized AdTech platforms of today — platforms like <strong><em>Snapchat</em></strong>, <strong><em>YouTube</em></strong>, <strong><em>Hulu</em></strong>, <strong><em>Spotify</em></strong>, and <strong><em>Pandora</em></strong>.</p><p>Between <strong><em>Adweek</em></strong>, <strong><em>Ad Age</em></strong>, and later <strong><em>Third Door Media</em></strong>/<strong><em>Marketing Land</em></strong>, Tim has a plethora of articles and fireside chats predicting future trends that ended up becoming reality! In 2015, Tim wrote about <a href="https://adage.com/article/digital/facebook-retailers-set-shop-site/299509"><em>Facebook wanting retailers to use their Facebook Pages as their primary storefront</em></a>.</p><blockquote>That prediction recently proved true</blockquote><p>As retailers have started to realize that consumers may never be going back outside like they used to, <strong><em>Shopify’s</em></strong>— <em>eCommerce management platform </em>— stock prices have <strong><em>soared </em></strong>as more businesses move online permanently. One of our AdTech consultants at SVC recently worked with <strong><em>Facebook</em></strong> on building and releasing their new Shopify integration that allows for COVID-affected businesses to move online permanently.</p><blockquote><strong><em>As consultants who worked with Facebook, we can vouch that Tim’s prediction for this was 100% spot-on!</em></strong></blockquote><h4>Now</h4><p>Tim is currently writing for <strong><em>Digiday </em></strong>and he’s rapidly approaching his third year as a Senior Media Reporter. Over the years Tim has built shown thought leadership in the space — <em>making him an obvious choice for this Mini-Series!</em></p><p>Tim averages at least one new article per day. Some of his top recent articles include:</p><ul><li><a href="https://digiday.com/future-of-tv/why-warnermedia-still-has-so-many-hbo-branded-streaming-services-a-study-of-a-branding-and-distribution-headache/"><em>Why WarnerMedia still has so many HBO-branded streaming services: a study of a branding and distribution headache</em></a></li><li><a href="https://digiday.com/future-of-tv/a-show-you-can-play-twitch-moves-deeper-into-producing-live-interactive-shows/"><em>Twitch moves deeper into TV-like programming (with interactivity)</em></a></li><li><a href="https://digiday.com/future-of-tv/how-the-future-of-tv-and-streaming-has-been-reshaped-so-far-by-2020/">How the future of TV and streaming has been reshaped so far by 2020</a></li><li><a href="https://digiday.com/future-of-tv/shut-out-of-fire-tv-and-roku-peacock-is-the-latest-example-of-the-arrival-power-moves-to-streaming/">Shut out of Fire TV and Roku, Peacock is the latest example of the arrival power moves to streaming</a></li><li><a href="https://www.adweek.com/digital/the-new-media-wars/"><em>AT&amp;T and Verizon Are Making Very Different Bets on What 5G Will Mean for Consumers and Content</em></a></li><li><a href="https://digiday.com/media/reddit-overhauls-ad-sales-with-a-new-boss-from-pinterest/"><em>Reddit overhauls ad sales, with a new boss from Pinterest</em></a></li></ul><h4>Our Fan Favorite</h4><p>Our non-paywalled fan-favorite is his recent article on <a href="https://digiday.com/future-of-tv/viacomcbss-upfronts-pitch-to-advertisers-we-wont-bludgeon-people-with-the-same-ad/"><em>ViacomCBS</em></a>! <strong>Something that should be noted is that the Viacom and CBS merger was a necessary merger in hindsight.</strong> This merger allows ViacomCBS (<em>and Pluto TV</em>) to compete with Comcast (<em>which has acquired several streaming platforms recently such as Now TV and XUMO</em>).</p><p>Another point that we liked from his article is his statement on the AVOD/Digitial TV industry expecting to reduce average ad time. As we’ve seen this year with Peacock TV’s launch, <strong>the industry is definitely moving towards this direction!</strong></p><p>Follow him on Twitter: <a href="https://twitter.com/petersontee">https://twitter.com/petersontee</a></p><p>Reach out to him over email: <a href="/service/mailto:tim@digiday.com">tim@digiday.com</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9236b35baeda" width="1" height="1" alt=""><hr><p><a href="https://medium.com/this-week-in-adtech/top-journalists-in-adtech-mini-series-tim-peterson-9236b35baeda">Top Journalists in AdTech Mini-Series: Tim Peterson</a> was originally published in <a href="https://medium.com/this-week-in-adtech">This Week In AdTech</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Top Journalists in AdTech Mini-Series: Patrick Kulp]]></title>
            <link>https://medium.com/this-week-in-adtech/top-journalists-in-adtech-mini-series-patrick-kulp-f203aa13f22?source=rss----2f232efba4af---4</link>
            <guid isPermaLink="false">https://medium.com/p/f203aa13f22</guid>
            <category><![CDATA[branding]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[advertising]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[marketing]]></category>
            <dc:creator><![CDATA[Shane Austrie]]></dc:creator>
            <pubDate>Mon, 10 Aug 2020 17:13:26 GMT</pubDate>
            <atom:updated>2020-08-10T17:13:25.792Z</atom:updated>
            <content:encoded><![CDATA[<h4><strong>Patrick Kulp</strong> writer at Adweek…</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*956NqsjkKHPCSIJpzl_WDA.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@climatereality?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">The Climate Reality Project</a> on <a href="https://unsplash.com/s/photos/journalist?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><blockquote>The only Mini-Series doing quick, two-minute highlights of some of the best digital writers in Marketing, Branding, and AdTech!</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/1*rTgUmhD8N3-k4gGNquhvRQ.jpeg" /></figure><h4>Overview</h4><p>Patrick Kulp is currently a writer at Adweek. He covers stories ranging from AI startups to 21st-century consumer trends.</p><h4>Background</h4><p>Patrick attended the <strong>University of California: Santa Barbara</strong> for undergrad. While at <strong>UCSB</strong>, he majored in Political Science.</p><p>He has shown an interest in covering current events ever since his college days. During his late college years, he would cover positive social protesting— going back even to the Berkeley/Oakland protests of 2014.</p><p>Patrick held several positions during his college years: from being an Editorial intern for both <strong><em>Noozhawm</em></strong> and the <strong><em>Pacific Coast Business Times</em></strong> to being the Digital Managing Editor for UCSB’s 90-year-old news publication —<strong><em> The Daily Nexus.</em></strong></p><p>He joined <strong><em>Mashable</em></strong> in 2014, where he worked his way up to the position of <strong><em>Business Reporter</em></strong>. During his three — <em>nearly four</em> — years at <strong><em>Mashable</em></strong>, he covered topics such as:</p><ul><li><strong><em>Business:</em></strong> The <strong>WeWork</strong> founders and the <strong>YouTube</strong> boycotts</li><li><strong><em>Tech:</em></strong> The <strong>Winklevoss</strong> twins and <strong>Bitcoin’s IPO</strong></li></ul><p>One notable prediction that Patrick made back in 2017 was the rise of <strong><em>Amazon Alexa</em></strong>. Though the Amazon speaker was seen as a toy back in 2016/2017, now you can find an Alexa in practically every household — over 70% of smart speakers are now Alexa-enabled.</p><blockquote><em>With the COVID-lockdown in the US now becoming a permanent way of life, advertisers have recently started looking into Alexa as a way to offer personalized, branded, marketing experiences.</em></blockquote><h4>Now</h4><p>Since <strong><em>Mashable</em></strong>, Patrick has been writing for <strong><em>Adweek</em></strong> for nearly three years!</p><p>We’ve been following his writing for several years since he joined <strong><em>Adweek</em></strong>. He’s been writing about 4 or 5 articles per week, bringing some of his past experience and knowledge at <strong><em>Mashable</em></strong> to his current writing at <strong><em>Adweek</em></strong>.</p><p>Some of the top articles he has written over the years at <strong><em>Adweek </em></strong>include:</p><ul><li><a href="https://www.adweek.com/digital/why-weather-data-is-the-hottest-new-commodity-for-brands-and-businesses/"><em>Why Weather Data Is the Hottest New Commodity for Brands and Businesses</em></a></li><li><a href="https://www.adweek.com/creativity/we-asked-todays-most-advanced-storytelling-ai-to-co-write-a-story-about-itself/"><em>We Asked Today’s Most Advanced Storytelling AI to Co-Write a Story About Itself</em></a></li><li><a href="https://www.adweek.com/digital/the-new-media-wars/"><em>AT&amp;T and Verizon Are Making Very Different Bets on What 5G Will Mean for Consumers and Content</em></a></li></ul><p>His recent writing has been about the advancement of AI in the entertainment sector. Some of the topics of discussion have been <strong><em>Intel</em></strong>, <strong><em>Nvidia</em></strong>, <strong><em>Warner Bros</em></strong>, <strong><em>Open AI</em></strong>, and many other recognizable names in the industry.</p><h4>Our Fan Favorite</h4><p>Minus the usual Adweek paywall, our fan-favorite from him is his recent article on <a href="https://www.adweek.com/creativity/experimental-new-ai-can-autocomplete-images-with-the-same-technology-as-predictive-text/">AI that can autocomplete images</a>.</p><p>We love his in-depth look into the exact AI technology implemented and how it’s usage is no-longer text-based, but also image-based!</p><p>Over recent years, GAN-like AI models have started piquing the interest of the marketing industry for creating new and unique visual content — a good example being memes!</p><p><em>One thing that we wanted to note is that this same technology is also being used to help Linear TV show the best ads in real-time for live TV.</em></p><p>Follow him on Twitter: <a href="http://twitter.com/patrickkulp">http://twitter.com/patrickkulp</a></p><p>Reach out to him over email: <a href="/service/mailto:patrick.kulp@adweek.com">patrick.kulp@adweek.com</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f203aa13f22" width="1" height="1" alt=""><hr><p><a href="https://medium.com/this-week-in-adtech/top-journalists-in-adtech-mini-series-patrick-kulp-f203aa13f22">Top Journalists in AdTech Mini-Series: Patrick Kulp</a> was originally published in <a href="https://medium.com/this-week-in-adtech">This Week In AdTech</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Ads: Entertain or Lose the Game]]></title>
            <link>https://medium.com/this-week-in-adtech/ads-entertain-or-lose-the-game-dd3138806d50?source=rss----2f232efba4af---4</link>
            <guid isPermaLink="false">https://medium.com/p/dd3138806d50</guid>
            <category><![CDATA[marketing]]></category>
            <category><![CDATA[adtech]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[streaming]]></category>
            <category><![CDATA[advertising]]></category>
            <dc:creator><![CDATA[Lacey Cohen]]></dc:creator>
            <pubDate>Tue, 28 Jul 2020 17:12:51 GMT</pubDate>
            <atom:updated>2020-07-28T16:31:27.188Z</atom:updated>
            <content:encoded><![CDATA[<p>To the average GenZer and/or millennial, ads are tedious, seen as background noise, and provide a break in their TV show to scroll through Instagram for a few minutes. In fact, many younger people have installed AdBlocker apps on their web browsers to eliminate ads completely — the need for a social media break is not as strong as it maybe once was! Advertisements serve as a strong source of revenue for companies, and they fill the commercial gaps in TV programming that exist anyway. You may as well optimize them so that your users don’t grow entirely bored. Many streaming services provide a no ad subscription option, but if yours doesn’t, then you might need to think a little harder about how to ensure these ads aren’t a deterrent.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*WabuSmUqju6H8h8A" /><figcaption>Photo by<a href="https://unsplash.com/@mbaumi?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> Mika Baumeister</a> on<a href="https://unsplash.com/s/photos/streaming?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> Unsplash</a></figcaption></figure><p>Remember in high school English class when you learned about Logos, Ethos, and Pathos? Well, it turns out that may have been a pretty important lesson after all. Ads that work are ads that evoke emotion in your viewers. Take the Doritos ads during the Super Bowl for example: funny, lighthearted, and are almost always guaranteed to make you smile. Or the Extra Gum ads? Where the boyfriend makes the girlfriend origami birds out of the gum wrappers, or the father watches his daughter go off to college? Those are gold star examples; <em>seek those out, and </em><strong><em>use them</em></strong>.</p><p>Additionally, your users are used to multi-tasking like crazy, if you don’t capture their attention within a few seconds, then you run the risk of ruining the ad experience. A strong attention getter is crucial for an enjoyable ad experience; your users are not going to wait around for your ad to get better, they will simply skip over to the next tab while they wait for it to play out.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*oeepJWu8ohXqIWRZ" /><figcaption>Photo by<a href="https://unsplash.com/@jmkong?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> Aaron Sebastian</a> on<a href="https://unsplash.com/s/photos/times-square?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> Unsplash</a></figcaption></figure><p>These two factors combined will be enough. What that means is, if your ads make viewers laugh, or they tug at the heart strings, and they do it fast enough to keep your viewers’ attention, it won’t really matter as much whether the product itself is one that your audience would be interested in.</p><p>Showcasing an entertaining ad is imperative in engaging your audience. Long gone are the ads where a salesperson speaks monotonically at the camera, forcing the audience to listen while they wait for their show to come back. Nowadays, if your audience is bored, they <em>will</em> leave. All you need to do to avoid that is grab their attention (fast!), and establish an entertaining story. With those two factors in play, your audience is sure to enjoy their ad journey.</p><h3><a href="https://medium.com/@diana_10807">Diana Gaspar</a> is a co-creator of this article.</h3><p><em>If you’d like to connect with us, feel free to reach out on LinkedIn:</em></p><p><a href="https://www.linkedin.com/in/lacey-cohen/">https://www.linkedin.com/in/lacey-cohen</a>/</p><p><a href="https://www.linkedin.com/in/dianapgaspar/">https://www.linkedin.com/in/dianapgaspar/</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dd3138806d50" width="1" height="1" alt=""><hr><p><a href="https://medium.com/this-week-in-adtech/ads-entertain-or-lose-the-game-dd3138806d50">Ads: Entertain or Lose the Game</a> was originally published in <a href="https://medium.com/this-week-in-adtech">This Week In AdTech</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Contextual Brand Safety]]></title>
            <link>https://medium.com/this-week-in-adtech/contextual-brand-safety-98e6380c5abb?source=rss----2f232efba4af---4</link>
            <guid isPermaLink="false">https://medium.com/p/98e6380c5abb</guid>
            <category><![CDATA[streaming]]></category>
            <category><![CDATA[advertising]]></category>
            <category><![CDATA[adtech]]></category>
            <category><![CDATA[branding]]></category>
            <category><![CDATA[marketing]]></category>
            <dc:creator><![CDATA[Shane Austrie]]></dc:creator>
            <pubDate>Tue, 28 Jul 2020 00:51:48 GMT</pubDate>
            <atom:updated>2020-08-16T06:56:41.120Z</atom:updated>
            <content:encoded><![CDATA[<h4>Why brands are scared to advertise on your platform in 2020…</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8muSEFd497mkj0VwJgIWgw.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@kristianegelund?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Kristian Egelund</a> on <a href="https://unsplash.com/s/photos/brand?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><p>In 2020, we already know that ads are supposed to be contextually relevant. Long gone are the days where ads simply needed to target the right audience — regardless of time/location. With ever-decreasing conversion rates, <em>we now realize that ads need to target the right audience, </em><strong><em>at the right moment</em></strong><em>.</em></p><p>With the increased social awareness of <strong>Gen Z</strong>, brands now also care about the connotations that are being implicitly associated with them while advertising with your platform. <em>More specifically, </em><strong><em>brands care about the</em> <em>content your users are viewing before, during, and after their ad is seen</em>.</strong></p><blockquote>“Brand safety is a set of measures that aim to protect the image and reputation of brands from the negative or damaging influence of questionable or inappropriate content when advertising online.” <strong>— Internet Advertising Bureau (<em>IAB</em>)</strong></blockquote><blockquote>If you think brands don’t care about brand safety, just look at the immediate action that KFC took after facing a large backlash in the UK for a “<em>finger-licking good</em>” ad during COVID</blockquote><p>Even YouTube is dealing with this issue. Brands have been complaining a lot more in 2020 due to their YouTube ads being shown during extremist videos (<em>e.g. political, racial, etc.</em>). However, because YouTube is owned by Google, YouTube can get away with it — <strong><em>your platform can’t.</em></strong></p><p>Advertisers pulled their ads from <em>YouTube — literally the largest streaming site — </em>for months, due to YouTube showing their ads next to non-brand-safe content. YouTube lost hundreds of millions of dollars due to that small protest. <strong><em>How does your platform adapt? What would brands do if your platform ends up in a brand safety incident?</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RBWjpxnLlxwJG9OjZlAwow.png" /><figcaption>Credit Tom Fishburn</figcaption></figure><p><strong><em>If you’re looking for articles on how to actually achieve brand safety, check out these two articles:</em></strong></p><ul><li>Simplified Contextual Video Ads Technical Paper: <a href="https://medium.com/this-week-in-adtech/simplified-technical-paper-ml-based-contextual-video-ads-on-demand-live-ee8f9b661059">https://medium.com/this-week-in-adtech/simplified-technical-paper-ml-based-contextual-video-ads-on-demand-live-ee8f9b661059</a></li><li>Simplified ML-powered Brand Safety Technical Paper: <a href="https://medium.com/this-week-in-adtech/using-computer-vision-nlp-for-brand-safety-9872a93000c8">https://medium.com/this-week-in-adtech/using-computer-vision-nlp-for-brand-safety-9872a93000c8</a></li></ul><p>Brand safety isn’t vertical or platform-specific. Brand safety affects online news publications, music streaming, video streaming, and even podcasts (<em>now that podcasts are mainstream</em>).</p><p>Brands care about the genre of your visual, musical, and/or written content. They care about if the content features explicit language (<em>i.e. swearing</em>), hate speech, as well as if the scenes, writing, or lyrics before and after the ad create negative emotions for the user.</p><p>In a large study by Cheq AI, results indicated that <strong>intent to buy dropped by 200% after watching inappropriate content</strong>, as well as the brand quality is perceived <em>7 times worst</em>. Additionally, consumers were 450% more likely to feel that the brand did not care about them.</p><p>After Cheq AI’s study, there have been a plethora of other studies showing the dramatic effects of context as it relates to brand safety — and brands are starting to realize this too. From Pepsi to L’Oreal, 2019/2020 has been a wake-up call in the marketing world. Topics, words, and phrases that used to be brand-safe can instantly change due to new current events — <strong>and brands expect advertising platforms to be able to instantaneously adapt or be left behind.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=98e6380c5abb" width="1" height="1" alt=""><hr><p><a href="https://medium.com/this-week-in-adtech/contextual-brand-safety-98e6380c5abb">Contextual Brand Safety</a> was originally published in <a href="https://medium.com/this-week-in-adtech">This Week In AdTech</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Using Computer Vision & NLP For Brand Safety]]></title>
            <link>https://medium.com/this-week-in-adtech/using-computer-vision-nlp-for-brand-safety-9872a93000c8?source=rss----2f232efba4af---4</link>
            <guid isPermaLink="false">https://medium.com/p/9872a93000c8</guid>
            <category><![CDATA[advertising]]></category>
            <category><![CDATA[adtech-industry]]></category>
            <category><![CDATA[branding]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[enterprise-software]]></category>
            <dc:creator><![CDATA[KEONI MURRAY]]></dc:creator>
            <pubDate>Mon, 27 Jul 2020 20:25:33 GMT</pubDate>
            <atom:updated>2020-07-28T02:40:23.165Z</atom:updated>
            <content:encoded><![CDATA[<h4>Another article in the simplified technical paper series…</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TnDHNe2nV0Xbot0lvkAY-A.png" /></figure><p>As advertisers have more platforms than ever to reach their desired audience, brand safety has steadily become an increasing concern.</p><p>During my time as a consultant, I have had the opportunity to work on multiple projects where my team and I were tasked with building customized systems designed for filtering content that brands would not want to be affiliated with — <em>as well as, making the system smart/adaptable enough to work for any brand.</em></p><blockquote><strong><em>We’ve recently adapted this system to work on live content as well. However, this article primarily focuses on the version that was centered around on-demand content.</em></strong></blockquote><p>This article is a simplified overview of the internal white paper Silicon Valley Consulting has written (<em>and is constantly improving</em>). Given the length and complexity of this system, this article will be part one of a multi-part, simplified series.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QCaM1WUQBfxcDStmXBXXsA.png" /></figure><p>The above image is an overview of what the architecture would look like in a production environment. <strong>Within this article, we will cover a brief overview of each service in this system.</strong></p><h3>Pre-Processing</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/400/1*FQL8NMdNlQ_VSJetISIOzw.gif" /></figure><p>The pre-processing stage of this overall system takes in the several attributes of a video as input:</p><ul><li><strong>Visual data</strong></li><li><strong>Audio data</strong></li><li><strong>Metadata</strong> (<em>actors, genre, description, etc.</em>)</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JnVZvhQXJb_JG7thB0TJEg.png" /></figure><p>During this stage, a video is grabbed from a file system (<em>whether it’s AWS S3 or Hadoop</em>) then sent over to the pre-processing service. The pre-processing service is designed to simplify the three attributes of a video for easier ML analysis later.</p><p>For the visual data input, the goal is to reduce the amount of redundant/unnecessary data, as well as convert this data into suitable numerical tensors.</p><p>First, we take all frames associated with a scene (<em>there’s an ML-model that determines what’s a “scene”</em>), then having a secondary model group frames in a scene by similarity. Each group is then analyzed by a third ML model (<em>hint: this model is based on GANs</em>), that attempts to find the three best groups to represent the beginning, middle, and end of a scene.</p><p>The groups are then “simplified” and stitched together, in order to recreate a simplification of an entire scene. Each simplified scene is then converted into a tensor (<em>with every frame being a matrix</em>) and is then sent to the rest of the classifiers in the brand awareness system to be analyzed.</p><p>For the audio data, we simply use the transcript associated with the video. We break up the transcript based on the scene, and then we do basic NLP pre-processing such as removing filler words, punctuations, stemming, etc.</p><p>For the metadata, we use a combination of one-hot encoding, label encoding, bucketing, and a few proprietary methods in order to best utilize the metadata.</p><h3>Building Our Dataset</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*NlEozMliA1jsGqlGEn_FkA.jpeg" /></figure><p>For building the dataset to train all of our models, we used a combination of internal data (<em>provided by our partners</em>) and public data.</p><p>Some of our public data for the video pre-processing system is trained from sources such as:</p><ol><li>Facebook.com</li><li>Reddit.com</li><li>Twitter.com</li><li>Unsplash.com</li></ol><p>Building these datasets included designing a data warehouse and pipeline that would be able to collect the data we need from a variety of internal sources; for this, we used Airflow and Google Big Query.</p><h3>Classifiers</h3><p>The classifiers after the pre-processing step are designed to classify and detect scenes that fall into the following categories:</p><ol><li>Violence</li><li>Crime/Illegal</li><li>Alcohol/Tobacco/Drugs</li><li>Sexual content &amp; Nudity</li><li>Profanity, Vulgar, Gore</li><li>Hate</li><li>Disaster</li><li>Cyber Crime/Malware</li><li>Illness</li><li><em>*And a few other categories that I can not disclose*</em></li></ol><p>What’s listed above are common categories that most brands do not want to be associated with. <strong><em>However, keep in mind, brands will usually have their own custom categories of content that they do not want to be associate with — which is why it’s important to make this system adaptable</em></strong>.</p><p>Each scene will be scored against each category (<em>collectively</em> <em>creating a numerical vector</em>), then simplified into a single brand-safety score for a scene. The scores for every scene can be collectively simplified further, in order to represent the scoring of an entire video.</p><h4>Hate Speech Classifier</h4><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fgiphy.com%2Fembed%2FOAHvbQLumpmta%2Ftwitter%2Fiframe&amp;display_name=Giphy&amp;url=https%3A%2F%2Fmedia.giphy.com%2Fmedia%2FOAHvbQLumpmta%2Fgiphy.gif&amp;image=https%3A%2F%2Fi.giphy.com%2Fmedia%2FOAHvbQLumpmta%2Fgiphy.gif&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=giphy" width="435" height="326" frameborder="0" scrolling="no"><a href="https://medium.com/media/d3d1219c151913c229a8ef07e8e48060/href">https://medium.com/media/d3d1219c151913c229a8ef07e8e48060/href</a></iframe><blockquote>Let’s start with the definition of hate speech: “Hate speech is any communication that disparages a target group of people based on some characteristics such as race, color, ethnicity, gender, sexual orientation, nationality, religion, or other characteristics”</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/398/1*-zv7LtHgDWYx6L8c78SS8Q.png" /></figure><p>The hate speech classifier uses NLP (<em>Natural Language Processing</em>) to check if audio data from a scene contains hate speech. Verbal and written content is numerically converted via a pre-trained <strong><em>GloVe</em></strong> vectorizer, then fed through a custom NLP model.</p><h4>Training Overview</h4><p>For on-demand content, we simply used the transcript. However, in order to make this stage work for live content (<em>not just on-demand content which already has transcripts</em>), we used a combination of text from social media and the results from a speech to text service (<em>for converting audio from a live video stream into text</em>). After building the dataset, we included a new DAG in Airflow for cleaning the text in the dataset.</p><p>Once the dataset was cleaned, we used what is called a pre-trained <strong><em>GloVe</em></strong> embedding. This simply associated a word in a vector space of N-length.</p><p>After the embedding phase, we used a custom Neural Network architecture. We used an <strong><em>LSTM</em></strong> (<em>Long Short Term Memory</em>) model for predicting the next word(<em>s</em>) (<em>we also have a modified version for detecting the next several sentences</em>), followed by models for detecting explicit and implicit meanings.</p><p>Originally, we simply used an LSTM model to do sequence classification (<em>i.e. hate speech or not hate speech</em>). However, in order to handle live content, we need to predict sentences in advance. This advance prediction is necessary in order to give the other systems outside of brand safety (<em>e.g. video ad recommendation models</em>) the time necessary to select the best ad for this scene.</p><blockquote>Note — In a more up to date version of this paper, we’ve started taking advantage of the implicit signals in audio, rather than just text. We’ve been able to successfully utilize implicit signals such as tone of voice and sarcasm.</blockquote><h3>Hate Logo Detector Overview</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/1*EoMQEKkJMA3wO0xtHoKIjA.gif" /></figure><p>The hate logo detector simply detects hate logos in a video scene.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/348/1*PItlTIwld1uvvNMncDzwUA.png" /></figure><h4>Training overview</h4><p>The first step in training was collecting the data. During this stage, we created new DAG’s in Airflow to collect hate logos from multiple pre-defined websites across the web (<em>this was surprisingly easy</em>).</p><p>During the training stage, we used an internal algorithm to validate the quality of the images (<em>e.g make sure they are not blurry, etc</em>.), then proceeded to convert each image into numerical values.</p><p>The ML model we used for training is <strong><em>SSD</em></strong> (<em>Single Shot Multibox Detector</em>). We choose <strong><em>SSD</em></strong> over <strong><em>YOLO</em></strong><em> </em>(<em>which is commonly used for object detection</em>) due to the speed and improved accuracy of <strong><em>SSD</em></strong> for this use case. <strong><em>We are experimenting with the SeNet model which is known for improving accuracy by over 25%. </em></strong>We used a similar technique to train the other object detectors as well.</p><h3>Scene Detection Service</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/328/1*oFRUfRDXMscOuseqNQKeog.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PZ_kmrO9csJojBSQ5q1GbA.png" /></figure><p>The Mature Scene Detection System represents multiple classifiers (<em>over 70 to be exact</em>). In order to make this part of the system work for live content, we had swap out the classifiers we were originally using, for lighter weight models.</p><p>For the simplicity of this article, we only show the most individually important/common classifiers within the Mature Scene Detection service. These six classifiers are moderately straight forward, so I won’t go into details about them. However, in future articles, I will go into more detail on what some of the more unique classifiers in this service do.</p><p>The numerical output of all classifiers are used in a vector of predetermined size (<em>based on the number of classifiers</em>).</p><h3>Merge Score</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sy9GXD7neQkzMdHX0ivwtQ.png" /></figure><p>The Merge Score service in this overall system collects the output of the Hate Speech Detector, the Hate Logo Detector, as well as the Mature Scene Detector.</p><p>Each input, <em>if a vector</em>, gets converted into a single numerical value — turning the overall input from a matrix to an output that is a vector. Additionally, each numerical value in the vector is then scaled so that the next stage is not affected by the original disparity in scalar ranges.</p><h3>Final Score</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*cqt7ICI47k_ZHclM7eMhGg.png" /></figure><p>The Final Score service is an MLP (<em>i.e. Multi-Layer Perception</em>) model that weighs the various inputs in the vector, in order to determine the brand-safe level of a scene.</p><p><strong>The MLP model produces a floating-point score between -1 and 1.</strong></p><p>A score of <em>1</em> represents a scene that is perfectly brand-safe — a scene with zero negative implicit connotations<em> </em>or explicit content.</p><p>A score of <em>0</em> represents a scene that is moderately brand-safe — a scene with zero explicit content but may have some possibly negative connations.</p><p>A score of <em>-1</em> represents a scene that absolutely not brand-safe — a scene that has explicit content that most people would not find acceptable.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*puA4kbaH5XsFGdTl_K7-Ew.png" /></figure><p>Once these scores are collected, they are merged together into a JSON object, then cached into Redis for quick and dynamic use by the video ad selection/recommender system.</p><p>This is a really complex system to build and this article does not do enough justice to all of the details that make this overall system work. However, hopefully, this gives your team an idea of how you should approach architecting an ML system for filtering video content that is not brand-safe.</p><p>One thing that should be noted is that several key aspects are left out of this article:</p><ul><li><strong>The exact models, layer architecture, and hyperparameters (<em>e.g. loss functions</em>) used.</strong></li><li><strong>Additional classifiers that are necessary in order to ensure <em>high brand safety.</em></strong></li><li><strong>The exact infrastructure needed to reduce latency — which is <em>highly</em> necessary if this is to be used in production, especially for live content.</strong></li><li><strong>Technical fall back strategies needed in order to handle edge cases that <em>WILL</em> occur <em>ONLY</em> in production.</strong></li></ul><p>Linkedin — <a href="https://www.linkedin.com/in/keonimurray/">https://www.linkedin.com/in/keonimurray/</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9872a93000c8" width="1" height="1" alt=""><hr><p><a href="https://medium.com/this-week-in-adtech/using-computer-vision-nlp-for-brand-safety-9872a93000c8">Using Computer Vision &amp; NLP For Brand Safety</a> was originally published in <a href="https://medium.com/this-week-in-adtech">This Week In AdTech</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Advertising After the Pause]]></title>
            <link>https://medium.com/this-week-in-adtech/advertising-after-the-pause-bcd1429c6918?source=rss----2f232efba4af---4</link>
            <guid isPermaLink="false">https://medium.com/p/bcd1429c6918</guid>
            <category><![CDATA[streaming]]></category>
            <category><![CDATA[streaming-service]]></category>
            <category><![CDATA[advertising]]></category>
            <category><![CDATA[adtech]]></category>
            <category><![CDATA[marketing]]></category>
            <dc:creator><![CDATA[Franklin Feng]]></dc:creator>
            <pubDate>Fri, 24 Jul 2020 13:47:28 GMT</pubDate>
            <atom:updated>2020-07-23T18:40:45.478Z</atom:updated>
            <content:encoded><![CDATA[<p><em>A Savvy Approach for Streaming Platforms to Partner Revenue Generation With a Seamless Viewing Experience</em></p><p>Advertisements. They seem to be everywhere in our society, but perhaps none more frequently than on streaming platforms. A typical program that is 60 minutes in length would average out to about <em>10 minutes in ads</em>. As viewers, we are often frustrated by persistent disruptions to our content and advertising that intrudes into our experience. Here’s a question: <em>how often do you pause a stream to grab something to eat or to use the restroom?</em> Data shows it’s common, and streaming platforms have taken notice. Introducing a new method to displaying ads: <strong>Pause ads</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*_mYZo0bOB8V4deeF" /></figure><p>Let’s briefly explain what pause ads are and how they offer a fluid viewing experience through minimizing intrusion:</p><p>Pause ads eliminate the format of inserting commercials before, during, or after a video by allowing you to enjoy an <em>uninterrupted play-through</em> of their content. <strong>Only when the stream is paused would a sponsor display its advertising</strong>, filling up the stagnant screen until you decide to press Play on the video again.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*N2kjRtJMyyLJdiYc" /></figure><p>Hulu was one of the first streaming platforms to incorporate pause ads, partnering with Coca-Cola and Charmin to deliver “subtle and relevant” advertising while you take a quick break from the content. <em>The goal is to reduce obtrusion into the streaming content while implementing a conscious product that appears in common activities you would do when paused.</em> This hopefully increases the relevance of the advertised brand when you’re in the store shopping next time.</p><p>Strategists see pause ads to be a viable pathway for generating advertising revenue. Hulu hopes to see pause ads rake in <strong>“50 percent of Hulu’s ad revenue within the next three years”</strong> (Anthony Ha), while competitors like Netflix are taking note of pausing trends in their consumers. The streaming service recently found that the “peak pause time across the nation is at 7:29 p.m” (Shaymus McLaughlin).</p><p>So what is the future of advertising on streaming platforms? It very well may be utilizing the valuable time when you hit the Pause button.</p><p><strong>Resources:</strong></p><p><em>Ha, Anthony. “Hulu Announces a New Ad Unit That Appears When You Pause.” TechCrunch, TechCrunch, 31 Jan. 2019, techcrunch.com/2019/01/31/hulu-pause-ad/.</em></p><p><em>McLaughlin, Shaymus. “Netflix Remembers Every Time You Pause a Show (and a Lot of Other Info).” Bring Me The News, 8 Mar. 2018, bringmethenews.com/news/netflix-remembers-every-time-you-pause-a-show-and-its-a-lot-on-halloween.</em></p><p><a href="https://medium.com/@margo_26957"><strong>Margo Boxer</strong></a><strong> contributed to the reporting of this article.</strong></p><p><em>If you would like to discuss further, feel free to contact us on LinkedIn:</em></p><p><a href="http://www.linkedin.com/in/franklinfeng">https://www.linkedin.com/in/franklinfeng</a></p><p><a href="https://www.linkedin.com/in/margo-boxer-680a24195">https://www.linkedin.com/in/margo-boxer-680a24195</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bcd1429c6918" width="1" height="1" alt=""><hr><p><a href="https://medium.com/this-week-in-adtech/advertising-after-the-pause-bcd1429c6918">Advertising After the Pause</a> was originally published in <a href="https://medium.com/this-week-in-adtech">This Week In AdTech</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Expanding your streaming service while maintaining your niche]]></title>
            <link>https://medium.com/this-week-in-adtech/expanding-your-streaming-service-while-maintaining-your-niche-cc42e769e057?source=rss----2f232efba4af---4</link>
            <guid isPermaLink="false">https://medium.com/p/cc42e769e057</guid>
            <category><![CDATA[business-development]]></category>
            <category><![CDATA[advertising]]></category>
            <category><![CDATA[streaming]]></category>
            <category><![CDATA[user-experience]]></category>
            <category><![CDATA[business-strategy]]></category>
            <dc:creator><![CDATA[Cindy Syren]]></dc:creator>
            <pubDate>Fri, 24 Jul 2020 13:46:51 GMT</pubDate>
            <atom:updated>2020-07-27T22:06:53.240Z</atom:updated>
            <content:encoded><![CDATA[<h3>Keeping your customers in mind when you advertise</h3><p>We are living in an age of rapid online growth as the internet expands to eclipse traditional technology. At the forefront of this conversation is the transition of the entertainment industry as it shifts away from television, blu ray, and radio. As competition between streaming services grows steeper, companies are brainstorming ways to set themselves apart from their competitors.</p><p>Streaming services now differentiate themselves based on content and user experience. Netflix, Hulu, Amazon Prime and HBO — the ‘Big Four’ of online streaming — offer a wide variety of content across many categories. Then, there are genre-specific streaming services, which have recently broken through the long standing “Big Four” barrier of entry to the upper echelons of the world of streaming. Services like Disney+ and CrunchyRoll, which specialize specifically in family entertainment and anime/manga, respectively, have a <strong>clear-cut target audience</strong> and a <strong>distinct brand</strong>, which establishes them as serious competitors to other major streaming services.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*onSaV6Wi5uh32iLJ" /><figcaption>Graphic: James Bareham/Polygon</figcaption></figure><p>Disney pulled most of its content from other streaming services when it created Disney+, drawing subscribers away from Netflix and Hulu. Netflix itself has begun producing “Netflix Originals” — shows and movies that are only available to Netflix subscribers. Just as streaming services broke away from traditional television, companies are slowly scaling back their business models to tap into niche markets. Content-specific streaming services are undoubtedly the near-future of the streaming wars.</p><p>At this time, it is critical for a streaming company to maximize revenue if they wish to remain ‘in the game’. For smaller, niche streaming services it is important to find the optimal balance between ad revenue and subscription costs. Newer streaming services should seize this transitionary period in the industry to explore novel advertising opportunities which could make all the difference in the growth of a company.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*zYXMyCvrY6a-PtM6" /></figure><p>When it comes to advertisements, companies <strong>must</strong> focus on maximizing user engagement and experience. User’s want to see ads that <em>matter to them. </em>Fortunately, if you are a genre-specific streaming service you are in luck! Genre-specific streaming services <em>know </em>their target audience. From there, it is about fine tuning an individual’s experience. Show your users ads that they <a href="https://medium.com/this-week-in-adtech/ads-can-be-annoying-but-yours-dont-have-to-be-97d176a2c441">won’t be annoyed at</a>. Show your users ads that they will <em>actually want to click on.</em></p><p>If your company is breaking into the realm of streaming with a niche target audience, it is <strong>inevitable</strong> you will need to leverage ads to generate revenue. Before a company can worry about competing among the big players, they must first set themselves apart from the smaller ones. Advertising may seem like just another detail for a streaming service, but a good advertising strategy will go a long way.</p><p><em>Feel free to connect with me on </em><a href="https://www.linkedin.com/in/cynthia-syren/"><em>LinkedIn</em></a><em> to continue the conversation.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cc42e769e057" width="1" height="1" alt=""><hr><p><a href="https://medium.com/this-week-in-adtech/expanding-your-streaming-service-while-maintaining-your-niche-cc42e769e057">Expanding your streaming service while maintaining your niche</a> was originally published in <a href="https://medium.com/this-week-in-adtech">This Week In AdTech</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Stop Trying To Control Your Users]]></title>
            <link>https://medium.com/this-week-in-adtech/stop-trying-to-control-your-users-a070c403cf20?source=rss----2f232efba4af---4</link>
            <guid isPermaLink="false">https://medium.com/p/a070c403cf20</guid>
            <category><![CDATA[streaming]]></category>
            <category><![CDATA[marketing]]></category>
            <category><![CDATA[adtech]]></category>
            <category><![CDATA[streaming-service]]></category>
            <category><![CDATA[advertising]]></category>
            <dc:creator><![CDATA[Shane Austrie]]></dc:creator>
            <pubDate>Wed, 22 Jul 2020 04:01:09 GMT</pubDate>
            <atom:updated>2020-07-22T04:01:09.459Z</atom:updated>
            <content:encoded><![CDATA[<h4>It may have been acceptable in the ’90s and early ’00s, but not in 2020 and beyond…</h4><p><strong><em>Note: These tips are written in the lens of the streaming world, but the underlying message is universal…</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7X384bWId5kRZ3m-FmH6MQ.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@mq1?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Michał Lis</a> on <a href="https://unsplash.com/s/photos/walking-away-from-tv?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><blockquote>If your users are sharing streaming accounts, <strong><em>then let them</em></strong>.</blockquote><p>Trying to force them to pay individually may do more damage than good (<em>e.g. revenue, market share, etc.</em>). <strong><em>Instead, focus on how to maximize the amount of profit based on their current behavior.</em></strong></p><p>If it’s common for 3 users to share one $9.99 streaming account, then that might be the market telling you that your streaming service is only worth $3.33 per account (<em>whether you accept that fact is up to you…</em>).</p><blockquote>However, instead of changing your prices to $3.33 and enforcing a one user per account policy — what if you simply increased the number of *<strong>good</strong>* ads shown to the users sharing that one account (if it’s a paid account that still has ads like Hulu and Quibi).</blockquote><p>Hulu and Netflix fight this problem by allowing multiple users per subscription, but simply capping the max number of users and/or the number of concurrent streams happening on an account at once.</p><p>With Hulu specifically, since they have a subscription plan that is ad-supported, they are winning either way:</p><ul><li><strong>They’re making over 10x the revenue (<em>in comparison to the annual revenue of a subscription</em>) by showing ads to all users on the account.</strong></li><li><strong>Additionally, even when no-one is using the shared account during some months, the end-user is still paying Hulu automatically since it’s a monthly paid subscription.</strong></li></ul><p>If you try controlling your users, <em>going too much against the grain</em>, you’ll end up losing your users — just look at <strong><em>Snapchat</em></strong>, a once-beloved app to all Gen Z…</p><p><strong>If you would like to discuss more on this, feel free to contact me on LinkedIn: </strong><a href="https://www.linkedin.com/in/shaneaustrie"><strong><em>https://www.linkedin.com/in/shaneaustrie</em></strong></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a070c403cf20" width="1" height="1" alt=""><hr><p><a href="https://medium.com/this-week-in-adtech/stop-trying-to-control-your-users-a070c403cf20">Stop Trying To Control Your Users</a> was originally published in <a href="https://medium.com/this-week-in-adtech">This Week In AdTech</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Simplified Technical Paper: ML-based Contextual Video Ads (on-demand & live)]]></title>
            <link>https://medium.com/this-week-in-adtech/simplified-technical-paper-ml-based-contextual-video-ads-on-demand-live-ee8f9b661059?source=rss----2f232efba4af---4</link>
            <guid isPermaLink="false">https://medium.com/p/ee8f9b661059</guid>
            <category><![CDATA[advertising]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[adtech]]></category>
            <category><![CDATA[marketing]]></category>
            <category><![CDATA[video-streaming-service]]></category>
            <dc:creator><![CDATA[Shane Austrie]]></dc:creator>
            <pubDate>Wed, 22 Jul 2020 02:33:21 GMT</pubDate>
            <atom:updated>2020-07-25T00:12:15.983Z</atom:updated>
            <content:encoded><![CDATA[<h4>How to do contextual video ads in live &amp; on-demand content streaming (2020/2021)…</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qJD9r5UsS1E9Oj6CfMHDOQ.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@morningbrew?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Morning Brew</a> on <a href="https://unsplash.com/s/photos/hulu-hbo?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><p>This Machine Learning system is based on what’s used at streaming companies/services like <em>CBS Interactive </em>and several other top ad-supported streaming platforms.</p><p><strong>Companies like CBS Interactive (<em>as well as their sibling companies</em>) have shown an over 14% gain in ad engagement when using the predecessor of this system.</strong></p><blockquote>This system focuses on ensuring that video ads are relevant to the last scene and/or frame.</blockquote><p>Additionally, this system can be used for both <strong>video-on-demand </strong>(<em>VOD</em>) content, as well as <strong>live-stream content </strong>(<em>e.g. live baseball, breaking news, etc.</em>).</p><h3>Table of Contents:</h3><h4>* Offline Video Ad Pre-Processing</h4><h4>* Ensemble Hybrid Recommendation System</h4><h4>* On-Demand &amp; Live Video Ad Serving</h4><h4>* Conclusion</h4><p><strong><em>For the sake of brevity, in order to keep this Simplified White Paper as a 5-minute quick read, we’ve simplified some of the ML terminologies and generalized the models.</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6h_0SQxFEFBV6NW1_ogPSg.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@herfrenchness?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Clarisse Croset</a> on <a href="https://unsplash.com/s/photos/machine-learning?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><h3>Offline Video Ad Pre-Processing</h3><p>When an advertiser uploads an ad, it is specially encoded using an auto-trained/generative model (<em>e.g. a modified version of GANs</em>), <strong>in order to reduce the size and allow for frame-by-frame and scene-by-scene ML analysis.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9pAE__wkr-d7um-GChi9BA.png" /><figcaption>The video ad being converted into compressed numerical data</figcaption></figure><p>At the end of each day (<em>or it can be more frequent based on your business needs</em>), an offline batch/Spark job uses a combination of advertiser data (<em>pre-trained models produce this data on an advertiser-by-advertiser basis using various proprietary data sources</em>), NLP (<em>Natural Language Processing</em>) models, and Video Analysis models to turn the encoded video ad into usable numerical data for live recommendations.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RIjI6cmE8tzgy8h5gRV0zw.png" /><figcaption>The encoded video data being used by both the Video Analysis Models and the NLP Models, then turned useable data that will be stored in a database for live production</figcaption></figure><p>The NLP models handle sentiment analysis, topic detection, and other language subtleties (<em>double meanings, accents, sarcasm, etc</em>).</p><p>The Video model does frame-by-frame (<em>as well as scene-by-scene</em>) entity detection, and multi-dimensional engaging story detection (<em>what’s going on in the foreground of the scene, what’s happening on the background of the scene, and how “engaging” is each story</em>).</p><p>Both of these models are trained using a combination of private company-specific data, as well as through <em>Transfer Learning</em> from open models.</p><p>There are also a few key post-processing steps that happen after each model is done analyzing, in order to <strong><em>*effectively*</em></strong> combine the two system outputs (<em>i.e.</em> <em>NLP Result + Video Result</em>).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/517/1*Udo-3hRTZjDDPu3kZteuYA.jpeg" /><figcaption>*Generic Image</figcaption></figure><h3>Ensemble Hybrid Recommendation System</h3><p><strong><em>For On-Demand Video Content:</em></strong> Several video ads are matched with various pieces of content, that is then filtered on a scene-by-scene basis via the heavy-duty version of the recommender system.</p><p><strong><em>For Live Video Content:</em></strong> A large number of ads are matched based on the live video’s entertainment category, then is filtered on a frame-by-frame basis via our light-weight version of the recommender system.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/531/1*s-lyuSVskmSkaY6kdxmvtg.png" /><figcaption>*Somewhat Generic Image</figcaption></figure><h4>Recommender System Overview (<em>both On-Demand and Live</em>):</h4><p>For contextual video ads, a <strong>Forest-based Recommender Ecosystem</strong> should be used (<em>several shallowly trained/generalized recommender systems, controlled by a head recommender system</em>).</p><blockquote>Note, this does <strong>not</strong> mean that the recommender systems are Tree-based, but that the <strong>overall architecture/voting process of this system is similar to that of an ensemble, or specifically, similar to models like Random Forest, Gradient Boosting Machine, and XGBoost.</strong></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eoNExY1tI4Lh5tpaAsdLuA.png" /></figure><p><strong>*Generalized Recommender System:</strong> A recommender system to obtain a generalized video ad to video content compatibility score, as well as a video ad to scene/frame compatibility score. <em>We’ll save the exact model used for this system for a different article.</em></p><p>For <strong>on-demand content</strong>, the final output of this model is a list of best content and scenes for a video ad. This list adjusts based on the popularity of a piece of content. We prefer content that is being viewed frequently, in order to give advertisers their ad views as soon as possible.</p><p>For <strong>live content</strong>, the final output of this model is a list of video ads for the current frame, allowing for instantly relevant video ads. If relevancy is flexible, this model can also output the best video ads for a scene (last <em>~45 seconds</em>).</p><p>If the similarity score is below a certain threshold, and no video ads are compatible, we have the option to either show a default ad (<em>e.g. an internal ad like how Hulu advertises their other original programming</em>) or to simply not to show any ad for this time slot.</p><p><strong>*User-Specific, Head/Master Recommender Model:</strong> This model is partially based on the <em>Contextual Bandit ML model</em> in order to offer high, user-level personalization (<em>e.g. personalizing off the user’s most recent data (e.g. mood, recent viewing history, etc.)</em>).</p><p>This model uses user data to filter down the results of the several general recommender models that it controls. Once the choices are filtered, this model will produce a list of video ads to show to the user — <strong><em>with the best one being the one that the user actually sees.</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IgWreevzLW-ZWVCo_RsZIg.png" /><figcaption>The recommender system choosing a video ad to place in the live content stream</figcaption></figure><h3>Conclusion</h3><p>For the sake of brevity, this article leaves out the exact infrastructural details that ensure speed and reliability — especially when working with live-stream content.</p><p>The infrastructure needed involves moving past simply using redundancy and pre-computed offline results; but instead, heavily utilizing both <strong>in-memory storage</strong> (e.g. Redis <em>for handling millions of API requests per second</em>) and a <strong>Functional Infrastructure</strong> <strong>architecture</strong> (<em>e.g. Cloud-Based Infrastructure, Spark on Kubernetes, and Microservices</em>).</p><p>Additionally, we chose not to cover the effective way to pre-process and post-process your content. However, these steps <strong><em>are</em></strong> necessary in order to ensure that the content is numerically compatible with the generalized recommender systems that produce the compatibility score.</p><p>Supplemental Material:</p><ul><li><strong>CBS Interactive Technical Talk:</strong> <a href="https://youtu.be/hX71H78UYAc">https://youtu.be/hX71H78UYAc</a></li><li><strong>Story-Based Classification of Movie Scenes: </strong><a href="https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0228579">https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0228579</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ee8f9b661059" width="1" height="1" alt=""><hr><p><a href="https://medium.com/this-week-in-adtech/simplified-technical-paper-ml-based-contextual-video-ads-on-demand-live-ee8f9b661059">Simplified Technical Paper: ML-based Contextual Video Ads (on-demand &amp; live)</a> was originally published in <a href="https://medium.com/this-week-in-adtech">This Week In AdTech</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>