Gupta Media Case Study PDF

9 -5 2 0 -0 3 1...

9 -5 2 0 -0 3 1 REV: FEBRUARY 21, 2023 V. KASTURI RANGAN COURTNEY HAN Gupta Media: Performance Marketing in the Digital Age In December 2010, Gogi Gupta landed in Miami as one of four outside speakers invited to present at Sony Music’s internal summit. His Boston-based digital marketing company was barely seven years old, and dwarfed in size compared to iTunes, VEVO and Spotify, the other companies invited to speak at the event. When it was his turn to speak, Gupta got up before a grand ballroom full of Sony executives with a bottle of water. Then, slowly pouring its contents into a cup, he turned to his audience, “Have you ever pumped up your media budget only to see it go down the drain?” Gupta waited for a ripple of nodding heads to spread across the room before dramatically flipping over the cup of water. To the audience’s surprise, the cup was empty. 1 Gupta went on to make his now familiar pitch. “The old adage – half of advertising works, but we don’t know which half – is no longer true,” Gupta explained. “We know precisely how each intervention works. That’s the promise of digital marketing.” 2 Nearly a decade later, by early 2019, Gupta Media had built a strong reputation promoting music labels and their Grammy-award-winning artists who ranged from Lady Gaga to Adele. The firm was a Twitter Official Partner, media advisor to entertainment companies such as Universal Music, Sony and Amazon Music, as well as the Cleveland Cavaliers. 3 The company had outgrown its co-working space at the Cambridge Innovation Center and relocated into a prime downtown Boston location with sixty employees, most below the age of thirty, who sat in a vast open floor office space surrounded by gold records and gig posters on the walls. “Our new space was designed from the ground up in 2017, with plenty of room to work, play and relax,” Gupta said. 4 There were no props or banners promoting advertising creatives, no waiting area for guests or even a receptionist. Gupta, his senior account managers and analysts tucked their laptops under their arms as they moved from meeting to meeting. The office resembled a software development shop more than a traditional ad agency. Gogi Gupta arrived at the office early one May morning in 2019 with a full day of meetings ahead. Two major accounts demanded his attention. First, the Governors Ball Music Festival, an annual three- day event on Randalls Island in New York, was fast approaching. Ticket sales were sluggish compared to prior years and with four weeks remaining, the team had nearly spent an amount nearing the previous year’s budget but delivered far fewer tickets. Tom Russell, head of Founder’s Entertainment, wanted to confer with Gupta and Alex Palmer, the Governors Ball campaign lead at Gupta Media, to discuss how to boost sales in the last crucial weeks. Professor V. Kasturi Rangan and Research Associate Courtney Han prepared this case. It was reviewed and approved before publication by a company designate. Funding for the development of this case was provided by Harvard Business School and not by the company. Certain details have been disguised. HBS cases are developed solely as the basis for class discussion. Cases are not intended to serve as endorsements, sources of primary data, or illustrations of effective or ineffective management. Copyright © 2019, 2020, 2021, 2023 President and Fellows of Harvard College. To order copies or request permission to reproduce materials, call 1-800-545-7685, write Harvard Business School Publishing, Boston, MA 02163, or go to www.hbsp.harvard.edu. This publication may not be digitized, photocopied, or otherwise reproduced, posted, or transmitted, without the permission of Harvard Business School. This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to Mar 2025.  520-031 Gupta Media: Performance Marketing in the Digital Age Across the office, a different challenge loomed with Fender Play. The musical instrument company Fender Musical Instrument Corporation (FMIC) had hired Gupta Media to promote its new online guitar learning app called Fender Play. For the past year and a half, Gupta Media was deeply involved in promoting the product, which became one of the agency’s largest campaigns. Fender’s CMO Evan Jones recently called Gupta to relay the Fender executive team’s desire to see Fender Play turn into a stand-alone product. In other words, the Gupta campaign would be responsible for generating not only user trials but also conversions and revenues. Gogi Gupta was to meet with Jordan Maddocks, his lead for the Fender account, to decide how they would revise strategy for ad purchases in light of Fender’s request. There was a third issue on Gupta’s mind. Last week, he had received a call from the marketing branch of Red Bull, asking if Gupta Media would be interested in pitching a media campaign to promote the brand. Red Bull’s marketing strategy focused on building customer relationships by identifying with a lifestyle image of extreme sports and cool, daring activities. Gupta Media had excelled at delivering advertising performance where metrics were readily available, but the agency had less experience measuring the impact of digital advertising on offline sales, especially with the kind of broad distribution associated with a brand like Red Bull. He wondered what Gupta Media’s value proposition could be. The Digital Advertising Ecosystem For decades, the bulk of advertising sales relied on large, upfront investments in traditional media such as print and TV. The advent of the Internet introduced new marketing opportunities for advertisers online. In only two and a half decades, 5 the digital advertising industry grew to rival traditional advertising in sales, with digital media projected to surpass traditional advertising revenues in 2019, largely driven by declines in TV and print. 6 In 2018, out of a total U.S. media spend of $223.7 billion, digital advertising accounted for $107.5 billion, just under half of the overall market, with a 22% increase alone from 2017. 7 The rise was largely driven by growth in mobile and video ads, which reached $69.9 billion and $17.3 billion, respectively. The digital ad market was broadly comprised of buyers or “advertisers,” e.g., Procter & Gamble, and sellers or “publishers,” such as Google and Facebook. Advertisers could manage their marketing needs internally or hire agencies to plan and execute a portion or all of their campaigns. In the early days of digital marketing, advertisers often purchased ad placements directly from the publisher, following a “direct buy” model common in traditional print and television advertising. After negotiating a price and volume, typically measured on a cost per thousand impression a (CPM) basis, the ads would appear on the publisher’s websites. Direct buys didn’t allow advertisers to target specific audiences or adjust their strategies during the campaign based on how viewers responded to the ads— advertisers could only specify where and how many ads were shown. In April 2005, Right Media launched an online auction where advertisers could bid for ad impressions and publishers could award winners based on the highest offer. 8 Also called an ad exchange, the new marketplace used automated algorithms and real-time bidding (RTB), a method that allowed ad impressions to be bought, sold and displayed practically instantaneously, at any hour of day. 9 Soon ad exchanges such as Google’s DoubleClick, Facebook Exchange and OpenX opened and began selling ad spots to advertisers based on bid price, ad relevance and quality. The exchanges were equipped with tools that enabled advertisers to submit bids based on data about a user’s past behavior a One impression counts as every time an ad is delivered or exposed to a web user. 2 This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to Mar 2025.  Gupta Media: Performance Marketing in the Digital Age 520-031 on the publisher’s site, an important advancement that allowed advertisers to segment and target audiences more precisely than in direct buying. Furthermore, unlike the analog world where an advertiser allocated a lump-sum budget towards marketing, advertisers operating in the digital space could specify daily highest bids and a budget, and ad exchanges such as Facebook and Google would optimize media spending accordingly. The new automated auction approach was called programmatic ad buying and quickly became the preferred method for placing ads online. By 2019, programmatic ad sales accounted for 80% of total digital display advertising revenues. 10 The largest ad exchanges used second-price auctions, where the winner paid the price of the next highest bidder’s bid. Ad quality and relevance for the intended audience were important to publishers’ reputation. Since publishers wanted users to receive relevant content, publishers considered the quality and relevance of ads to the target audience when calculating the winning bid. A more relevant ad, based on CTR (click through rate) or engagement rate would win an auction over a higher priced bid, but less relevant, ad. The publishers (Google, Facebook) did this to ensure that users viewed advertising as useful and not an interruption to their social media activities. Exhibits 1 and 2 provide brief details on how advertising auctions are conducted at the two largest digital media publisher platforms, Facebook and Google. To manage and analyze data for media campaigns, advertisers often used Data Management Platforms (DMPs) that they connected to a Demand-side Platform (DSP) where they could specify their media purchasing needs and execute bids on various ad exchanges. 11 DSP software came with varying levels of target market customization, machine learning capabilities and data integration options. 12 Advertisers and media agencies could draw on a wide range of tools to enhance the quality and targeting of their ads. Analytics companies (Quantcast, Nielsen), for instance, sold data about ad placement and user clicks-through rates and Verification companies (Convertro, Tagman) compiled data into legible performance reports. Technology platforms called Ad Servers (DoubleClick, Atlas) provided data on user engagement and conversion. Creative optimization companies (Tumri, AdReady) improved digital ad functionalities. 13 See Exhibit 3 for a thumbnail sketch of players in the digital advertising industry. Rapid technological advancements, new social media platforms, disruptive digital business models and privacy regulations like the European Union’s General Data Protection Regulation (GDPR) pushed rapid changes in the digital ad industry. A typical media campaign in 2019 ran on different digital devices—computers, mobile phones, tablets—and across different channels—search engines, social media, display ads and TV—thus multiplying the complexity and scale of feedback that advertisers could use to refine the targeting efficiency of their ads. Whereas advertisers once possessed a limited understanding of what led users to their content, by 2019, advertisers could know whether target audiences liked an ad, left a comment, watched a video or moved an item to the shopping cart. 14 Advertisers used cookies b to track and follow user traffic on their sites. Usually this was proprietary data and not made available to publishers. Facebook, Google, and other publishers provided code that advertisers and/or any organization could install on their websites (pixels c) or apps (SDK) voluntarily, to track the activity on their web sites or apps, respectively. Many major publishers such as Facebook, Google, Twitter, and Snapchat kept user activity on their sites proprietary, earning their sites the reputation of being “walled gardens.” 15 Publishers made such data available to the first party owner b Cookies are small text files that users automatically downloaded when visiting a website. They helped websites identify users and their behavior, but could also be deleted by the user and did not translate across devices. c Pixels tracked user behavior on websites and through social share icons (such as Likes) and enabled websites like Google, Facebook and Twitter to “listen in” on user behavior. 3 This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to Mar 2025.  520-031 Gupta Media: Performance Marketing in the Digital Age of that data freely, but not others. Not coincidentally, such companies were also the biggest digital media sellers. Advertisers were reliant on a small handful of publishers to reach their target customers. In 2018, Google held 38% and Facebook 22% of the total digital ad revenue share in the United States. 16 The variety of data and measurement tools improved advertisers’ ability to target and convert audiences to customers, yet challenges remained. Managing and making use of various sources of customer data could become unwieldly. Companies couldn’t always control when, where or what type of content their ads ran alongside, especially when purchasing ads on Google’s Display Ad network. Fraud from clicks by Internet robots (bots) remained a concern, as well as the pervasiveness of ad blockers. 17 Another contentious issue was that of attribution. A targeted customer might click on an ad but not make a purchase right away. How soon after the last click did an ad convert an interested person into a customer—the same day, a week, or a month? And how could off-line conversions be measured? In spite of these and other issues, there was a growing consensus that digital marketing provided better cause-and-effect measurement for advertising spending than traditional forms that prompted advertisers to turn their attention online. Bidding for Digital Media Three pricing models were commonly used in programmatic ad buying. Advertisers paid by the number of ad exposures when they used the cost per thousand impressions (CPM) model, a former industry standard and similar to the pricing model used in traditional media advertising. The digital advertising industry has been shifting to performance-based pricing models in recent years that better captured an ad’s effectiveness and increased risk-sharing between the advertiser and publisher. The cost per click or pay per click (CPC or PPC) pricing model charged advertisers when their ads were clicked, and measured user engagement with ads, such as clicking on a link for an advertiser’s website, viewing promotional materials or watching a video. A third model, cost per action (CPA), tracked when a customer completed a given “action” specified by the advertiser, such as making a purchase, downloading an app or signing up for a list-serve. Advertisers could specify an attribution window, or the time (7, 14, days) between when the customer first engaged with the advertisement and when they completed the action. The formula for converting ad impressions to purchases is given by: Impressions X CTR X Action Rate d For example, a hypothetical advertiser seeking 1,000 impressions and paying $10 to place them, is said to be paying $10 CPM or 1 cent per impression. Further if the CTR to the advertiser’s website is 10%, that would amount to 100 clicks, or 10 cents CPC. If 1% of those who clicked converted to a purchase, then the advertiser’s $10 budget will result in one purchase or a CPA of $10. Starting in 2006, performance-based models began to surpass the CPM model in popularity. 18 In 2013, approximately 65% of online advertising revenues were priced by performance models, 19 a share that fell only slightly to 62% by 2018. 20 Ad priced by the CPM model stayed relatively consistent between 2013 and 2018 at roughly a third of total ad revenues. 21 d Where impressions X CTR = Clicks/ Visitors, and Visitors X conversion rate = # of Actions 4 This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to Mar 2025.  Gupta Media: Performance Marketing in the Digital Age 520-031 Gogi Gupta Born in India, Gogi Gupta grew up in Pittsburgh and Buffalo. He graduated from Cornell University in 2000, where he studied public policy. “I was a country bumpkin from Buffalo,” Gupta joked. “I took my Bs and all of the smart kids went to office hours and got As.” 22 Good with numbers, he thought he would be “the guy Goldman Sachs would call up for advice about algorithms.” 23 After a stint in management consulting, Gupta left to work for a small technology company in Cambridge, Massachusetts that closed during the dot-com slowdown in early 2000. Instead of looking elsewhere again, Gupta decided to start a venture of his own. Despite admitting to a lack of talent in music, Gupta saw an opportunity within the music industry to help labels better market their artists online. As an early entrant into the industry’s digital marketing space, Gupta was able to carve out a niche for himself. “I started on the publishing side, trying to help some of my favorite sites monetize their traffic,” Gupta said. “The sell side is much harder than being on the buy side.” 24 When Gupta landed Hollywood Records, part of Disney Music Group, as a client, his business began to take off. In his view, record labels had too many fans and not enough customers. “When Eminem was the biggest social media star and Recovery was the biggest album that year, we worked the math out and found that only about 9% of his US fans bought the album. There were about 32 million social media fans there. What we want is to get that 9% number to 15%.” 25 Gupta’s data analytics team built a solid understanding about how people consumed music, with segments broken down by age, gender, location, online spending habits, music preferences, down to the types of YouTube videos that fans of particular artists watched. “When we started promoting Lady Gaga before she was a star, we knew how to find a core audience that was influential and a good fit for her: gay men in New York who frequent dance clubs,” Gupta explained. Gupta took that knowledge and used it to target ads towards very specific audiences with a predisposition to enjoy Lady Gaga’s music, and iterated from there. “Instead of spending large sums for one big blockbuster ad during the Super Bowl, companies can spend smaller amounts on multiple targeted campaigns and get exact data on viewer response, from how many people saw the ad, to how many clicked through to the artist, to how many made a purchase.” 26 Gupta Media: A Full-Service Marketing Agency In a typical campaign, the client might provide Gupta Media with creative content and first-party data such as email lists, mobile phone numbers and past purchaser data. Gupta Media segmented customers based on the degree of their exposure and engagement with the client’s product to generate an addressable audience of people already familiar with the product alongside a strategy for finding and exposing the product to new audiences. When executing the strategy, Gupta Media sometimes augmented first-party data with second- and third- party data. e This could involve purchasing data on music video views of Governors Ball artists from YouTube (second-party data) and combining it with credit card records (third-party data) of New York state residents who purchased concert tickets in the past year. The resulting cross-section of people who viewed YouTube videos of artists slated to perform at the festival and recently spent e First-party data is that owned by the advertiser. Second-party data is owned by another company and available for purchase. Third-party data is commonly aggregated from various sources and typically includes consumer behavioral and/or demographic data. 5 This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to Mar 2025.  520-031 Gupta Media: Performance Marketing in the Digital Age money on concert tickets, could form a particular audience segment. Gupta Media placed native f ads and gauged user engagement on the top media platforms including Facebook, Google, Twitter and Snapchat. By closely monitoring ad performance, Gupta argued that it could attain higher Click- through and Action Rates, therefore lowering CPC. At its Boston offices, analysts could be seen poring through live data, sometimes by the hour, to revise bidding and placement strategies. The advertising industry’s average commission rate for digital media spanned 6-20%, but Gupta Media’s superior performance enabled the company to garner commissions at the top end of that range. 27 “I tell clients, if we are not significantly better than others then don’t pay us,” Gupta says, “we are data-driven, so the number one thing is that we are transparent about the performance for every ad unit we push out. If you buy an ad from us, you’ll know exactly how it’s doing. We provide our clients real-time data for every ad unit.” 28 Clients were given access to check their ad performance on the Gupta Media software platform during campaigns. By mid-2019, the firm had placed some half billion dollars of advertisements, including $60 million worth of digital media on behalf of 150 clients in 2018 alone. 29 Roughly a third of its clients accounted for two thirds of its revenues. In an effort to expand its capacities and compete with companies such as Swirl, MuteSix, Fetch and MediaCom, the agency created a creative team that designed ads in-house to add to its suite of services. An Example of a Gupta Media Campaign: George Ezra For twenty-one months, between July 2017 to April 2019, Gupta Media worked with Columbia Records UK to promote British pop musician George Ezra’s second album, Staying at Tamara’s. Ezra had released a debut album four years earlier, and it performed well, debuting at number three on the UK Albums chart before eventually climbing to number one and ending the year as the third best- selling UK album of 2014. 30 Jess Dashner, Account Director and head of the Tamara’s campaign, explained the overall targeting and messaging strategy behind the campaign. “We were constantly trying to expand the fan base by adding new audiences who would eventually convert from casual listeners to hardcore fans, where most of the singles are streamed and albums bought,” Dashner said. 31 Gupta Media broke down audiences into three tiers. Tier 1: Primary or “super” fans: This segment captured George Ezra’s established fan base. Since these fans were already well aware of George and his music, Gupta Media focused on persuading them to buy Ezra’s new music. Gupta Media broke down this segment further into four categories that they referred to as “pillars.” Internally referred to as the Four Pillars strategy, the agency applied this approach to estimate the size of its core addressable audience for all of its campaigns. 1) Searchers. This segment included people actively searching for George Ezra on different platforms. Gupta Media prioritized ad placements for this audience segment since they were already actively anticipating the new album. For instance, people who searched for “George Ezra new album” received an ad with a link to pre-order the album directly above the search results. With one click, the searcher could buy or stream songs and albums. 2) Connected fans. Members of this segment had previously sought and received information about George Ezra, such as subscribing to his Email list, following him on Instagram or liking his page f Native advertising is a form of online advertising designed to match the form of a post by a user of the platform. 6 This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to Mar 2025.  Gupta Media: Performance Marketing in the Digital Age 520-031 on Facebook. Gupta Media served these fans ads for his new songs/album, and tickets for his concerts. 3) Short-term retargeting audience. Retargeting allowed Gupta Media to find users who expressed specific interest in George Ezra by visiting his Website, watching a video, or sampling a song. Gupta Media served these customers with promotions that encouraged them to purchase the music. 4) Previous buyers. This segment included people who demonstrated a prior willingness to buy George Ezra’s music. They received ads that reminded them about Ezra’s upcoming album. Compared to Tier 1 audiences, Tiers 2 and 3 included the segments of people who were less familiar with George Ezra but had demonstrated a predisposition for his or similar music through their past actions and behaviors online. Gupta Media served ads that introduced the artist and his music to these audiences without suggesting that they purchase. Tier 2: Casual Fans: This population was aware of George Ezra and his music but not necessarily followers and typically didn’t spend money on album pre-orders, streaming or Ezra’s events. Gupta Media focused on promoting singles to these audiences in the hopes that they would like what they heard and demonstrate more engagement with his music. Tier 3: New Audiences: This tier included people who liked similar artists and genres to George Ezra as well as viewers of shows that Ezra performed on. Gupta Media focused on a “light touch campaign” by serving ads to individuals with similar demographic profiles as Ezra’s fan base, in this case, urban, young, slightly female-leaning fans of Indie rock. A/B Testing Gupta Media used A/B testing methods across all three tiers to identify the advertising that performed best in converting audiences to the desired action. For example, Gupta Media randomly served one of three different ads to Tier 1 audiences and tracked outcomes via metrics such as ten- second video engagement, sound-on viewing, and click-through rates. Since “I’d Buy It” performed the best, that became the ad that was served at scale to Tier 1 audiences. Table 1 Testing Creative ads for Effectiveness IG Stories 10s 10s View % Sound Video % Link Ad Preview Impressions CTR Creative Views Rate on Watched Clicks https://vimeo.com/g “I’d Buy it” uptamedia/review/3 1,687,087 103,955 6.16% 84.01% 18.66% 10,182 0.60% 63914385/29c7f73dce https://vimeo.com/g No. 1 uptamedia/review/3 181,735 3,305 1.82% 58.67% 9.90% 648 0.36% Boomerang 63914392/ebcaec194c Official https://vimeo.com/g Video/TV uptamedia/review/3 2,001,548 29,980 1.50% 65.20% 6.15% 5,095 0.25% Advert 63914406/29f10a8dca Source: Gupta Media. 7 This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to Mar 2025.  520-031 Gupta Media: Performance Marketing in the Digital Age Ad #1 – “I’d Buy It” – This ad was a 1st person creative showing George Ezra watching his own “commercial” on his phone. The camera then panned to him excitedly telling the camera “I’d Buy it.” Ad #2 – “No. 1 Boomerang” – This ad highlighted the news that the album was at #1 while also mimicking the “Boomerang” style of video which was popular on Instagram. The ad copy was in the 1st person, but George Ezra did not directly talk to the camera, as in Ad #1. Ad #3 – “Official Video/TV Advert” – This ad was a 3rd person creative (“out now” vs. “get my album now”) and was much closer in style to what the consumer would normally see as a commercial on TV. It was more polished than the previous two versions, making it look professionally produced, but that also tended to make it less “thumb stopping” as consumers tapped through their Instagram stories. It had more the look and feel of an ad and less like organic content. Overall Campaign The full Ezra campaign was structured around key moments of Ezra’s album promotion schedule that were determined by the record label. These moments took place around once a month and included single –track and music video releases, show appearances, and concert dates. For example, in the first six months of the campaign, Gupta Media raised awareness about George Ezra’s upcoming album to existing fans using ads such as: https://vimeo.com/guptamedia/review/363915351/e1f1111006 The company ran ads for songs from Ezra’s first album since it was common for fans to listen to older songs around the time of a new music release: https://vimeo.com/guptamedia/review/363913835/cd6365ffba Three tracks from the album were also released at different points throughout the pre-order period as incentives to encourage album pre-orders: https://vimeo.com/guptamedia/review/363914373/17b9e9a583 Over the course of the campaign, Gupta Media sought to move audiences from lower tiers into tier 1. The agency would purchase impressions to show a music video clip to potentially interested people in tier 2 and then retarget those who watched at least two minutes of the video (tier 1) with an ad encouraging them to buy or stream a single. Those who purchased the single might be then be served an ad to purchase a concert ticket. Over the nearly 22 months of the campaign, Gupta Media served up nearly 121 million impressions and nearly 1.736 million link clicks to these audiences. Staying at Tamara’s debuted on March 2018 and reached #1 in its first week. 32 See Exhibit 4 for a summary of efforts made on different platforms during the campaign. Decisions Regarding Governors Ball and Fender Play The Governors Ball Music Festival Sales Challenge Gupta Media and Founders Entertainment had partnered for the past eight years to drive digital strategy, media buying and design support for the Governors Ball, a large three-day music festival in New York that took place at the beginning of June. The festival had featured major acts such as Nas (2013), Drake (2015), Beck (2016) and Eminem (2018). A three-day ticket started at $299 plus $30 service fee, and a one-day ticket at $129 plus $20 service fee. The Randalls Island venue could hold 60,000 8 This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to Mar 2025.  Gupta Media: Performance Marketing in the Digital Age 520-031 people per day, and average gross margins after talent fees, venue rental and facilities contracting was 25%. Exhibit 5 shows a poster of the 2019 line up. Gupta Media strategized, produced advertisements, creative copy and models on ticket sales for the event. An in-house team at Gupta Media executed the plan, buying ads on a daily basis and optimizing placement on all major publishing platforms. In 2018, Gupta Media sold a total of 31,387 tickets equaling 76,964 “ticket days” of attendance with an advertising budget of $332,284. Gupta Media used a conservative 7-day attribution window to account for conversion, making it likely that even more tickets were sold as a result of advertising spend but unaccounted for. The agency’s advertising efforts drove about half of overall ticket sales while organic advertising such as PR, email and social media drove the other half. By contrast, in 2019, with four weeks before the start of the concert, only some 13,444 tickets had been sold and $271,377 spent. In the last four weeks leading up to the 2018 festival, advertising expenditures totaled $57,929 to sell 11,229 tickets. If the same were to repeat in 2019, the total advertising spend would be roughly the same as last year but ticket sales would still fall well short. Click on the separate attachment, Gupta Media Spreadsheet Supplement (HBS No. 520-706), to download the 2018 (full year) and 2019 (with 4 weeks remaining) data on advertising and sales. Another supplement, Gupta Media (B) (HBS No. 520-039) provides the promotional materials from 2018 and 2019 that are referred to in the data spreadsheet. The Spreadsheet Supplement contains nine data tabs: 1. Metrics and definitions 2. Demographic data of the NY metropolitan area 3. 2019 sales data with four weeks remaining 4. 2019 demographic data of buyers (Facebook and Instagram only) 5. 2019 buyer data by platform. (Twitter was discontinued after 4 weeks) 6. 2018 full 22 weeks of sales data 7. 2018 demographic data of buyers (Facebook and Instagram only) 8. 2018 buyer data by platform 9. Mock Up of the Facebook bidding interface page Gogi Gupta and Alex Palmer, head of the Governors Ball campaign at Gupta Media, planned to adjust bids and expenditures on a daily basis leading up to the event. They would meet with Tom Russell, head of Founder’s Entertainment and a long-term collaborator of Gupta’s, to discuss the larger ad sales strategy. What advertising spend should Gupta Media recommend to Tom Russell for the next four weeks? How many tickets would that sell? Who should they target and how would they justify the expenditure? The Fender Play Marketing Challenge With $500 million in 2018 revenues, 33 Fender Musical Instruments Corporation (FMIC), was a premier manufacturer of high quality, upmarket guitars, basses, amplifiers and related equipment. The company boasted a long history supplying instruments to award-winning musicians. Fender was 9 This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to Mar 2025.  520-031 Gupta Media: Performance Marketing in the Digital Age facing changes to its business lines, as music retailers like Guitar Center closed stores and music genres such as rap, and R&B gained in popularity. As a result, Fender began to reach out directly to build relationships with potential end customers. Its new app, Fender Play, taught guitar skills to beginning and intermediate players through a series of video lessons. For the past year and a half, Gupta Media ran ads for Fender Play across Facebook, Google Search, Twitter, YouTube and Snapchat, and built a stable customer base of nearly 245,000 users. The campaign grew to become one of Gupta Media’s largest accounts, with a little over seven-figure annualized marketing budget. See Exhibit 6 for images from the Fender Play campaign. Jordan Maddocks, the data analytics lead at Gupta Media and Ryan McGee, Director of Growth and Acquisitions at FMIC reviewed the campaign’s most recent performance data and found that advertising spend averaging roughly $102,769 per week translated to an estimated $177,350 in incremental (ad driven plus organic) potential lifetime customer revenues. Currently potential consumers received a free two-week trial, after which they could choose to sign up for either an annual plan of $100 or a monthly plan of $10. The advertisements when viewed by a mobile phone user, would direct them to click through to their respective app store (iOS or Android depending on their phone device) to order the free trial. The ads could also direct them to click through to the Fender Play web site directly to order the free trial (including when they accessed the ads on their computer). At the end of the two-week trial period, users would be reminded to convert to a paying subscription or opt out. The conversion rate varied by which store they went to for the trial, iOS, Android or the company’s web site, and by whether they chose the monthly or annual subscription plan. Exhibit 7A provides the pricing and LTV customer value data and Exhibit 7B provides the campaign’s performance data. Until more recently Jordan Maddocks had followed a media strategy of maximizing subscribers for Fender Play assuming that the company would prefer this strategy as a way of converting lesson customers to purchase guitars. Apparently there was a change in direction, and Gogi Gupta had recently received a request from Fender to focus effort on boosting lesson sales, which had stagnated at roughly $175,000 over the last several periods. Fender wanted more users but more importantly wanted a significant increase in sales revenues of over 25% without increasing the ad spend. Return on Ad Spend (ROAS) would be the new metric of choice. The Gupta team had previously run some advertisements using discount codes and Maddocks studied how such discounts affected user trial rates. Discounts would apply as long as the subscribers renewed, but in any case for not more than a year. Knowing that both user quality and the cost of ad inventory varied by device (Web, iOS, Android), he used historical campaign data to make projections on how 10% to 50% discounts would affect trials down to the device level (see Exhibit 8). Exhibit 9 shows the estimated media costs (averaged across media) for various levels of exposure and subsequent clicks. Maddocks did some preliminary projections at two levels of price discounts 30% and 50% and as expected, the number of incremental trials went up, and so did the ROAS, but the revenues were flat or even declining. See Exhibits 10 and 11. This he knew would not address the new strategy at Fender. Maddocks wanted to find the optimal combination of price discounts (could be different for monthly and annual products) and advertisements served that could raise revenues significantly, without increasing total ad spend from the current $102,000. What should he recommend to Fender? 10 This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to Mar 2025.  Gupta Media: Performance Marketing in the Digital Age 520-031 What to do about Red Bull? Confident that Maddocks and Palmer were hard at work to address the Governors Ball and Fender Play marketing challenges, Gogi Gupta returned to his desk where he was reminded about Red Bull. The marketing team was waiting for his response to their pitch invitation. Credited with creating the energy drink category, Red Bull was a €6.28 billion (sales) global brand that sold in over 120 countries and remained the #1 energy drink in the United States. It sold over seven billion cans every year through bars, clubs, supermarkets, gas stations, vending machines and liquor stores, but nothing directly to the consumer. The brand’s value proposition was “giving wings to people and ideas” and “setting milestones in sports and culture.” 34 See Exhibit 12 for examples of their brand image. Red Bull invested heavily in creating and maintaining its brand but had recently faced challenges by new entrants like Monster. 35 The company recruited extreme sports athletes as brand ambassadors, from mountain biking and BMX (bike racing on a dirt track) to windsurfing and parkour (obstacle courses). They hosted extreme sports events such as the Red Bull Cliff Diving World Series, Air Race and Crashed Ice along with stunts including the Stratos space diving project and the renowned Red Bull Flugtag, where participants created their own flying machines and tested them by diving off a cliff. 36 They owned soccer teams in Europe and Brazil, and recently built the apps Red Bull TV, RBMA Radio and Bike Unchained, 37 as well as a social gaming platform with its own in-game island. 38 Gupta Media had built its reputation developing precise metrics that targeted audiences at the bottom of the marketing funnel. Measuring success in the top of funnel space, one typically dominated by TV ads, was much harder. How would they connect offline and online marketing, and measure performance? What could their unique selling point be for such an established brand? Even if a marketing campaign was possible, was it reasonable for a performance marketing firm like Gupta Media to take on this kind of broad media campaign? What should Gogi Gupta include in his pitch? Gupta thought back to his time at the Sony Music summit. Undaunted by the size and experience of his competitors, he had managed to make a lasting impression, but Red Bull would be a different kind of challenge, with a less clear connection between strategy and outcomes. Could the tools that made Gupta Media so successful at converting sales for top artists and music festivals also work for a consumer brand like Red Bull? 11 This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to Mar 2025.  520-031 Gupta Media: Performance Marketing in the Digital Age Exhibit 1 How Facebook and Instagram Advertising Works Facebook boasted over one billion daily active users to its Website and two billion users when one included Facebook’s other tools and apps (Messenger, Instagram, WhatsApp). Facebook segmented its audiences into three categories. Saved audiences comprised the total pool of Facebook users, for whom Facebook maintained information on its users’ location, interests, age, gender, language, income and social media activity. Custom audiences were comprised of the advertiser's first-party data gathered from email list- serve subscribers, phone numbers, device IDs, etc. of people who previously engaged with the advertiser’s content. Hash emails could be uploaded and matched with members of Facebook’s saved audiences by a unique identifier as long as they were members of Facebook. Lookalike audiences were generated by Facebook’s proprietary algorithm to include people with profiles similar to the advertiser’s custom audience. This match occurred by considering dozens of preference and interest variables only known to Facebook. Advertisers could upload first-party data onto the platform, creating a “custom audience,” within Facebook, and Facebook would report what percentage were one-to-one matches with its saved audience. Alternatively, by matching engagement profiles of custom audiences to Facebook’s lookalike audience, Facebook could target 1% to 10% of saved audiences that most closely resembled the custom audience. This could be done by geography or other demographic variables requested by the advertiser. Advertisers could specify among a wide range of variables to identify a target audience of their choice.a Facebook claimed to consider three factors in bids. First and most critically, the bid amount mattered. Low bids for highly popular ad spots were unlikely to do well. Facebook’s algorithms also considered variables such as “likes,” “comments,” “shares,” and “clicks,” and weighed them against negative comments to calculate a relevance score for every ad. The third metric, called estimated action rates, was a proprietary algorithm that measured the likelihood of the targeted customer ignoring or responding negatively to an ad. Once an advertiser entered a bid into the Facebook interface, an automated algorithm took the three factors into account and awarded the bid to the advertiser with the highest overall score. The winner was determined by a second-price auction, with the winner paying one cent more than the next highest bidder. Facebook generated periodic reports for the advertiser, to provide detailed data, including video completion rates and a normalized “dwell time” metric on a picture, or how long people stopped at a given ad when scrolling through the Newsfeed relative to other ads. Source: “Facebook Ads Guide,” https://www.facebook.com/business/ads-guide, accessed September 26, 2019. a Facebook sets a minimum of 100 people for its target audience to protect user privacy. 12 This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to Mar 2025.  Gupta Media: Performance Marketing in the Digital Age 520-031 Exhibit 2 How Google Advertising Works Google offered advertisers two ad networks, Search and Display, to reach audiences. The Search network placed advertisers’ material above organic results when a person searched for a particular product or service on Google’s homepage. Google Ads (formerly Google AdWords) allowed advertisers to bid on a set of keywords so that when such words appeared in Google, their clickable ads would appear in Google’s search results. Their order was determined by their auction rankings, which is explained in the next paragraph. The display network included the pool of all Websites that agreed to display Google ads using Google AdSense. These were typically banner ads that appeared on the margins of the screen. Google estimated that nearly three million websites hosted spots for ads. Advertisers who bid for ads in the display network were not targeting users already searching for the advertiser’s products or services, but exposing their ads to anyone who visited a Website in the display network. Google ran similar but separate ad auctions for the two networks to determine what ads to show, and the order of appearance. In the Search network auction, advertisers identified keywords from their advertisements, specified a preferred pricing metric, bid amount and a list of keyword groupings paired with their ads. Google then entered the keyword from the advertiser’s account it considered most relevant into the auction with the maximum bid the advertiser specified. An auction began when a user entered a search query. A computer algorithm conducted an instantaneous auction among advertisers whose keywords overlapped with the user’s search query and ranked advertisers based on bid size and Ad Quality Score. Google’s Quality Score was a proprietary metric that indicated how relevant and useful an ad was to the customer. Since the advertisement was served on its search page, ad rank determined the order of ad placement. The highest ad rank scorer that received the top position in the display was the auction winner and paid the CPC (cost per click) according to the formula shown in the illustration below. Ad Rank Score = CPC BID X Quality Score The highest combined Maximum price bid Proprietary metric that determines CPC Bid x Quality Score specified for the keyword. ad relevancy to the user (potential receives the highest rank. CTR, relevance, quality of landing page). The higher the quality score, the better. Price Paid= Ad Rank Score of next ranked bidder / Own Quality Score + $0.01 Max Bid Quality Score Ad Rank Score Actual CPC Advertiser 1 $2.00 10 20 16/10 + 0.01 = $1.61 Advertiser 2 $4.00 4 16 12/4 + 0.01 = $3.01 Advertiser 3 $6.00 2 12 8/2 + 0.01 = $4.01 Advertiser 4 $8.00 1 8 Highest CPC Source: “Auction - Google Ads Help,” https://support.google.com/google-ads/answer/142918, accessed September 26, 2019. 13 This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to Mar 2025.  520-031 Gupta Media: Performance Marketing in the Digital Age Exhibit 3 The Display Ad Tech Landscape Source: David Christian, “Display Advertising Technology Landscape.” https://prezi.com/katuvp2rkyk_/the-display- advertising-technology-landscape/, accessed August 15, 2019. 14 This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to Mar 2025.  Gupta Media: Performance Marketing in the Digital Age 520-031 Exhibit 4 Summary of Digital Efforts for the George Ezra Campaign Source: Company documents. 15 This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to Mar 2025.  520-031 Gupta Media: Performance Marketing in the Digital Age Exhibit 5 Governors Ball 2019 Line Up Source: Company documents. 16 This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to Mar 2025.  Gupta Media: Performance Marketing in the Digital Age 520-031 Exhibit 6 Fender Play Sample Ads Source: Company documents. 17 This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to Mar 2025.  520-031 -18- Exhibit 7a Life Time Value of Acquired Customer Avg. Price of Avg. # of Months App Store Avg. Revenue per Estimated LTV per Free Trial to Paid Estimated LTV SKU Price Plan per Month per Active user Fees month After Fees Active Paying User Conversion Rate per Trial Start Annual Web $100.00 $8.33 19.0 0% $8.33 $158.33 73.3% $116.11 Annual iOS $100.00 $8.33 16.5 30% $5.83 $96.25 49.9% $48.03 Annual Android $100.00 $8.33 14.2 30% $5.83 $82.83 37.0% $30.65 Monthly Web $10.00 $10.00 8.0 0% $10.00 $80.00 75.4% $60.32 Monthly iOS $10.00 $10.00 7.0 30% $7.00 $49.00 57.0% $27.93 Monthly Android $10.00 $10.00 4.0 30% $7.00 $28.00 53.0% $14.84 Source: Company documents. Exhibit 7b Campaign Weekly Performance Data (Average of Last 4 Periods) Discount Trial Advt. Click Advt. Ad* Total* Total Estimated Estimated Mar 2025. Product Platform SKU CPC Advt. Cost ROAS Amount Rate Volume Trials Trials % Trials CPA LTV per Trial Revenue Annual Annual Web 2.70% 4,000 $1.82 $7,280.00 108 44% 245 $29.66 $116.11 $28,500.00 291% Web Annual Annual iOS 0% 3.20% 6,000 $1.30 $7,800.00 192 44% 436 $17.88 $48.03 $20,958.00 169% iOS Annual Annual Android 2.40% 6,000 $1.10 $6,600.00 144 44% 327 $20.17 $30.65 $10,030.36 52% Android Monthly Monthly Web 3.20% 12,000 $1.54 $18,499.20 384 44% 873 $21.20 $60.32 $52,642.91 185% Web Monthly Monthly iOS 0% 3.50% 18,000 $1.65 $29,700.00 630 44% 1,432 $20.74 $27.93 $39,990.68 35% iOS Monthly Monthly Android 3.40% 22,000 $1.50 $32,890.00 748 44% 1,700 $19.35 $14.84 $25,228.00 -23% Android Totals - 3.24% 68,000 $1.51 $102,769.20 2,206 44% 5,014 $20.50 $37.37 $177,350.00 73% Source: Company documents. Note: *User Trials were both organic and that attributed to digital advertising. Total Trials is the number combining both. The proportion due to advertising varied from 40% to 48%, and is averaged to 44% for convenience. This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to  520-031 -19- Exhibit 8 Price Discount and Estimated Trial Rate Price Discount and Estimated Trial Rate. SKU 0% 10% 20% 30% 40% 50% Annual Web 2.70% 2.85% 3.15% 3.50% 4.75% 6.25% Annual iOS 3.20% 3.35% 3.50% 3.85% 4.25% 6.00% Annual Android 2.40% 2.60% 2.75% 2.90% 3.15% 4.50% Monthly Web 3.20% 4.00% 4.25% 4.40% 4.70% 5.50% Monthly iOS 3.50% 4.05% 4.15% 4.35% 4.50% 5.00% Monthly Android 3.40% 4.00% 4.10% 4.30% 4.45% 4.75% Mar 2025. Source: Company documents. This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to  520-031 -20- Exhibit 9 Weekly Click Volume and Estimated Effect on CPC Weekly Click Volume and Estimated Effect on CPC SKU 100 2,000 4,000 6,000 8,000 10,000 12,000 14,000 16,000 18,000 22,000 Annual Web $ 0.98 $ 1.40 $ 1.82 $ 2.66 $ 3.50 $ 4.97 $ 6.22 $ 8.40 $ 12.60 $ 22.40 $ 42.56 Annual iOS $ 0.65 $ 1.11 $ 1.17 $ 1.30 $ 1.30 $ 2.08 $ 2.60 $ 4.55 $ 6.50 $ 11.05 $ 23.14 Annual Android $ 0.55 $ 0.83 $ 0.94 $ 1.10 $ 1.10 $ 1.65 $ 2.20 $ 2.97 $ 4.40 $ 6.60 $ 12.10 Monthly Web $ 0.38 $ 0.71 $ 0.94 $ 0.99 $ 0.99 $ 1.26 $ 1.54 $ 1.99 $ 2.82 $ 3.76 $ 6.96 Monthly iOS $ 0.30 $ 0.53 $ 0.68 $ 0.71 $ 0.71 $ 0.83 $ 0.90 $ 1.08 $ 1.20 $ 1.65 $ 2.10 Monthly Android $ 0.26 $ 0.46 $ 0.59 $ 0.65 $ 0.65 $ 0.75 $ 0.83 $ 0.85 $ 0.88 $ 1.08 $ 1.50 Mar 2025. Source: Company documents. Note: As Exhibit 9 and the accompanying figure shows, as a marketer attempts to garner more clicks, the cost per click goes up, because it is harder to reach beyond those initial enthusiasts and attract new potential customers to click on the advertisement. Conversely, however, as the discount rate goes up, more consumers would be tempted to click to explore the advertising offer, and hence cost per click will decline. Gupta Media estimated the following: For discount rates of 10%, 20%, 30%, 40% and 50%, the CPC would drop by approximately 5%, 8%, 12%, 17%, and 20% respectively from the numbers shown in Exhibit 9 (and the chart). This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to  520-031 -21- Exhibit 10 Projected Performance at 30% Price Discount (In interpreting the experiments, Jordan Maddock assumed that the trial rates would hold steady in the range of clicks (4,000 to 22,000) that were being tested. The CPC numbers, however, have been adjusted down by 12% from Exhibit 9 as explained by the footnote b) in Exhibit 9). Discount Trial Advt. Click Advt. Ad* Total* Total Estimated Estimated Product Platform SKU CPC Advt. Cost ROAS Amount Rate Volume Trials Trials % Trials CPA LTV per Trial Revenue Annual Annual Web 3.50% 4,000 $1.62 $6,479.20 140 44% 318 $20.36 $94.11 $29,944.44 362% Web Annual Annual iOS 30% 3.85% 6,000 $1.14 $6,864.00 231 44% 525 $13.07 $37.55 $19,713.62 187% iOS Annual Annual Android 2.90% 6,000 $0.97 $5,808.00 174 44% 395 $14.69 $22.88 $9,047.34 56% Android Monthly Monthly Web 4.40% 12,000 $1.37 $16,464.29 528 44% 1,200 $13.72 $42.22 $50,668.80 208% Web Monthly Monthly iOS 30% 4.35% 18,000 $1.45 $26,136.00 783 44% 1,780 $14.69 $19.55 $34,791.89 33% iOS Monthly Monthly Android 4.30% 22,000 $1.32 $28,943.00 946 44% 2,150 $13.46 $10.39 $22,334.20 -23% Android Totals - 4.12% 68,000 $1.33 $90.694.69 2,802 44% 6,368 $14.24 $26.15 $166,500.00 84% Mar 2025. Source: Company documents. Note: * User Trials were both organic and that attributed to digital advertising. Total Trials is the number combining both. The proportion due to advertising varied from 40% to 48%, and is averaged to 44% for convenience. This document is authorized for use only in Dr Meng-Mei - Maggie CHEN's CHEN - Case 2507 Marketing for hospitality and service firms / MiHM at EHL Haute Ecole SA from Sep 2024 to  520-031 -22- Exhibit 11 Projected Performance at 50% Price Discount ((In interpreting the experiments, Jordan Maddock assumed that the trial rates would hold steady in the range of clicks (4,000 to 22,000) that were being tested. The CPC numbers, however, have been adjusted down by 20% from Exhibit 9 as explained by the footnote b) in Exhibit 9). Discount Trial Advt. Click Advt. Ad* Total* Total Estimated Estimated Product Platform SKU CPC Advt. Cost ROAS Amount Rate Volume Trials Trials % Trials CPA LTV per Trial Revenue Annual Annual Web 6.25% 4,000 $1.46 $5,824.00 250 44% 568 $10.25 $79.44 $45,138.89 675% Web Annual Annual iOS 50% 6.00% 6,000 $1.01 $6,084.00 360 44% 818 $7.44 $30.56 $25,006.70 311% iOS Annual Annual Android 4.50% 6,000 $0.86 $5,148.00 270 44% 614 $8.39 $17.70 $10,860.34 111% Android Monthly Monthly Web 5.50% 12,000 $1.23 $14,799.36 660 44% 1,500 $9.87 $30.16 $45,240.00 206% Web