Description

Amidst an unusually strong year for the ad tech industry, Rubicon Project quickly decided to merge with Telaria in an all stock deal. Using/accepting stocks as currency is a telling sign. Both companies have been extensively marketed by bankers before. Telaria is the old Tremor Video, which would have been sold in entirety if there were buyers. But no sophisticated buyers showed enough appetite. Tremor ended up selling its DSP business and retaining its SSP. Alas, times have changed. Riding the connected TV buzzword, the rebranded Telaria is now a stock market darling trading at close to 9X revenues. Rubicon, on the other hand, following several years of near death experiences, stabilized revenues for now and re-emerged as a seemingly real business. Bulls even start to salivate at the “inherent operating leverage”. Now with the Telaria/CTV bandwagon the stock is shooting for the moon. While I applaud the herculean tasks that both companies were able to pull off, I would like to kindly remind the bulls that swimming naked remains a civil act only when tides are high. And tides do recede - sometimes sooner than you think. 

“The Trade Desk of the sell-side” is a myth. I am not an expert in the Trade Desk (TTD), but at 20X revenues it is a little too cheap for my taste. But because of its valuation, everybody in the ad tech dog race aspires to be a TTD. The Rubicon/Telaria duo of course would advertise the deal as such - except that post header bidding, the sell-side became entirely commoditized and having a customer no longer means the same thing as in the good old days. In the old days (waterfall days), publishers used several SSPs concurrently, but bids were awarded according to historical performance (price, yield, etc). The winners kept winning and Rubicon had some happy times. Then it missed a beat and tripped over when publishers, hungry for cash, ran to header bidding, even at the expense of sacrificing user experience. Now SSPs compete for EVERY single impression, real time. Rubicon, hemorrhaging market share, rushed to build its own product which was once called Fastlane. Then they stopped mentioning Fastlane altogether. What happened? Xandr Monetize (the rebranded AppNexus’ SSP business following its acquisition by AT&T) did the unthinkable by open sourcing its header bidding product (now called Prebid). AT&T has the money and is free to experiment, afterall. The Prebid wrapper costs $0 and publishers adopted it in droves. Rubicon painfully realized that it had to recognize a permanent loss of capital and joined the Prebid alliance. 

Whereas there was no clear rationale to use multiple DSPs over the long run (except for perhaps disliking Google), multihoming is the norm in the SSP world - especially post header bidding. An average publisher uses six SSPs to maximize their revenues. Google (DoubleClick) and Amazon (who joined the race recently and swept the market in Amazon fashion) are clearly the market leaders, and the rest (AppNexus, Rubicon, OpenX, MoPub, PubMatic, Oath, SpotX....) are stuck in a race to the bottom. The industry shivers every single time DoubleClick (Google) changes its bidding mechanism, sometimes they find themselves in better positions and sometimes worse. 

Critically, Rubicon charges 13% of takerate on average (which has been going down). Sell side commissions at this level in other industries, especially those with high volume transactions, is unthinkable. If you look around, commission charges across all industries are coming down because of good old rules of technology and competition. Investment banking, a typical type of white glove high touch transactions, charges 3% to 7%. The residential real estate industry - decent value/decent volume - charges 1% to 3% (and is going down). Stock brokerage - high volume low value added - just became an industry with 0% commission. SSPs essentially provide the plumbing work in the labyrinth of tech stack. And it only provides plumbing for one side of the house. The fees will come down. 

Browsers are finally turning the screws on third party cookies. Early January, Google dropped a bomb in the industry. It will phase out third party cookies on Chrome in two years. Third party cookies provide a more granular view of the audience. The more detailed the profile of that eyeball is, the higher the CPM. Since Chrome is the number one browser in most markets, eliminating third party cookies poses a serious challenge for publishers. When Safari introduced Intelligent Tracking Prevention (ITP), CPM for those eyeballs dropped by 60% because of loss of more accurate targeting information. Similarly, when Firefox introduced similar features, observed CPMs fell off a cliff in Germany (where Firefox is the #2 desktop browser after Chrome). Killing third party cookies in Chrome will pose an existential threat for a lot of smaller publishers. While the companies will tell you that they have been preparing for this moment for years and will be prepared when it happens, the truth is, they have nothing yet. To quote Rubicon’s own Carol Starr: “So far there are no truly viable cookie-less ID solutions at scale.” CPMs for independent publishers (bread and butter of SSPs) will drop, period. 

Google is also forcing out intrusive video ads in Chrome. Recently, the Coalition for Better
Ads announced a new set of standards. “The Coalition has announced that website owners should stop showing these ads to their site visitors in the next four months. Following the Coalition’s lead, beginning August 5, 2020, Chrome will expand its user protections and stop showing all ads on sites in any country that repeatedly show these disruptive ads.” So what are the examples of “intrusive video ads”? a) “Long, non-skippable pre-roll ads or groups of ads longer than 31 seconds that appear before a video and that cannot be skipped within the first 5 seconds.”; b) Mid-roll ads of any duration that appear in the middle of a video, interrupting the user’s experience; c) Image or text ads that appear on top of a playing video and are in the middle 1/3 of the video player window or cover more than 20 percent of the video content. The bad news is that these happen to be some of the more “viewable” ads and might enjoy a higher CPM. And by the way, much of Rubicon and Telaria’s growth came from video. Importantly, together with killing third party cookies this represents just one of the many ways of getting screwed when you directly compete with Google. What multiples should you pay for these businesses? 

New regulations are coming, for real. California introduced their California Consumer Privacy Act (CCPA), which just took effect 1/1/2020. While it took 18 months for GDPR to kick in, CCPA goes into effect immediately (starting July 1, companies can be sued retroactively for violations dating from Jan 1). CCPA represents a monumental shift as it is the first of its kind in the US. Some advertisers have already pulled campaigns from California (a large economy) to evaluate the fallout. 

While GDPR only represented a drizzle so far for ad techs and publishers, CCPA will be a tsunami. For example, any company that sells California residents’ personal information (the definition of a sale is very broad) is required to put a “clear and conspicuous” link on its homepage titled “Do Not Sell My Personal Information” for people to request the company to stop selling their information. The law mandated a $7,500 fine per violation per person, with no upward limit. In contrast, GDPR sets a limit which is 4% of global revenues or EUR 20MM, whichever is higher. Thus if you are a small company (most ad techs and most publishers are), you will be disproportionately affected under CCPA. Additionally, any person who lives in California can sue the violator. If you are a real cynical person, it is not unthinkable to imagine Google and Amazons of the world could secretly hire people to sue their competitors out of business. Well - don’t be cynical, people are typically nice and God-loving in California. 

Some people will argue that CCPA is only for California, but they failed to recognize that: 1) it is hard to not have anything to do with California when you are a US based company; 2) the IAB is actually lobbying hard to institute federal level regulations. Why does the IAB want more regulations? Because the alternative is the nightmare scenario of dealing with 50 different state level regulations (they are coming fast). Thus it is 100% certain that the regulatory environment is quickly getting much more uncomfortable for this industry. When GDPR went into effect, half of third party data vendors went out of business because nobody wanted to deal with them before understanding their compliance position. And that was GDPR. Be prepared when the Yankees come. (Don’t say Rubicon/Telaria will emerge as the winner - you don’t know that.)

Connected TV will not be a big business. It is true that ad budget is moving towards connected TV. It is also true that Telaria has a working product that gained some market share. But CTV has far fewer inventory than what people commonly think. In a typical watching session, there is a very finite number of video ads that can be inserted without destroying audience experience. Also, running a campaign on CTV requires much less impressions as CTV provides more refined targeting than traditional TV.  Importantly, most of the CTV inventory is in those “walled gardens”. Four companies account for >75% of connected TV inventories: Netflix (which shows no ads), Amazon, Hulu, and Youtube. Many CTV buys are directly placed with publishers. With direct buys, publishers get minimum spending commitment, and can budle inventory across devices. Programmatic buys are used but usually small. Importantly, currently CPMs are priced such that ad buyers are encouraged to buy directly. CPM for direct CTV usually costs between $20 to $30; for Programmatic CTV, $30 to $40. The difference partly reflects the ad tech tax but also is an intentional pricing strategy to: 1) encourage direct buys; 2) inflate inventory pricing across the board. 

All the players are increasingly protecting their moat - Amazon bought Sizmek’s ad server business; Roku bought Dataxu and aims to build its own ecosystem. Roku also started limiting third party access to IP addresses - which is crucial as CTV sits in a cookieless environment. Third party SSPs will remain as outsiders. 

Hulu does have a two year contract with Telaria to support its programmatic private marketplace. Telaria handles some of the programmatic plugins such as viewability, brand safety verification, audience segmentation and measurement across other media. But the core tasks are owned by Hulu’s own ad server. Hulu is also not paying a lot of money to Telaria (Telaria is a $69MM business and doesn’t have a single customer representing more than 10% of revenues). If Hulu is paying or anticipates to pay anything meaningful to Telaria, they would have bought the company outright (Telaria, as we know, was available for sale). The Amazon/Sizmek deal is a recent and very relevant precedent. Remember that when the RUBI transaction was announced, Telaraia was only trading at a $350MM valuation. In the merger proxy, apparently both RUBI and Telaria explored alternatives but nobody was interested. 

If Hulu, the one of the largest sources of CTV inventory that exists today, only spends no more than a couple of million dollars each year with Telaria, how big will the CTV opportunity be? 

Lastly, Telaria’s own projections were uninspiring, according to the merger proxy: 

TLRA Projections 

Remember when you try to sell the company, it is not the moment to be modest and conservative. With that in mind, Telaria only expects to close between $13MM to $18MM of new deals every year over the next few years. Its standalone valuation today: over $600MM. 

As the stocks went up, insiders kept dumping shares. The chart below shows insider activities over the past two years for Rubicon. For Telaria, there has been no significant activity, but it is noteworthy that Canaan Partners and W Capital, both early backers of the firm and thoroughly understand Telaria’s tech stack, sold their shares in 2018/2019. W Capital, in particular, sold their stocks back to the company at $2.53/share in December 2018. Of course they look very stupid in retrospect, but I sympathisize with them as nobody would have anticipated how crazy the public market could have become.  

Rubicon Project Insider Sales

Based on the current valuation, the combined company is trading at more than 5X revenues and over $1.3BN market cap, whereas in reality it should trade at 1-2X revenues. What is the last time you see a private market ad tech transaction valued more than $1BN? When Sizmek filed for bankruptcy in 2019, its ad server business (#2 after Google) was sold to Amazon for $30MM to $40MM, and its DSP/DMP business was sold to Zeta for a mere $15MM. In the same year, Roku bought Dataxu (DSP business) for $150MM. In 2017, Rocket Fuel was sold to Sizmek for $145MM (for those who wonder how did the enlarged Sizmek end up with so little value in a bankruptcy, yes, it was a total disaster). The same year, YuMe, after a long soul searching journey with no credible buyers, merged with Rhythmone for about $150MM. Also the same year, Tremor (Telaria’s predecessor) sold its DSP business for $50MM. One year earlier, in 2016, TubeMogul, one of the last deals predating the ad tech crisis, was sold to Adobe for $440MM. 2018 did see the AT&T/AppNexus transaction which was valued at $1.6BN, but it was an outlier and as we know, people do occasionally overpay. Gluttony used to be a sin in the industry, but people are (mostly) on diet now. 

Comp Valuation in the Proxy

Both Rubicon and Telaria have been looking for an exit for a long time. They got married because both had a good year and suddenly both were trading at ridiculous valuations that nobody else would pay. Facing mounting regulatory pressures, it made perfect sense to get together and claim a victory. They were even willing to trigger section 382 despite that both claimed they were to earn big free cash flows and supposedly big profits soon: as of 12/31/2018, RUBI had $286MM of federal and $167MM of state NOLs; TLRA, $109MM of federal and $63.1MM of state).  

Ultimately, Rubicon/Telaria are both economically sensitive businesses that will get hit hard in an eventual downturn. An industry that helps monetize remnant inventories is not where you want to hide when sh*t hits the fan. (And we are starting to find out that random viruses do hit the world, from time to time.) Also, SSPs typically have terrible working capital cycles. They get paid by the DSPs, who in turn get paid by the agencies, who in turn get paid by the advertisers. Add 1-6 months every time you add an intermediary - that’s how Rubicon and Telaria regularly end up having over 400 days of sales outstanding. Telaria’s case is more curious - not only it has a huge AR balance, it had $120MM of AP as of 3Q19 (against $69MM of revenues in 2019), and that balance has been rapidly increasing. The company has been using AP to finance its business (which so far generates no real EBITDA). When the market tightens, the new couple will have to figure out their family finance solutions together or otherwise they will face another soul searching moment soon. Sadly, marriages do fail more often than naught.

Catalyst

When both companies announce their 2020/2021 outlook.