Description

The Trade Desk (TTD) offers a demand-side platform (DSP) used by advertising agencies for programmatic buying of digital ad inventory. Programmatic ad buying involves using software to automate traditionally manual processes. It allows for real-time auctions of ad inventory. The company went public less than one year ago.

TTD primarily makes money by taking a percentage of the gross ad spend on their platform; in 2016, TTD’s revenues represented 20% of their gross ad spend. The company is growing rapidly: revenues increased from $45mm in 2014 to $203mm in 2016. TTD possesses equally impressive margins: 2016 GAAP EBIT margins were 28%.

Background on DSPs:

TTD is one of at least 40 DSPs. Management claims they are the largest independent DSP and are taking share. Indeed, data from Magna Global implies 5-6% of programmatic spend in 2016 occurred within TTD’s software, up from 3-4% in 2015. DSPs principally differ in: (a) the exclusivity of and scope of their inventory; and (b) their access to data, the use of which refines ad buying.

They secondarily differ in (c) ease-of-use, (d) technology, (e) whether they are a managed service or self-serve, (f) their customer service, and (g) whether they sell only to agencies (e.g., Omnicom, Publicis) or to both agencies and advertisers (e.g., Nike, P&G). Today, DSPs do not differ significantly in their take rates of gross spend through their platforms (discussed further under risks).

Inventory: Buyers can only bid on coveted, exclusive inventory (e.g., YouTube, Facebook, and Hulu) through each property’s parent company. On the other hand, based on conversations with competitors and customers, incumbent omnichannel DSPs without exclusive inventory, of which TTD is one of five, do not meaningfully differ in terms of inventory. (The other incumbents are MediaMath, Rocket Fuel, Turn, and DataXu.) Some DSPs act as point solutions, focusing on inventory and performance within particular channels (e.g., TubeMogul, acquired by Adobe, for video). Notably, other DSPs—e.g., Google’s DoubleClick Bid Manager (DBM)—allow a trader to not only purchase exclusive inventory but also purchase omnichannel inventory.

Data: Proprietary data can enhance ad targeting. According to their websites and corroborated by our industry conversations, DBM, Amazon Advertising Platform, Yahoo’s Brightroll, and AOL ONE possess proprietary data. We did not find evidence that TTD offers proprietary data at this time. The data available in TTD’s software are third-party data (also available in other DSPs) and, with permission, the advertiser’s first-party data. According to a programmatic trader, the first- and third-party data themselves impact ad campaign performance more than differences in DSPs’ use of them.

Going forward, buyers will likely continue to use DSPs that possess exclusive inventory and data. Agencies we spoke to view other incumbent omnichannel DSPs (including TTD) as more interchangeable.

Total addressable market (TAM):

Defining TTD’s TAM as the amount of programmatic spend available to them, their TAM depends on three factors, the first of which is likely to be a tailwind and the latter two of which are likely to be headwinds:   

Growth in programmatic spend. According to Magna Global, programmatic spend is expected to grow from $19bn in 2016 to $42bn in 2020 (22% CAGR). A risk (discussed below) is that newly available inventory, particularly in video, becomes exclusive for purchase through DSPs owned by media companies.

The amount of programmatic spend processed by DSPs with exclusive inventory/data vs. omnichannel DSPs without. Advertisers and agencies are more interested in exclusive inventory unavailable through omnichannel DSPs such as TTD. As one industry source told us, advertisers scrutinize the performances of Google and Facebook campaigns, which represent the majority of their digital advertising spend, whereas campaigns for non-exclusive inventory receive less attention as long as the campaign numbers are met.

The percentage of programmatic spend managed by agencies as opposed to advertisers in-house. One industry source said 75-80% of media purchases today (programmatic or not) flow through agencies. Agencies are under pressure to justify their fees. One industry source estimates agencies took 60% of their advertisers’ programmatic media spend five years ago, whereas they take 25% today. Another expects advertisers will bring programmatic trading in-house, leaving agencies primarily with responsibility over creative work, not trading. Since TTD focuses on agencies only, migration of programmatic spend from agencies to advertisers would decrease TTD’s TAM (discussed more in risks below).

Competition with other omnichannel DSPs:

TTD has primarily been taking market share, it seems, from other omnichannel DSPs lacking exclusive inventory or data. In 2016, while global programmatic spend grew 27% according to Magna Global, TTD’s revenue grew 76%. Based on conversations with competitors and public disclosures, MediaMath and RocketFuel grew year-over-year in 2016 at <5% and (7)%, respectively. It appears Turn has been laying off employees, while growth at DataXu is slower than TTD’s.

Conversations with competitors, agencies, and others in the industry suggest TTD’s excellent customer service and commitment to serve agencies only (i.e., TTD will not go around the agencies to directly sell to the advertiser) differentiate TTD from other omnichannel DSPs. Customer service matters because traders, who are typically recent college graduates, require significant training and because issues inevitably arise during a campaign. Multiple people told us the price of third-party data in TTD’s platform is meaningfully lower (<1/2x) that of competitors’.

According to its 2016 10K, TTD enjoyed a 95%+ client retention rate in 2015 and 2016. It appears customers who leave TTD are those who no longer want to employ programmatic buying anymore, rather than those who switch to a new DSP.

What are the switching costs? There are two types of switching. The first type involves a customer’s reallocation of media spend between existing DSPs, which is not difficult assuming the same functionality exists in the new DSP as the previous DSP. The second type involves onboarding a new DSP in lieu of an existing one. In this case, the time needed to recollect first-party data is the most meaningful switching cost. First-party data cannot be transferred between DSPs, and collecting a sufficient sample takes two to four weeks. Meanwhile, however, the newly onboarded DSP can run the 40-50% of campaigns that do not employ first-party data. Training staff on a new DSP takes up to one week. 

What are barriers to entry? On earnings calls, TTD’s management has claimed that only with scale can a DSP afford to monitor all inventory available on SSPs and exchanges.

Key risks:

Incumbents may cut take rates. On earnings calls, TTD’s CEO guided to TTD’s take rate falling over time, attributing it to volume discounts and increasing spend on TV. But it could fall further. We learned that agencies would use one DSP’s price cut to negotiate price cuts elsewhere. It appears the agency holds more negotiating power than TTD. Most people we spoke with expect take rates to decline. One industry source expects the take rate to fall to high single digits. That said, we found no evidence at this point that any major DSP plans to undercut on price. One individual said Amazon’s DSP is priced in-line with the market.

Competition from a TTD clone. As mentioned, TTD stands out among omnichannel DSPs without exclusive inventory and data because of its agency-only focus, customer service, and cheaper third-party data. Only the last seems particularly difficult to replicate. It’s unclear how TTD was able to negotiate access to cheaper third-party data.

Consolidation of media spend on DSPs that not only have exclusive inventory or data but also omnichannel inventory access. Since buyers prefer to use as few DSPs as possible and it’s possible to purchase the same non-exclusive inventory through DBM as through TTD, why won’t buyers use DBM for everything?

Four industry sources we spoke to do not believe all programmatic spend will consolidate on DBM. They think there will be room for an omnichannel DSP that lacks exclusive inventory or data. Two supporting reasons: (a) Trader preferences matter, and DBM is not categorically preferred to all other DSPs. (b) Buyers are reluctant to cede even more negotiating power to Google in the form of media spend that could be run through another platform.

However, none of the four individuals, when asked about consolidation of spend on DBM, believe that buyers want to avoid DBM due to the possibility that DBM preferentially buys Google inventory. In fact, Google uses its position as a supplier of inventory to entice buyers to spend through DBM over other DSPs.

Furthermore, Google’s position as both an SSP and a DSP makes the buyer more confident that their inventory is non-fraudulent. Pure DSPs, on the other hand, are known for having more fraud.

If the vast majority of programmatic media spend goes toward DSPs with exclusive inventory (e.g., Google, Facebook), omnichannel DSPs without exclusive inventory (such as TTD) may be left with an uninspiring TAM.

Customers may move away from TTD if they do not sufficiently focus on developing:

1. Machine learning (ML) and artificial intelligence (AI) tools for automating campaign management. Based on conversations we had, TTD does not stand out currently on the basis of its ML tools. >50% of the people we’ve spoken to think AI will play a significant role in programmatic buying within the next 3-5 years.

2. Effective cross-device targeting. One source claimed that performing cross-device targeting well would require an incumbent to rewrite its entire architecture. Another claims cross-device targeting is comparable between DBM and TTD today; however, by virtue of its data sources (e.g., Chrome, Android, Gmail), Google is better positioned than any TTD integration partner to track a single user across devices.

3. Offsetting these concerns is the fact that TTD is now one of the largest DSPs and can afford to spend more on R&D than smaller DSPs.

TTD’s strategy is to sell only to ad agencies, a customer base which is itself under pressure to justify its fees to advertisers. Additionally, if ad trading moves away from agencies and toward in-house brands, TTD’s growth prospects would be affected if it maintains its promise to sell to agencies only.

Valuation:

At $52, based on consensus estimates for 2018, TTD trades at ~6x TEV/revenue and ~20x TEV/EBITDA, a not unreasonable valuation for such a rapidly growing, high margin business if one can get comfortable with the risks elaborated above. 

Catalyst

Earnings releases detailing continued market share gains and margin preservation.