DSP Segmentation

We’ve released the agenda for MarketectureLive. We’re deep-diving on identity, AI, and more.

Marketecture Podcast: Ron Jacobson on the current state of measurement

Ron is the CEO of Rockerbox, an independent measurement company. He takes us through the currrent state of measurement, from Media Mix Models (MMM) to incrementality.

DSP Segmentation

Last week a lot of us were confronted with the news of StackAdapt’s enormous $235 million funding round (on a $2.5 billion valuation) and our natural inclination was basically WTAF?

Here’s was my hot take, which aged poorly (but I’m all about transparency here!):

By the numbers, StackAdapt is reported to have $500 million in net revenue, and about $150 million in net operating profit. These numbers are extraordinary for the DSP business and makes them one of the most profitable companies in the sector.

Profitability aside, how big are they compared to the competition? In general I like to evaluate DSPs based on gross media spend since that removes differences in business models from the evaluation and gives us the closest estimation of actual market share. Using my scientific back-of-the-envelope estimate of a 20% take rate, a $500 million net revenue would yield a gross media number of $2.5 billion for StackAdapt, which indeed would make it the fourth biggest DSP after DV360, TTD, and Amazon. As I stated in my tweet above, that doesn’t really comport with what industry observers see in the market — SA doesn’t feel like the fourth largest DSP.

Which brings me to the subject of this newsletter, segmentation. We often forget that the advertising business is enormous (est $1 trillion globally) and even if you hone in on digital advertising it’s still huge. Hone in further to digital display advertising and its still huge. Hone in on just the DSP market, and its still pretty huge. My SWAG for the size of all DSPs (ex-China) combined is ~$35 billion in gross media.

Within this huge market are many varieties of companies that may compete with each other, or may not, depending on the segment of customer they target. Different segments have different economic structures, such as take rates and transparency, and that’s perfectly fine and good for everyone.

Some of the segmentation break-points in the DSP market:

• Large, traditional brands: Use holdco agencies, prefer leading DSPs, push for transparency and low take rates

• Regional agencies: Need managed services, not take-rate sensitive

• CTV buyers: Specialized data and reporting for CTV, access to premium inventory

• Mobile app publishers: Need ease of use, creative services, and measure everything based on ROAS

The business also segments geographically — it is pretty much impossible to use a single DSP across the globe.

Which brings us back to StackAdapt. My understanding is they focus on the regional agency and small business segments of the market, and as such they likely have a much higher take rate than we expect from a DSP. If their take rate was, say 50%, that would yield a gross media total of $1 billion, which would still be very big but would put them into the same rough marketshare bracket as the other second-tier DSPs, including Yahoo DSP, Beeswax, Basis, MadHive, and Simplifi. Big, but not unreasonable. Also, if you accept my SWAG of $35 billion, then SA’s market share would only be 2.85% — a lot of upside in the DSP business.

Are some segments better than others

In a separate thread on Twitter/X there was a discussion of what Yahoo DSP would be worth as a spin-out given the valuation of StackAdapt. The poster (who I apologize, but I cannot locate) was implying that Yahoo’s asset would be worth far more. Maybe, maybe not.

The assumption underlying the comment was (I think) that Yahoo would have a higher relative valuation because it is in the more exciting segment of the market — the mainstream agency-centric segment — where they are presumably chipping away at the Trade Desk and other DSPs. Maybe, maybe not.

Without knowing anything about the financial situation of Yahoo’s DSP (do I use the possessive or not?) I can paint two different pictures of the value of that asset vis-a-vie StackAdapt.

Just the numbers

If you’re a PE firm or other financial investor you would look at key drivers to the economics of these businesses:

• Gross media spend

• Spend growth

• Take rate (and trend)

• Net profitability

Given the numbers we’ve seen for StackAdapt, I have a very hard time believing that Yahoo DSP would stack up.

The synergies

Of course, not every company is valued purely on financials, and this is where the segmentation comes to play. Is there more opportunity available to a buyer by owning a broad-based agency-centric DSP like Yahoo, vs one that sells to smaller and mid-sized clients?

Maybe. While the DSP business on its own is attractive (fast growth, big market, opportunities for profitability), it has not generally been a spring board for horizontal expansion. I can’t think of a good example where a DSP has successfully expanded into related markets (DMP, CDP, rich media, creative management, CRM?). I also cannot really think of an acquired DSP giving its new owner cross-product synergies, with the exception of Google’s DV360 getting financially bundled with other products.

My overall take is that the main difference in investment attractiveness between DSP segments is driven by their total addressable market (TAM) rather than some inherent value in some kinds of customers over others.

Shouldn’t this all consolidate?

Another somewhat lazy point of view on the DSP market is that the leaders will eventually take all and there will be consolidation. This more or less happened over the past ten years within the traditional, agency-driven mainstream DSPs, and much of TTD’s growth has been at the expense of companies like MediaMath, Turn, Rocketfuel, etc. who all fell by the wayside, each seemingly for totally different reasons.

Those companies that TTD vanquished, though, were direct competitors within the same segment. Crossing segments is much, much more difficult, since the product requirements are not the same, the sales processes are different, the economics are different. Unless something forces consolidation (access to capital, access to data, regulations, etc), there’s no reason to expect it.

The bottom line: if you find a unique set of customers and service them, you can do pretty well in this business.

Reading list

• StackAdapt raises $235 million (!!) and has revenue of $500 million (!!) at a $2.5b billion valuation (!!!) (link)

• tvScientific Series B (w/ participation by Roku (link)

• Outbrain-Teads closes and the new name is Teads (link)

• Coinbase acquires Spindl, on chain attribution company (link)

• Trump kills de minimus exception to tariffs – Temu RIP? (link)

• Steve Yap leaves google to become CRO of Perion (link)