#91: Open Web Market Sizing
An imperfect exercise, but good enough nonetheless.
AdTech Economic Forum (ATEF) 2025: The Davos of AdTech is returning to The New York Times Center on March 19, 2025. We are curating an experience like no other in adtech. Secure your spot today at the intersection of adtech, economics, and finance!
The topic discussed today was shown at ATEF last April and caused quite a discussion. Why? Because at ATEF enlightenment our economic unit of production. We produce it with four raw materials: Content, Conversation, Connections, and Collaboration.
On August 27, eMarketer sent out its chart of the day. The chart illustrates various buckets of ad spending across the digital and traditional divide.
In this post, Quo Vadis takes eMarketer’s high-level view as a starting point and enhances it into a granular and relative view of media money flows. The intention is to show our readers where the dough is tending to increasingly flow and what it means for the “Open Web,” venture investment, and M&A.
Like all things in advertising land, market sizing is an imperfect exercise. We hope you find our approach instructive enough to be useful.
Rule of Thumb: At the end, ask yourself: “How wrong could we be and by how much does it change our answer?
The Main Thing
The main thing we are curious about is estimating the absolute size and relative size of what industry folks generally refer to as “open web” display ads — aka boxes on websites and in apps.
This topic is critically important because when people talk about “adtech” and adtech companies, they are mostly talking about the planning, activating, processing, and monitoring of ads that get served to boxes on websites (and apps).
These are the companies we track in our AdTech 18 equal-dollar portfolio. Whether they are public or the many other non-public adtech companies, they are all competing (or trying to compete) to capture and retain a share of finite media flows.
Brian Morrissey put it perfectly on The Rebooting:
"I mean we are fighting about display advertising… I mean come on… that’s not where the market is going. The best ad products have been created by walled gardens not by this programmtic archipelago.”
Media Money Flows US
First things first. Our media money flow estimates are based on US media spending. In the future, we’ll provide a global view, but even with just a US view, we think it’s a good enough (albeit imperfect) proxy across the G20 (ex-China) markets where most ad dollars are allocated.
Total media spend and the digital divide
Let’s start at the top with $390 billion in total US media spending which is split between digital at $303 billion (78% and growing) and "traditional” which is $87B (22% and shrinking/morphing).
CTV eats at a buffet called Linear TV
CTV spending in the US is estimated at $29 billion. Not long ago, Linear TV was $88 billion ($29B + $59B) but now it’s 33% smaller mostly because of CTV cannibalization. Some day, in the distant future (roughly 60 years from now), linear TV viewership will decay to near zero and CTV (or something entirely new) will be considered “traditional.”
Yes, you heard that right. Theory and observation suggest that what goes up will eventually come back down. But how long will it take?
For example, adoption curves tend to be durable in the sense that the rise up to peak adoption (e.g. TV viewing circa 2019) started in earnest 83 years ago in 1941 with the very first TV ad for Bulova watches. If adoption decay from the peak happens as it normally does, then we estimate that will take several decades for linear TV to fall to zero as CTV displacement continues to evolve.
Other traditional spending areas
The rest of traditional media spending flows to radio, out-of-home, and print totalling $27 billion or 7% of total US media spending. That’s roughly the same share open web programmatic captures today. We’ll get to that in a minute.
Breaking out the “digital” components
Let’s start with eMarketer’s footnote about digital ad spending delineation and definitions:
Digital ad spending includes banner ads and other (static display ads such as Facebook's News Feed Ads and X's Promoted Posts), classified ads, email (embedded ads only), mobile messaging (SMS, MMS, and P2P messaging), rich media (including in-stream and outstream video ads), search ads (including contextual text links, paid inclusion, paid listings, and SEO), sponsorships, lead generation (referrals).
That’s a lot to chew on so let’s digest it one piece at a time starting with search ad spending.
Search
“Nothing last forever and we both know hearts can change.” Guns & Roses, November Rain.
Classic search ad spending in the US is estimated at $90 billion. We use the word “classic” because there are other types of “search-like” spending taking place across the digital ad scene like in retail media. We’ll cover that in a minute too.
Google’s FY23 global search advertising revenues were $175 billion (See FY23 10K page 63). Across Google’s entire portfolio of business lines, 48% is sourced in the US so we estimate Google search revenue in the US at $83 billion. Bing, Yahoo, other search players, SEO, etc get the remaining $31 billion. Google captures a 93% share in the US, but for how long will its product dominance remain? Pretty much every time you use ChatGPT you’re not using traditional search. That means fewer clicks and fewer referrals to the ad-supported open web.
Display Ads
Now we’ll turn to display ads — aka banner and video ads in boxes on social feeds, websites, and apps. We split the display ad world into three buckets: Social, Retail Media Media, and Open Web.
In the US, eMarketer says social ads attract $75 billion, or roughly 40% of the display market. According to Meta’s FY23 10K (page 103), Meta properties take in $53 billion in the US which is 72% of social display ad spending.
TikTok is the next biggest player with an estimated $16 billion (and rising) which is 20% of the social display ad market.
Short video platform TikTok is on track to overtake Meta, after reports that the company made $120bn worldwide in 2023, up 40% year-on-year, with $16bn of that coming from the US alone as it faces legislation that could force the sale of the company in order to continue operating in the country (Source: WARC).
After Snap Chat with $2.9 billion in US media flows, (FY23 page 88), that leaves around $3 billion and change for all the other social players like X, LinkedIn, etc.
Retail Media is not as straightforward to delineate. In September eMarketer estimated:
Walmart will claim 6.8% of all US retail media ad spend this year, amounting to $3.72 billion this year, according to our March 2024 forecast.
If eMarketer is close enough, that means the overall US retail media captures $55 billion in media flows ($3.72 ÷ 6.8%).
Amazon is the biggest player attracting $38 billion in retail media (aka commerce media) flows or 13% of total digital ad spending in the US and 67% of the retail media category.
Walmart and a long list of other onsite and offsite retail media options capture the remaining $17 billion. This is the space where Criteo increasingly winning and dominating.
The two most exciting things about this space are the precision of targetability across deterministic audience IDs and the total imprecise ability to attribute sales to one player versus another when several are on the media plan.
Understanding relative marginal contribution across many retail media endpoints is not overly difficult to accomplish. We wrote a how-to piece suggesting Shapley Values in January 2023. However, arriving at measurement nirvana does not happen without sacrifice. Any marketer bold enough to pause all retail spending for around 30 to 60 days (to cleanse conversion data) and restart with a Shapley state of mind will reach total enlightenment.
The Infamous Open Web
With everything spoken for across the diverse display ad market, we can now turn to what’s left over — Open Web display ads in little rectangular boxes. If our trail of preceding estimates is close enough, the open web can be deduced to be around a $48 billion market or 12% of total media spend in the US.
The first tranche to account for across open web ad spend is direct spending. That’s when a buyer buys guaranteed impressions (non-biddable) at a specific CPM to run at a specific time. We estimate direct spending to be around 36% of open web spend or $17 billion.
How do we get there? We assume that Jounce Media’s Top 100 “Bellwether” sellers represent around 90% of open inventory which is $43 billion. We then estimate on average a publisher sells 40% of all available inventory in this manner, which is $19 billion out of $53 billion.
“Gannett makes $330 million from programmatic and pays Google about $15 million for ad tech. Switching ad servers isn’t worth losing revenue. Roughly 25-40% is direct sold, rest is programmatic.” DOJ vs. Google
The interesting thing about direct-sold display ads is buyer rationale in the past vs now and in the future. In the past, direct buys we bought using high-level obtuse context signals. When 1-1 programmatic audience targeting came along, a lot of direct spending got funneled to the DSP-SSP infrastructure which was built on 3rd party cookies. Now that 3rd party cookies are gone, and targeting has moved closer to publishers, multi-pub direct buys with audience fidelity on a single campaign buy are not only possible but happening. That’s an underappreciated very big deal with massive ramifications across the ecosystem.
The remaining $31 billion is sold in programmatic auctions between an SSP (bid request) and DSP (bid response) or directly between a Buyer-DSP-Publisher setup with zero SSP involved or between a Buyer-SSP-Publisher setup with no DSP involved. This auctioned inventory could be bought and sold in open market auctions or private marketplace deal IDs which are getting increasingly produced by the flavor du jour called “curation.”
We think about the economics of curation as a classic surplus value proposition. For example, the Trade Desk consistently captures 20% take rates from gross ad spend. Let’s say they charge 10% for DSP tech fees and the other 10% comes from selling data which is an endemic form of curation. When a buyer uses a third party to curate “unique” or purportedly differentiated supply packages, 10% is freed up. The curation producer gets the lion’s share and, in theory, gives something back to the advertiser which increases working media.
In any case, the frenzy around curated deal IDs is a market response to the signal loss from 3rd party cookie deprecation which can potentially have profound effects on who gets what across the open web.
The Programmatic Open Web Pie
Let’s dig down a bit further to illustrate how the programmatic pie is getting split up. We’ll start with Google. Before the DOJ’s anti-trust trial, we would have had fuzzy data at best to make an educated guess about Google’s open web display revenue. The trial revealed documents that give us a much better idea.
Our media money flows analysis only cares about Google’s demand-side display ad businesses — Google Ads and Google’s DSP DV360. According to court documents, total demand-side gross ad spend in 2020 was $19.5 billion.
Looking at Google’s FY23 10K, we know that the total US revenue across all business lines was 48%. If we assume 48% also applies to Google’s demand-side display ad business, then US money flows through Google Ads and DV360 in 2020 was around $9.3 billion.
According to eMarketer, the display ad market grew by 31%, 8% and 12% in 2021, 2022 and 2023, respectively. Assuming Google’s demand-side business grew with the market, a plausible assumption without doubt, FY23 display ad spend through Google Ads and DV360 in the US was $14.7 billion.
For comparison sake, Trade Desk processed $9.4 billion in gross ad spend in FY23 which means it became a strong #2 player and could reach parity with Google when we see FY24 numbers in February.
Asked who is the largest DSP, Casale said “depends on the day, but it is a coin flip between DV360 and TTD.” DOJ vs. Google
The remainder of open web programmatic spending is captured by Amazon DSP (estimated at $1.5 billion) and Criteo ($700M in the US) leaving around $5.6 billion for everyone else in a very cluttered space.
We know from Outbrain’s recent acquisition that Teads is rather sizable processing around $600 million in gross ad spend with around half coming from the US.
Based on the relative headcount (“Associated Members” on LinkedIn), we estimate Kargo’s gross ad spend to be around $420 million (might be a bit light) which is mostly US we’d imagine.
About a year ago we published a best-guess valuation on Yahoo. Our 2023 estimate for Yahoo’s display ad revenue was $4.4 billion. Keep in mind that a large portion of Yahoo’s display revenue is direct-sold and a meaningful portion is likely generated from its DSP. It would not be surprising to learn that Yahoo’s DSP is in the “unicorn club” processing over $1 billion of media flows in the US. In fact, earlier this month Yahoo DSPs was named “Best Buy-Side Programmatic Platform” by Digiday.
Zeta is mostly a DSP ad network with a data store and did $700 million in the US in FY23 (FY 2023 10K, page 67)
Roku generated $500 million of total gross ad spend in the US in FY23 and a meaningful portion is likely display and/or in/out-stream video ads as opposed to pure CTV ads.
AdForm has a similar headcount to Kargo and reported gross billings of €335 million in its 2023 annual report (USD ~$370 million). A good portion is likely US-based.
Viant’s TTM gross ad spend is $243 million, mostly sourced in the US.
Prior to going private, AdTheorent processed $224 million in gross spending which is mainly US-based.
That’s a short list and already gets us over halfway to $5.6 billion. If we keep going down the long tail of DSPs we’ll get to $5.6 billion and run out of players.
All roads lead to M&A and open web consolidation
The term "adtech" refers to the tools and platforms used to manage and deliver digital advertising campaigns. While the roots of digital advertising trace back to the early 1990s with the advent of the internet, adtech as a distinct field began to take shape around 2007-2010. That was when adtech (aka programmatic advertising and audience-targeted real-time bidding (RTB) revolutionized the way digital ads were bought, sold, and delivered.
We think it’s reasonable to set AdTech’s birth date when Right Media or DoubleClick created the ad exchange in 2007. From that point in time, adtech experienced a build-up M&A phase through ~2021 when many adtech companies went public. In between, there were more than a handful of sizable M&A transactions such as:
Oracle acquired Moat for $850 million in 2017 (build-up) only to shut it down early this year (tear down)
Private equity firm Vista bought Integral Ad Science in 2018 for $850 million and then took it public in 2021
Private equity firm Providence bought DoubleVerify in 2018 for $200 million and then took it public in 2021
In 2018, AT&T acquired Appnexus for $2 billion (build-up) which is now owned by Microsoft and called Xandr. Microsoft paid $1 billion to take Xandr off AT&T’s hands in (tear-down) in 2021.
More recently, MiQ was acquired by Bridgepoint Advisers through an LBO in September 2022 for a reported ~$1 billion. That deal happened at the peak of the build-up/tear-down cycle
The open web programmatic market is now in a tear-down M&A phase with a few big deals and many smaller deal sizes ranging from low to ok valuation multiples.
Outbrain acquired Teads for $1 billion this year
Zeta Global acquired LiveIntent last week for $250 million
Equative bought SSP Sharethrough for $82 million over the summer
Cadent took AdTheorent private for $324 million, also this past summer
DoubleVerify acquired Scibids for $121 million last summer
Last December, Perion acquired Hivestack for $100 million
Over the summer, Seedtag acquired Beachfront for an undisclosed amount. With 25 employees, a 16-year-old company should generate ~$350,000 per employee in net revenue and sell for around 2.5x or ~$23 million
As you can see, the open web adtech market appears to be right in the middle of a tear-down M&A phase. After this phase clears out, adtech will begin a new build-up era. We are already seeing it happen by looking at where venture capital is making investments into the opportunities shooting up from the fertile grounds of creative destruction.
With a new era of adtech investment dominated by AI applications, it’s a fair bet to assume increasing valuations as things unfold.
Fewer Players Across the Open Web Scene
“Open programmatic ad spend has only grown 3% since 2021, according to equity research firm Wolfe Research. Walled gardens have grown 10% in the same time frame.”
That’s what equity analyst Shweta Khajuria from Wolfe Research said at Programmatic I/O a few weeks ago in NYC. Let’s play out these growth rates over the next five years and then talk about how the open web programmatic is getting crowding out.
Today, if the $31 billion in open web programmatic spending is growing at 3% over the next five years while overall ad spend grows at 7%, then open web share will decrease from ~8% to ~6%. It’s still growing, but getting relatively smaller. If you’re looking for the main driver of the current tear-down M&A phase that would probably be a good one.
The Road Ahead for the Open Web
Traditional search shrinks while AI search expands. Who are the winners and losers?
Meta takes more share of social display (feels like a safe bet), TikTok also takes more share (if it remains viable/legal in the US), and everyone else grows but at a slower rate.
Amazon might take a bit more share of the retail media pie (and other non-advertising pies too) while the entire segment grows taking share from other buckets like the Open Web
Direct open web ad spending grows with innovation in multi-publisher audience-targeted media, Google gets out of the open web display, and Trade Desk grows and becomes the kingmaker across the open web (big fish, shrinking pond problem).
CTV grows as the sticky dependence and comfort of linear TV declines (live sports are everything) and YouTube continues to grow as a very safe bet for eyeballs and attention.
Disclaimer: This post, and any other post from Quo Vadis, should not be considered investment advice. This content is for informational purposes only. You should not construe this information, or any other material from Quo Vadis, as investment, financial, or any other form of advice.
Your webdev guides are top-notch! I started using EchoAPI for API simulation, and it’s been incredibly helpful for testing app features before full backend integration.