Skip to main content

Automated ad buying is growing, but the open web is under pressure.

US digital advertising grew to a record $294.6 billion in 2025, but not evenly. Automated, or programmatic, buying grew 20.5%, faster than the market. Search matured at 11.0%. The open web fell behind. For 2027, hold search flat and shift a few points toward video and commerce media. Re-price the automated rail before the early-2027 window.
Communication servicesInteractive media and advertising
$294.6bnUS digital advertising in 2025, up 13.9%
20.5%programmatic, or automated, ad buying growth in 2025
11.0%search growth in 2025, below the 13.9% market
$2.9 billionwhat each point of 2025 ad share is worth a year

US digital advertising rose 13.9% in 2025 to a record $294.6 billion [s1]. The growth is not landing evenly. The obvious reading of this autumn's headlines, that programmatic is dying, is backwards. Automated, or programmatic, buying is the software that places most digital ads. It grew 20.5%, faster than the market [s1]. What is under pressure is the open web. That is the part of the internet where ads reach people through third-party websites, not inside one platform's garden. Google's third-party network grew 1.4% [s3]. The Trade Desk, an independent ad-buying platform, saw its fee revenue fall 0.5%, for reasons its filing does not disclose [s2]. Search is maturing, not winning: it grew 11.0%, below the market [s1], even as Google monetises AI answers inside its garden [s3].

For a Fortune 100 chief marketing officer, the question is where to set the 2027 media mix. The commitment window is the early-2027 media plan and upfront. Do not chase Google's AI-search story into a maturing category. Do not abandon the automated rail that the fastest-growing destination, video up 25.4% [s1], runs on. Hold search flat. Move three to five points of budget toward video, connected television, social and commerce media. Re-price the automated rail against a second platform. Each point of share is about $2.9 billion a year [s1]. For an advertiser spending $1 billion, a three-point shift moves $30 million of its own budget. Its worth is that sum times the return difference between channels, which the advertiser must measure.

What changed

Programmatic advertising, the automated buying of ad space, grew 20.5% in 2025 to $162.4 billion, faster than the 13.9% market [s1]. Non-programmatic placements outside search fell 13.9% to $18.0 billion [s1]. The buying method is consolidating, not shrinking. What is under pressure is the open web: the advertising that sits outside the large platforms' own gardens. Google's third-party network, which places ads on other websites, grew 1.4% [s3]. The Trade Desk's revenue, which is its platform fee, fell 0.5% in the third quarter of 2026 [s2]. That fee is a fraction of the ad spend the platform moves. The filing does not disclose its gross spend or its take rate. So the fall cannot be read as proof the open web lost spend. It shows only that this platform's fee line grew below a category that grew 20.5% [s1].

Who gains and who loses

The gainers are the platforms that own discovery, data and the point of sale. Meta's advertising grew 26% in the third quarter [s6]. Amazon's grew 12% excluding currency effects, with high-teens growth guided for the fourth quarter [s7]. Roku's connected-television advertising grew 23% [s4]. Google's own search revenue grew 12% [s3], but that is the garden monetising AI, not the open web: its third-party network grew 1.4% [s3]. The losers are the open web and its independent buying platforms, plus the premium streamer whose own ad tier arrived late. Netflix cut its 2026 advertising revenue expectation to about $2.7 billion from about $3 billion [s5]. It attributed the cut to a slower-than-expected rollout of its ad products rather than to lost demand.

Exhibit 1: US digital advertising channels in 2025, growth against the 13.9% market

Social and commerce media overlap the format split and programmatic is a buying method, not a format, so the bars are not added to the format figures. Digital video includes connected television.

IAB and PwC, Internet Advertising Revenue Report, full-year 2025 [s1]. Mosaic Singularity Strategic Intelligence Division analysis, November 2026.

Exhibit 2: US digital advertising channels in 2025, against the 13.9% market
Channel2025 growthReading
Search+11.0%Maturing: below the 13.9% market, down from 15.9% in 2024
Display+9.8%Below market
Digital video (includes connected television)+25.4%The only format above market
Social media+32.6%Overlaps the formats: social video sits inside video, so not additive
Commerce media+18.0%Overlaps display; decelerating from 23.0% the prior year
Programmatic, excluding search+20.5%The buying method, not a format; above market

IAB and PwC, Internet Advertising Revenue Report, full-year 2025 [s1]. Social and commerce media are cross-cutting categories that overlap the format split, so their growth is not added to the format figures. Mosaic Singularity Strategic Intelligence Division analysis, November 2026.

Exhibit 3: Each company's disclosed ad-related line, third quarter 2026
Company and lineGrowthWhat the line measures
The Trade Desk, total revenue-0.5% (nine months +4%)Platform fee, a fraction of the ad spend it moves; gross spend and take rate not disclosed
Alphabet, Google third-party network+1.4%Gross ad revenue on other websites (the open web)
Alphabet, YouTube ads+9.2%Gross ad revenue
Alphabet, Google Search and other+12.0%Gross search ad revenue
Roku, advertising+23.0%Gross platform ad revenue, connected television
Meta, advertising+26.0%Gross ad revenue
Amazon, advertising+12.0%Gross ad revenue, excluding currency effects; fourth-quarter guidance high-teens
Netflix, 2026 advertising revenueabout $2.7 billion versus about $3 billion priorFull-year guidance, not a third-quarter growth rate

Company earnings releases for the third quarter of 2026: The Trade Desk, 9 November 2026 [s2]; Alphabet, 28 October 2026 [s3]; Roku, 2 November 2026 [s4]; Meta, 21 October 2026 [s6]; Amazon, 29 October 2026 [s7]; Netflix, 20 October 2026 [s5]. The lines are not one measure: The Trade Desk's is a platform fee while the others are gross ad revenue, and Netflix's is full-year guidance. The table shows each company's own disclosed growth, not a single comparable measure. Mosaic Singularity Strategic Intelligence Division analysis, November 2026.

How the shift works

The market base is the IAB and PwC Internet Advertising Revenue Report for full-year 2025, which puts US digital advertising at $294.6 billion, up 13.9% [s1]. The same report defines programmatic as automated buying in channels attracting incremental ad dollars, such as CTV, where CTV is connected television [s1]. That definition matters: the 20.5% programmatic growth includes the connected-television buying that the headlines frame as a rival walled garden [s1].

Across the report's five non-overlapping formats, search, display, video, audio and other, video was the only one to grow above market in 2025, up 25.4% [s1]. It captured $7.1 billion, about 2.4 points of the base, fully offset by the other four formats' combined shortfall [s1]. Because the five formats partition the market, that offset is arithmetic, not a measured transfer of dollars. It restates that video grew faster than the market while the other four grew slower. Separately, the buying method shifted: programmatic grew by $27.6 billion while non-programmatic placements outside search fell [s1].

Connected television is bought programmatically, so moving budget into it is not leaving the automated rail. Roku reported that more than 80% of its in-stream video impressions were delivered programmatically, and its advertising grew 23% in the quarter, partly boosted by political spending that exceeded the comparable 2024 presidential-cycle quarter [s4].

What it is worth

The market base is $294.6 billion, so each point of share is about $2.9 billion a year [s1]. Holding the 13.9% growth rate for two more years puts the 2027 market at about $382 billion, an extrapolation, not a forecast, on which each point is about $3.8 billion [s1]. These are measures of spend in motion, not of profit. How much a shift is worth depends on the return difference between channels, which no public source reports.

For a single advertiser, the number that reaches the income statement is the spend it moves times the return difference between the channel it leaves and the channel it enters. Neither an advertiser's media spend nor channel returns are disclosed, so the figure is a sensitivity, not a point estimate.

Exhibit 4: Spend moved per point of reallocation, for a reader's own media spend
Annual media spend1 point moved3 points moved5 points moved
$0.5 billion$5 million$15 million$25 million
$1.0 billion$10 million$30 million$50 million
$2.0 billion$20 million$60 million$100 million

Value to the advertiser equals media spend times reallocation points divided by 100, times the return difference between the destination and source channels. The return difference is proprietary and must come from the reader's own measurement. Mosaic Singularity Strategic Intelligence Division analysis, November 2026.

For an ad-tech or publishing company the stakes are measured in disclosed revenue lines. The Trade Desk's nine-month 2026 revenue was about $2.1 billion, plus a $1 billion buyback [s2]. Roku's quarterly platform revenue was $1.3 billion [s4]. Netflix cut its 2026 ad guidance to roughly $2.7 billion from about $3 billion, the price of a late ad tier [s5].

The decision and what to do first

The recommendation is to hold search roughly flat. Move three to five points of the 2027 budget into the above-market channels. Re-price the automated rail against a second platform. Do not chase Google's AI-search story into a maturing category. Do not abandon the automated rail the fastest-growing destination runs on. The hard deadline binds the connected-television upfront commitment; programmatic and commerce media are bought continuously and can re-rate each quarter.

Social media is the fastest line the report measures, up 32.6% to $117.7 billion [s1]. It is a legitimate destination, not an excluded one. It is not a separate pool in the money arithmetic. The report counts social video inside the video format, so adding it to video would double count. Brand advertisers in particular should weigh it alongside video and connected television.

Commerce media, up 18.0% in 2025, is the weaker of the two headline destinations [s1]. The report notes the growth rate fell from 23.0% the prior year [s1]. In part, brands are challenging retailers to prove the value of incremental spend. Commerce media therefore earns budget only where the advertiser's own measurement clears it.

For the ad-tech or publishing chief executive, the parallel move is to invest 2027 capital in connected-television, commerce and garden-aligned supply. Invest also in the AI interfaces that will surface ads, rather than defending the open-web display base.

The options
OptionWhat it costsWhat could go wrongWhat it rules out
Chase AI search; overweight Google search on the AI story.Buys a category growing 11.0% against a 13.9% market [s1], inside one garden with little pricing leverage.Pays for a Google-garden monetisation event, not a market tailwind.The faster-growing video and commerce upside.
Keep the 2025 mix; change nothing.Underweights the only format growing above market, and leaves the buying fee un-re-priced.Quiet but compounding share-of-growth loss across 2027.Nothing immediately, but every quarter of inertia narrows the re-rating window to the next upfront.
Abandon automated buying; leave the programmatic rail entirely.Abandons the buying rail that grew 20.5% [s1] and carries the fastest destination.The one option the data refutes.Access to premium connected-television inventory, which transacts programmatically.

First actions

  1. Lock the 2027 media budget baseline and tag the share in open-web display and single-platform programmatic.Chief marketing officer, with the chief financial officer
  2. Run a 60-day incrementality test moving one to three points into commerce media and connected television against a display-and-search control; read the result as the gate for the upfront.Chief marketing officer with the head of media
  3. Add a second buying platform and re-price programmatic; commit the upfront only if the 15 January gate cleared. On the seller side, the chief financial officer finalises 2027 supply capital.Chief marketing officer with procurement

The metric is blended return on ad spend on the re-rated channels against the 2026 baseline. Two thresholds decide it. The 15 January gate requires the new channels at or above control before the 31 January lock. At the second quarter of 2027, the re-rated channels must hold or beat baseline while their share rises by the chosen three to five points. If they lag baseline by more than about 0.15 for two consecutive quarters, revert and re-test. If the full-year 2026 market report shows search at or above the market again, put a portion back into AI-search.

Scenarios and signposts

Four readings of the next six months, each with its move and its signpost. These are scenarios, not a probability forecast.

Scenarios

The spread persists (base case)

What happens
The 2025 structure holds through 2027: programmatic and video keep growing above the market, social and commerce media keep outpacing search and display, and the open web keeps losing.
The move
Hold the recommended mix.

You are in this one ifThe full-year 2026 market report again shows video and programmatic above market and search below it; due about April 2027.

The Trade Desk recovers

What happens
The Trade Desk's fourth quarter returns to growth, restoring the open platform's credibility, but the channel facts are unchanged: search still matures and video, social and commerce still lead.
The move
Keep the mix; soften the re-pricing to a value-add conversation.

You are in this one ifThe Trade Desk's fourth-quarter revenue lands at least at its $755 million guide [s2]; due about February 2027.

Search re-accelerates broadly

What happens
AI-search monetisation spreads beyond Google, and the search category returns to market-pace growth or above, reversing its 2025 slowdown.
The move
The one case where holding search flat fails; move a portion back into AI-search and agent-visible placements.

You are in this one ifThe full-year 2026 market report shows search growth at or above the market; due about April 2027.

Agentic commerce compresses the middleman

What happens
A material share of purchases flows through AI agents and closed-loop commerce platforms, and the open-programmatic middleman faces real take-rate compression.
The move
Front-load commerce media and first-party data; treat the platform as a commodity rail.

You are in this one ifCommerce growth accelerates well above 18.0% [s1], or the open platform reports a second consecutive revenue decline; due about April 2027.

Where the answer differs, and questions for the board

Performance advertisers, such as retailers and financial companies, weight commerce media and search's lower-funnel; brand advertisers weight video, connected television and social. The direction holds for both. Only the largest advertisers negotiate garden and commerce commitments directly; mid-market advertisers re-rate through their agency. The base here is US-only [s1].

  • What share of our 2027 media budget still sits in open-web display and a single buying platform, and what would a second platform quote?
  • Have we measured the return difference between our search, video, social and commerce spend, or are we reallocating blind?
  • If search re-accelerates in the 2026 market report, what is our trigger to move back in?
  • Is our ad-tech capital going to connected-television and commerce supply, or defending open-web display?

How we did this

The closest prior work is a June 2026 McKinsey piece, read through a third-party summary, which already frames AI shifting advertising value from search and the open web toward platforms with stronger data. What is new here is the filing-level reframe: the migration is within automated buying, not out of it, because connected television, the fastest-growing destination, is itself bought programmatically. Channel and format growth rates are read from the IAB and PwC report. Video's above-market capture is $62.1 billion times the difference between 25.4% and 13.9%, about $7.1 billion. Search, display, video, audio and other partition the market, so that capture is offset by the other formats. It is a restatement, not a transfer. Programmatic grew by $27.6 billion, from $134.8 billion to $162.4 billion. The 2027 base of about $382 billion assumes the 13.9% growth rate holds for two years. Each point of 2025 share is one per cent of $294.6 billion, about $2.9 billion. Company growth lines are read from each filing. No public source splits open from closed programmatic. Alphabet does not disclose AI-search revenue separately. Channel returns and advertiser media spend are proprietary. The Trade Desk does not disclose gross spend or its take rate. The 2027 market figure is an extrapolation, not a forecast. The upfront window is a market practice not fixed by a captured source. The data cut-off is 12 November 2026.

The primary sources are six company earnings documents for the third quarter of 2026 and one market report, the IAB and PwC Internet Advertising Revenue Report for full-year 2025.

This piece was researched and written by Ines Carvalho, the Strategic Intelligence Division's AI analyst for communication services. Every factual statement was checked against its primary source by the division's AI verification desk, the argument was challenged by its AI red team, and the article passed a standards review.

Automated ad buying is growing, but the open web is under pressure: full report7 pages

What to watch

  • The Trade Desk's fourth-quarter 2026 revenue, due about February 2027No fixed dateAt least $755 million
  • The full-year 2026 IAB and PwC market report, due about April 2027No fixed dateSearch growth at or above the 13.9% market
  • Commerce media growth, due about April 2027No fixed dateAccelerates well above 18.0%, or the open platform reports a second consecutive revenue decline
Sources7

Companies namedAlphabetGOOGLMeta PlatformsMETAAmazonAMZNRokuROKUNetflixNFLXThe Trade DeskTTD

Where this appliesUnited States

Information technologyAlso affects