Key Takeaways and Introduction
- The best expansion channels beat the panel average by 50% to 90%. Being tapped out on Google and Meta is not the same as being out of efficient headroom.
- Snapchat and commerce retargeting DSPs pooled highest, at roughly 1.9x the expansion-channel benchmark, with Criteo at 2.3x.
- CTV offered the best combination of measured performance and scale headroom, at a 1.5x to 1.7x benchmark.
- Objective matters more than platform. Conversion-objective prospecting out-performed awareness and consideration buys on nearly every platform, often by 5x to 10x.
- The biggest expansion budgets pooled below benchmark. TikTok and affiliates absorbed the most spend and returned less per dollar than disciplined mid-size budgets on Snapchat, Reddit, and programmatic direct mail.
Every growth team hits the same wall: Google Search is capturing demand you already earned, Meta CPMs keep climbing, and the marginal ROAS on your next dollar in either platform is a fraction of your average. The question is not whether to diversify. It is which channel earns the next test budget.
Most “best channels to try” lists are guesswork. This one is measured. We analyzed the trailing 52 weeks (July 2025 to July 2026) of incrementality-calibrated performance across a panel of enterprise consumer brands representing more than $135M in measured spend on expansion channels, meaning everything except Google Search, Shopping, and PMax, and Meta, via Measured’s Cross-Channel Dashboard, calibrated with geo incrementality tests. CTV figures draw on the additional $25M-plus CTV panel from our companion piece, Top CTV Advertising Platforms for 2026.
Each channel below carries a Measured Incrementality Index: its pooled incremental ROAS (iROAS) relative to the expansion-channel benchmark across the whole panel (1.0x = panel average). Indexes are relative multiples only, and no client-identifiable data is disclosed.
This is a list, not a leaderboard. The channels are ordered to make the piece readable, not to declare a winner, because the right next test depends on your audience, your catalog, and your CRM file.
The Channels at a Glance
| Channel / Platform | Measured Incrementality Index |
|---|---|
| Snapchat | 1.9x |
| Commerce retargeting DSPs (Criteo, RTB House, AppLovin) | 1.9x pooled (Criteo 2.3x) |
| CTV: YouTube CTV and publisher-direct streaming | 1.5x to 1.7x |
| 1.6x | |
| Programmatic direct mail (PebblePost, PostPilot) | 1.6x |
| 1.1x | |
| Creator commerce platforms (ShopMy) | 1.3x |
| Affiliate networks (Impact, Rakuten, Awin, Partnerize, CJ) | 0.9x |
| TikTok | 0.7x pooled, top programs 2x to 3x |
| Podcast and streaming audio | 0.5x |
Three findings worth internalizing before the details:
- The best expansion channels beat the panel average by 50% to 90%. Being “tapped out” on Google and Meta is not the same as being out of efficient headroom.
- Objective matters more than platform. On nearly every platform, conversion-objective prospecting dramatically out-performed awareness and consideration buys of the same platform, often by 5x to 10x.
- Scale and efficiency trade off visibly. The biggest expansion budgets in our panel (TikTok, affiliate) pooled below benchmark, while disciplined mid-size budgets (Snapchat, Reddit, programmatic direct mail) pooled well above it.
Overview
The quiet winner in our data. Snapchat is routinely dismissed as a legacy social platform, but it pooled at nearly twice the panel benchmark, and conversion-objective Snapchat prospecting lines ran 3x to 8x benchmark at multiple brands. CPMs remain well below Meta’s, and its audience overlaps far less with Meta’s most-saturated segments than marketers assume.
Pros
- Highest pooled iROAS of any social platform in our panel.
- Conversion-objective campaigns performed at every brand in the panel that ran them.
- Cheap CPMs, with fast creative iteration using Meta-like tooling.
Cons
- Awareness-objective buys pooled far below benchmark, with one large brand’s awareness line indexing at roughly 0.1x. Buy for conversions.
- Audience skews young, so the ceiling depends on category fit.
Sources
Overview
The most surprising result in the panel. Commerce DSPs, long written off as last-click credit thieves, delivered incrementality-verified performance well above benchmark. Criteo pooled 2.3x benchmark, AppLovin’s ecommerce push delivered 2x including strong prospecting lines, and RTB House retargeting programs were among the highest-iROAS line items we measured anywhere.
Pros
- Top pooled indexes in the entire expansion panel.
- AppLovin prospecting proved genuinely incremental at multiple brands, not just retargeting.
- Fast to test: catalog feed plus pixel, live in days.
Cons
- Performance is bimodal. The same DSP indexed above 4x at one brand and near zero at another, so test with a geo holdout before scaling.
- Retargeting-heavy programs need incrementality validation more than any other buy, because platform attribution over-credits most here.
CTV: YouTube CTV and Publisher-Direct Streaming
Overview
From our CTV platform analysis: YouTube CTV pooled roughly 1.5x this panel’s benchmark, and managed and publisher-direct CTV programs (premium streaming inventory bought direct or via agency) pooled roughly 1.7x. CTV prospecting beat CTV retargeting by roughly 1.9x, and CTV as a whole out-performed linear TV by roughly 75% on measured iROAS. It is the largest-headroom channel on this list, because the reach ceiling is effectively national.
Worth noting on the buying side: publisher-direct premium inventory is now reachable without a DSP seat or an agency. Comcast’s Universal Ads puts 20-plus premium publishers, and since April 2026 Comcast-owned linear inventory, into a single self-serve workflow with no upfront commitment. Given that publisher-direct programs were the strongest sub-tactic in our CTV data, that is a materially cheaper way to run a first real CTV test than it was a year ago.
Pros
- One of the most broadly adopted channels in the panel, and proven incremental at meaningful scale.
- Prospecting-first, so it builds the top of funnel that Search then harvests.
- Cleanly geo-testable.
Cons
- Needs a real budget commitment, because sub-scale CTV tests in our panel often failed to register measurable lift.
- Platform-reported CTV attribution is heavily inflated, so measure independently.
Overview
Reddit was the panel’s consistency champion: conversion- and consideration-objective Reddit programs delivered solid positive iROAS at every brand in the panel that ran them, a claim no other platform here can match. High-intent community context appears to translate into genuinely incremental response, and both prospecting and retargeting lines performed.
Pros
- Positive measured performance at every brand in the panel that ran conversion objectives.
- Under-priced attention, with CPMs still low relative to engagement quality.
- Retargeting and prospecting both indexed above benchmark.
Cons
- Awareness-objective buys under-performed, a pattern across all social platforms in our data.
- Creative must feel native, because polished brand ads get punished.
- The scale ceiling is real for niche categories.
Sources
Overview
Direct mail is the most under-tested channel on this list, and the data says that is a mistake, with a sharp targeting caveat. Audience-targeted programmatic direct mail (PebblePost, PostPilot) pooled 1.6x benchmark, and the standout program, optimized CRM and lapsed-customer mailings, was one of the single highest-iROAS line items in the entire panel, at 9x-plus benchmark. Cold lookalike prospecting mail, by contrast, pooled below benchmark, and a large traditional catalog program landed at roughly 0.7x.
Pros
- CRM-triggered and retargeting mailings were top-decile performers.
- Zero auction pressure, so costs do not inflate as digital CPMs rise.
- Cleanly testable with holdout groups by design.
Cons
- Cold prospecting via lookalike mail under-performed consistently.
- Longer creative and production cycles than digital.
- Requires clean first-party data.
Sources
Overview
Pinterest is the reliability play. It is the most widely adopted expansion platform in the panel, it pooled just above benchmark on roughly $20M of measured spend, and, unusually, prospecting worked. Conversion-objective Pinterest prospecting indexed above benchmark at most brands, with the best programs (catalog, DPA, and Performance+ formats) reaching 2x to 3.5x benchmark. It rarely produced the panel’s top line item, but it almost never produced a disaster either.
Pros
- Broadest brand adoption and most predictable performance of any platform here.
- Conversion-objective prospecting genuinely incremental, not just retargeting.
- Performance+ and DPA formats measurably lifted efficiency.
Cons
- Traffic- and awareness-objective buys pooled near zero, so objective discipline matters.
- Ceiling on daily spend before efficiency decays in smaller categories.
Sources
Overview
Creator commissions bought through platforms like ShopMy, where creator links are tracked, commissioned, and optimizable, pooled 1.3x benchmark, with the best program near 2x. Contrast that with a large unstructured influencer program at another brand, which indexed at just 0.7x. The lesson is not “influencers work.” It is that platform-structured, commerce-tracked creator spend works.
Pros
- Best program in the panel ran roughly 1.8x benchmark.
- Pay-on-performance commission structures limit downside.
- Compounds with organic and social presence.
Cons
- Small panel sample, so treat the index as directional.
- Highly dependent on creator and brand fit, with results varying 3x across brands.
Sources
Overview
Affiliate pooled just under benchmark, but the average hides the story: content and creator affiliates ran up to 2x benchmark at multiple brands, while coupon, browser-extension, and cashback partners ranged from genuinely incremental to pure margin leakage depending on the brand. Affiliate is less a channel than a portfolio. The panel data says the difference between a 0.5x and a 2x program is partner curation, not the network you choose.
Pros
- Content and publisher partnerships were reliably above benchmark.
- CPA pricing caps downside while you learn.
- Every major network (Impact, Rakuten, Awin, Partnerize, CJ) had above-benchmark programs in the panel, so the network is not the variable.
Cons
- Coupon and extension partners often harvest existing demand, so incrementality-test them separately.
- Program-level averages are meaningless. Measure at the partner-type level.
Overview
TikTok is where the most expansion money went, and pooled, it under-performed the panel. But no channel had wider variance: conversion-objective prospecting with strong native creative ran 2x to 3x benchmark at several brands, Smart+ and GMV Max formats out-performed standard setups at multiple brands, while big-budget consideration and awareness lines at other brands indexed near zero. TikTok rewards operators and punishes tourists.
Pros
- Top-quartile programs were among the best social results in the panel.
- Smart+ and GMV Max automation measurably improved efficiency where adopted.
- Massive scale headroom, where the constraint is creative rather than audience.
Cons
- Pooled performance below benchmark: roughly half the panel’s TikTok spend earned materially less than the same dollars would have elsewhere.
- Awareness and consideration objectives were the panel’s biggest money pits.
- Requires always-on native creative production.
Sources
Overview
The audio picture is mixed. Pooled podcast spend indexed below benchmark, dragged by a large under-performing program at one brand, yet another brand’s podcast program measured well above benchmark, and a broadcast radio program elsewhere in the panel landed at 0.8x on meaningful spend. Audio can work, but in our panel it was the least predictable channel. Test it with a proper geo design and a realistic flight length before committing.
Pros
- The best podcast program in the panel indexed above benchmark, so host-read endorsement fit matters.
- Radio delivered near-benchmark efficiency at scale in one program, so do not dismiss broadcast.
- Low creative cost relative to video channels.
Cons
- Widest performance uncertainty of any channel in this list.
- Attribution vendors’ pixel and promo-code measurement materially over- and under-counts, so geo testing is essential.
Also Worth a Look
- Amazon Ads (advertising.amazon.com): indexed far below benchmark in our panel, but on a limited sample where the Amazon retail halo is not captured in DTC conversion data. If retail is material to you, measure it with a purpose-built design before judging.
- Traditional catalog: roughly 0.7x at seven-figure scale in a single program. It works as retention media and is expensive as prospecting.
- Native and display (Taboola, Yahoo Native, Zeta): scattered results, with retargeting lines occasionally strong and prospecting mostly below benchmark.
- Email and SMS (Klaviyo, Attentive): excluded from this analysis as owned channels, but panel iROAS on CRM sends was consistently high. Max these out before buying new media.
The Right Way To Enter a New Channel
The pattern across every entry above is identical: platforms do not succeed or fail, programs do. Conversion objectives beat awareness objectives, CRM-informed targeting beats cold lookalikes, and native creative beats repurposed assets. And every platform’s own attribution flattered itself. So enter new channels the way our best-performing panel brands do: size a real test budget, launch with a geo holdout design, read incrementality rather than platform ROAS at four to eight weeks, then scale winners into your media plan with saturation curves, not vibes. ChatGPT Ads is the newest example. Here’s how we’d approach it. Book a demo to see how Measured brands find, test, and validate new channels.
Frequently Asked Questions
How should I choose which channel to test first after Google and Meta? Match the channel to the asset you already have. If you have a young audience and saturated Meta prospecting, Snapchat pooled at 1.9x the expansion-channel benchmark in Measured’s panel. If you have a product catalog and meaningful site traffic, commerce retargeting DSPs pooled at similar levels, with Criteo at 2.3x. If you have a clean CRM file, programmatic direct mail pooled at 1.6x. If you need scale headroom rather than efficiency, CTV pooled at 1.5x to 1.7x with an effectively national reach ceiling. In every case, size the first test to detect lift and read it with a geo holdout rather than platform-reported ROAS. Is TikTok worth testing in 2026? Yes, with discipline. TikTok pooled below the expansion-channel benchmark in Measured’s panel, but top-quartile programs, meaning conversion-objective prospecting with native creative, often using Smart+ or GMV Max, ran 2x to 3x benchmark. The variance is the message: enter with a geo test and clear kill criteria. Does direct mail still work? Programmatic direct mail targeted at CRM audiences such as lapsed customers and cart abandoners pooled 1.6x the panel benchmark, and the best CRM-triggered program was one of the highest-iROAS line items Measured observed anywhere. Cold lookalike prospecting mail under-performed. Target warm, and measure with holdouts. How much budget does a fair channel test need? Enough spend, for long enough, to detect lift against your baseline volatility. In Measured’s panel, sub-scale tests, especially in CTV and audio, frequently failed to register measurable incremental sales. Size the test with a power analysis, not a round number. Why do these numbers differ from platform-reported ROAS? Platform attribution over-credits, especially on retargeting and branded-keyword adjacent buys. Every index in this article is based on incrementality-calibrated measurement, meaning geo experiments plus calibrated modeling, which is why some famously “high-ROAS” channels index lower here than their dashboards suggest. Methodology: Indexes reflect pooled incremental ROAS (incremental sales divided by media spend) by platform, from incrementality-calibrated measurement in Measured’s Cross-Channel Dashboard across a panel of enterprise consumer brands and 269 tactic-level programs totaling more than $135M in expansion-channel spend over the trailing 52 weeks ending July 2026, expressed relative to the pooled expansion-channel benchmark (1.0x = panel average). Google Search, Shopping, and PMax, Meta, and owned email and SMS are excluded from the benchmark. CTV figures reference the companion $25M-plus CTV panel. Platform groups with thin panel coverage are flagged as directional. No client-identifiable data is disclosed. Channel order does not represent a ranking.