When should we move budget from Meta to Google, TikTok or another channel, or back?
Our answer
Move budget off Meta only when its marginal incremental return falls below what a tested channel delivers. Measure it with spend-level tests or holdouts, not averages.
- Confidence
- Medium
- Strongest proof
- 3 controlled tests
- 12 datasets, 10 practitioners, 5 opinions or claims
- Sources
- 17 sources
- 30 statements
DNRCP
New here? Take the 1-minute tour
Why this is the answer
3 reasons and our judgmentMeta's last dollars can return far less than its average. Return on extra spend usually falls before the average does, so PMG, a media agency, advises measuring it by comparing revenue at two spend levels.
Shift money on tested returns, not reported ROAS. In Eightx's client examples, a skincare brand paused connected TV at 1.4 times incremental ROAS and moved the money to YouTube, which returned 1.9 times (Eightx, a fractional CFO firm).
Last-click reports push budget the wrong way. In the 2026 Labor Day week, Google showed nearly 3 times Meta's ROAS, inflated by closing credit, while Meta's new-customer cost was under half of Google's (Wicked Reports, an attribution software maker).
Our judgment
- Size a channel test to read at your order volume, not to a fixed amount. We follow Measured's power example (about six weeks at 5,000 orders a week, past 20 weeks at 1,250) and Common Thread Collective's suggested $10K AppLovin test, half of it AppLovin ad credit.
- AppLovin has the most 2026 tests behind it as a next channel, then YouTube. We base this on Haus's AppLovin tests, Common Thread Collective's 18 AppLovin tests, HexClad's holdouts, Haus's YouTube Demand Gen tests and Common Thread Collective's Q1 2026 YouTube results.
- Test new channels one at a time, in a period of comparable demand. Cody Plofker (launch with one channel and let expansion be earned) and Adam Gillman (get one channel working first) back one at a time. In Haus's Ritual case, a slow-period TikTok test showed no lift and a higher-demand retest showed 8% more incremental subscriptions.
- Re-test Meta's headroom when your cost per new customer rises faster than margin allows. Common Thread Collective advises testing the highest-spend channels before raising Q4 budgets. Northbeam's all-channel rise is context.
- Keep a new channel, AppLovin included, only when its incremental return clears your break-even target after margin, not just when a test shows lift. PMG adds spend only while the next dollar pays back after margin, Eightx's body care brand skipped connected TV when breaking even needed 2.4 times, and Drew Fallon reports margins going away at brands that added about $200K a month on AppLovin.
Our judgment, built from Cody Plofker, Adam Gillman, Measured, Common Thread Collective, Haus, HexClad, Northbeam, PMG, Eightx and Drew Fallon: our reading of these sources, not a finding from any one of them.
What the sources say
30 statements from 17 sourcesFunnel's measurement guide uses a made-up example: an extra $500 returning $200 has a marginal ROAS of 0.4 while blended ROAS stays near 2.13, so average ROAS hides the return on extra spend.
“Marginal ROAS on that incremental $500 is 0.4, or $0.40 per dollar.”
Drew Sanocki says a holdout test of 2025 Black Friday direct mail across many brands showed 3.89x incremental ROAS. Brand count and method are not given, and he appears tied to a direct-mail business.
“we holdout-tested 2025 bfcm mail across a ton of brands. 3.89x incremental roas.”
Meta reports a lift test for a car dealership marketer: people who saw Meta ads viewed 7.67 times as many vehicle listing pages and generated 14% more paid search traffic. No spend, sample or dates.
“Shifting budget out of Meta and into search would make search a less effective channel.”
Marin Istvanic says one client added Pinterest, TikTok and Snapchat, each reporting 7 ROAS, but total revenue did not change. If MER falls with the new spend while revenue stays flat, the channel is just taking credit.
“if you add a new channel and MER drops proportionally to the new spend while total revenue stays flat, that channel is stealing attribution.”
Ruggable's CMO says it moved about 10% of marketing spend from performance channels like Meta into influencers and CTV, including over $1 million into ShopMy creators. No cost or return figures are given.
“what got us here is not going to get us there”
Drew Fallon, who sees about 200 brands' books at Iris Finance, keeps seeing brands that Meta took to $50 million add $200K a month on AppLovin and lose their margin. He has not seen it work.
“Meta gets a brand to $50 million, they add $200K a month on AppLovin, and the margin goes away.”
Pushes back on scaling AppLovin: Drew Fallon describes brands that reach $50 million on Meta, put $200K a month into AppLovin and see their margin go away. An observation from about 200 brands, with no data.
“Across the 200 brands on Iris, Drew has not seen one scale AppLovin spend and make more money.”
Haus's geo tests find YouTube Demand Gen returns about the same as a brand's other channels (0.98x), and a brand's first YouTube test now lands at 0.92x, up from 0.64x. Haus calls these comparisons directional.
“In the Demand Gen era, Demand Gen tests land right at parity with the brand's other channels (0.98x).”
PMG, an agency, says marginal ROAS, the change in revenue over the change in spend, usually falls before average ROAS does. Stop adding spend once the next dollar no longer pays back after margin.
“Stop increasing a channel when the next dollar no longer pays back.”
Ridge CEO Sean Frank says AppLovin was Ridge's best channel for scale and results last Q4, measured as incremental, and Ridge plans to run it again.
“Northbeam, haus, post purchase, applovin was incremental. We will do it again this year.”
Measured's panel of channels tested after Google and Meta indexed AppLovin at 2.0x the panel's average incremental return, among the strongest channels in the panel.
“AppLovin prospecting proved genuinely incremental at multiple brands, not just retargeting.”
Wicked Reports says in Labor Day week 2026, Google's new-customer cost rose 11% to $179 while Meta's fell to $82, and warns that last click undercredits Meta. Order-verified, not an incrementality test.
“If you rebalanced budget off that view this week, you rewarded the closer and cut the opener.”
HexClad's Connor Rolain says AppLovin reaches the most new customers for HexClad: holdouts plus Northbeam showed 90% new customers, 36% higher ROAS and 27% lower CAC. The post carries an AppLovin credit offer.
“Which channel will help me reach the most new customers?”
CTC's Luke Austin weighs a new channel's staff and measurement cost against the 2 to 5% of mix it will likely hold, and says cheaper geo testing now justifies $15 to 20K channel tests.
WorkMagic, which sells mix modeling, says David Protein found its Meta attribution inflated by 36%, and after reallocating on incrementality revenue grew 34% and profit 37%. No method or period is given.
“David Protein found their Meta attribution inflated by 36%”
Cody Plofker's rule for a new brand: launch on one channel (Shopify plus Meta) and let CAC earn expansion, adding Amazon around a $250 CAC and TikTok Shop around a $150 CAC.
“launch with one channel (Shopify plus Meta) and let expansion be earned by CAC”
Cody Plofker says Jones Road, at about 80% paid social, plateaued for a few months on reach; reach campaigns were a band-aid, and it needed a new place to put budget, which led it to TV.
“But we really needed like another place where we could reach people and put budgets and continue to scale”
Adam Gillman advises founders to put most of their effort into getting one marketing channel working before adding others.
“my recommendation is to really put the majority of your effort into getting one channel to work”
Northbeam says Meta CPMs rose 10 to 20% on average from January 2026, a yearly pattern, and median CAC across all channels rose 10.34% year over year, with brands above $50M in revenue holding up better.
“CPMs increased 10%-20% on average starting January 2026, a normal trend I’ve seen several years in a row now.”
Measured, which sells geo testing, says incrementality tests commonly run four to eight weeks and almost never under four, since ending early misses lagged sales. Its illustrative Meta example needs about six weeks.
“almost no test should run under four weeks regardless of what the power math allows”
CTC's 18 AppLovin incrementality tests averaged 171%, against a 113.4% average for its Meta acquisition campaigns.
“Out of 18 AppLovin incrementality tests, only one came back below 100%. Five returned 225%+ incrementality. The average was 171%.”
CTC's H1 2026 report puts Meta's average incrementality across its clients at 113.4%, from below 25% to above 200%. It blames heavy reliance on Meta and advises diversifying, not cutting Meta.
“If your Meta incrementality is low, the answer is rarely to cut Meta. It is to diversify”
CTC puts Meta's average acquisition incrementality across its clients at 113.4%, down from a 120% benchmark. It falls below 100% when Meta already drives 60 to 70% of new-customer revenue.
“the average incrementality factor for Meta is 113.4%, if you wanna get precise”
Haus says Ritual's first TikTok test showed no lift, but a retest in a higher demand period, with new settings and creative, drove 8% more incremental subscriptions. All three changed at once.
“Had Ritual tested only during slower periods, they would have concluded that the channel simply didn't work.”
Haus's incrementality tests found AppLovin's DTC efficiency 1.11x the average of other channels the same brands tested, though results move toward the median as brands scale.
“AppLovin tests landed in the top quartile of channel performance 39% of the time, compared to an expected baseline of 25%.”
Eightx, a CFO firm, says CTV pays for DTC beauty brands only once Meta and Google saturate and a measurement plan is set. Below $15M revenue it rarely works, and a 90-day test costs $90K to $200K.
“Decide on a measurement approach before the first dollar of media.”
Measured's study found 64% of Meta's incremental conversions came from new or reactivated customers, so the figure includes lapsed buyers. On new-customer goals Meta's iROAS was 2.3 times search's.
“64% of Meta technologies' incremental conversions were attributed to new or reactivated customers”
CTC's Connor McLaughlin says AppLovin results vary widely, so each brand should run its own test, suggesting $5K of spend plus $5K of AppLovin credit. CTC is an AppLovin partner and offers a $5K spend match.
Sean Frank says to push Meta further up the funnel once a holdout proves it incremental, before adding new channels; after Meta, Ridge went to YouTube, and says AppLovin does well in Q4.
“Go up funnel as soon as you can prove it is incremental with a holdout. Do that before new channels.”
CTC's Q1 2026 benchmark of 299 brands, using its incrementality factors, puts Google brand search at 19.1x platform ROAS but 5.7x incremental, and Meta acquisition at 1.83x reported versus 2.07x.
“Google Brand search reports 19x ROAS but delivers just 5.7x after incrementality adjustment (0.30x IF).”
In the full answer
For readers who sign in- What to do this week3 steps
- The numbers behind it3 figures from cited sources
- Where each source standsA chart placing each source on the question
- The strongest disagreementThe best case against the answer, from a cited source
- How sure we areWhat its confidence rests on, and what would change the answer