Not if the reason is CPM. The case circulating this summer says Google impressions cost $100-plus per thousand while Meta’s cost $10 to $18, so budget should follow the cheaper impressions. But a Search impression is priced against declared intent and a feed impression is priced against attention; they are not the same product, and pricing them against each other is a category error. You are not buying impressions in either auction, you are buying customers, and the only comparison that survives contact with a live account is incremental cost per new customer.
The claim: pull budget out of Google, pour it into Meta
Threadpoint published the cleanest version of the argument in late May. In a May 27 post, Frank Field argued that “the mid-funnel is getting squeezed” and that advertisers are responding logically: pulling budget out of Google, where CPMs run $100 and up, and pouring it into Meta, where the same thousand impressions cost $10 to $18 1.
Steelman it first, because the observation underneath is real. Meta impressions are cheaper by roughly an order of magnitude, and for pure prospecting reach, where the job is putting a brand in front of people who have never heard of it, the CPM comparison has merit. Nobody should pay Search prices for awareness.
The narrative also has macro cover. eMarketer forecasts Meta’s 2026 worldwide ad revenue at $243.46 billion, a 26.8% share, against Google’s $239.54 billion at 26.4%, which would make this the first year Meta out-earns Google, with Meta growing 24.1% to Google’s 11.9% 2. The money is visibly moving toward attention, so the question operators keep typing, should I move budget from Google to Meta in 2026, sounds like it has an obvious answer.
It does not, because the comparison driving it is broken.
A Search impression and a feed impression are different products
Comparing Search CPMs to social CPMs is a category error, not a bargain. A Search impression exists because someone declared what they want, in words, seconds ago. A feed impression exists because someone paused mid-scroll. The $100-plus CPM buys a thousand statements of intent; the $15 CPM buys a thousand moments of attention. It is the storefront on the busiest street in town priced against a billboard on the highway. Both can pay. They are not substitutes, and a price gap between non-substitutes is not an arbitrage.
The aggregate data shows no Google exodus either. In Q1 2026, Google search ad spend grew 14% year over year, the highest growth rate in nearly two years, with flat average CPCs, and brand keyword CPCs declined 9% 3. A channel being drained by smart money does not post its fastest spend growth in two years at stable unit costs. Advertisers in aggregate are holding their Search positions, and the auction is not punishing them for it.
The arbitrage closes as fast as the money arrives
Cheap Meta CPMs are a moving target, and the post recommending the trade documents this itself. Threadpoint’s own portfolio data shows its weighted average Meta CPM rising from $15.07 in 2025 to $17.01 in 2026, up 13% year over year across 17 accounts from January through May, with its largest account up 20%, from $15.35 to $18.44 1. The same post cites a Ryze benchmark putting average Meta CPMs up 20% year over year, from $11.82 to $14.19 1.
That is what a popular trade looks like while it self-closes. Reallocated budget lands in Meta’s auction, bids against the budget that arrived last quarter, and inflates the price gap that justified the move. Anyone reallocating today is buying 2026 CPMs with a 2024 thesis. Meta impressions will stay cheaper than Search impressions, because attention will always price below declared intent, but the spread shrinks every month the trade stays crowded.
Incrementality re-ranks both channels, in both directions
Dashboards mis-rank channels in both directions, which is why CPM reallocations and ROAS reallocations are equally blind. Common Thread Collective’s Q1 2026 Channel Mix Benchmark, covering 299 DTC brands and $231 million in paid spend, found Google Brand search reporting a 19.07x ROAS while delivering 5.72x incrementally, a 0.30x incrementality factor, while Meta acquisition campaigns reported 1.83x and delivered 2.07x, a 1.13x factor 4.
Read carelessly, that looks like a case for the trade: Google flatters itself, Meta sandbags itself. The useful lesson is narrower. The reported number is wrong in both directions, so any reallocation rule built on platform-reported figures, whether CPM, CPC, or ROAS, steers by a bent compass. And the same benchmark shows brands already routing 58.7% of paid dollars to Meta acquisition 4. Meta is not an undiscovered inefficiency. It is the consensus position.
The decision rule that survives this data is the one the firm runs on: measure what a channel adds that would not have happened anyway, and rank channels on incremental cost per new customer. Everything else is unit-cost optics.
When the move to Meta is the right call
Sometimes the answer is yes, move it. Three conditions, all measurable. Search demand is saturated: impression share is maxed on the queries that convert, and the queries beyond them do not. Marginal return on Search dollars is declining in an incrementality test, not in the dashboard. And Meta holds up under the same test at the spend level you intend to run, not the level you tested at.
The firm is pro-concentration, so none of this is channel-hedging advice. Concentration decisions run on incrementality math, not on which channel’s impressions look cheap. We have moved budget across this exact border ourselves, in the opposite direction: in the B2B SaaS account that killed LinkedIn and grew pipeline, budget came out of the channel with the impressive-sounding reach story and into Google, because the incremental numbers said so. The math ranked the channels; the unit costs would have ranked them wrong. That indifference to per-channel optics is easier to hold when compensation does not scale with spend, which is most of our argument with the percentage-of-spend model.
The disagreement is about the decision rule, not the data
Threadpoint’s numbers deserve to be taken at face value, and so does the forecast behind the broader narrative. Meta CPM inflation is real, the eMarketer projection tells a true story about where attention and ad dollars are heading, and none of that is in dispute here. The dispute is what an operator should do with it. “Impressions are cheaper over there” is a fact. “Therefore move the budget” is a decision rule, and it is the wrong one, because the impression was never the product.
If your incrementality math says Meta wins at the margin, move the budget and do not look back. If the CPM chart is the only thing saying it, you are about to pay a rising price for the wrong unit.