Every beauty founder I talk to has the same instinct about channel cost. Meta is the cheap one. Google is where you go when you can afford it.
The first half of that is true. The second half is backwards.
We pulled the trailing 30 days across the 25 beauty and DTC accounts we manage, roughly $1,990,883 of media, and measured every account on the same basis. This is the channel level companion to the account level work we have published on what CPM tells you about rising CAC and on weighted CAC. Here is what came out.
The short version
$32.40 median CPM on Google Ads across 20 beauty accounts. Meta came in at $22.90 across 14. Google carries a 41% premium to reach the same person.
$24.29 median cost per acquisition on Google. Meta came in at $53.93. Google acquires 55% cheaper.
$122.46 median average order value on Meta. On Google, $118.65. A gap of 3.2%.
Every figure is a median across accounts, not a blended average. One large account cannot drag them.
The reach premium is real
Google impressions cost more in beauty, and it is not marginal. $32.40 against $22.90 per thousand is a 41% difference, and it holds across verticals. Google skincare runs $31.30 CPM. Google haircare runs $42.17, the most expensive inventory we buy anywhere. Our Meta advertising cost benchmarks for beauty brands give you the range your own account should be judged inside.
The spread is wide on both platforms. The middle 50% of Google beauty accounts sit between $20.67 and $46.21 CPM. On Meta the same band runs $15.69 to $34.95. So there are Google accounts buying cheaper than the Meta median, and Meta accounts paying more than the Google median. Channel sets the centre of gravity, not your outcome.
If CPM were the whole story, the advice would write itself. It is not the whole story.
The acquisition cost runs the other way
Here is the flip. Median cost per acquisition:
| Platform | CPA | Middle 50% of accounts |
|---|---|---|
| $24.29 | $16.66 to $54.33 | |
| Meta | $53.93 | $30.78 to $80.30 |
Google is 55% cheaper per customer. Not a rounding difference. More than twice as efficient at the point that actually hits your P&L.
Return on ad spend says the same thing. Google beauty medians at 4.44. Meta at 2.10. Over 90 days the gap widens: Google 5.48, Meta 1.77. If you are setting a number to steer against, start with how to set a target ROAS for beauty brands rather than with the platform default, and our ROAS calculator does the arithmetic.
The mechanism is not mysterious. On Google you are buying a person who has already decided they want something and typed it. On Meta you are buying attention and then manufacturing the want. Both are legitimate. One of them is a shorter trip to the checkout.
Look at the traffic metrics and you can see it happen. Google beauty clickthrough is 1.53% against Meta at 1.86%. Fewer people click on Google. But 5.19% of them convert, against 4.24% on Meta, and the Google conversions cost less than half as much to produce.
Fewer, better, cheaper. That is what intent buys you.
The caveat that matters
Every platform counts its own conversions, and every platform is generous to itself. Google is reporting Google conversions. Meta is reporting Meta conversions, and Meta credits purchases to view through impressions that Google would never claim. We have written about this gap at length in the attribution gap and the three CAC numbers in beauty.
So the direction of this finding is solid. The precise multiple is soft. If you ran both platforms through a single source of truth, some of the gap would close.
It would not close all the way. The pattern is consistent across 25 accounts, two windows and every vertical we measure, and it points the same direction every time. Treat 55% as the optimistic end of a real effect, not as a number to put in a board deck.
What channel does not change
Median average order value came in at $122.46 on Meta and $118.65 on Google. Across 25 accounts and about $1M of media in a month, the channel moves basket size by 3.2%.
This is the finding I would push hardest on, because it contradicts a lot of what gets sold. Channel mix is not an AOV strategy. If you want bigger orders, the levers are bundling, size upgrades, subscription and merchandising. They sit on your site and in your product roadmap, not in an ad account.
The corollary is useful too. Because AOV holds roughly constant across channels, the CPA difference passes almost cleanly through to contribution margin. A customer acquired on Google at $24.29 against one acquired on Meta at $53.93 spends about the same on the first order. The $29.64 difference is margin. That is also why we argue for judging the engine on MER and new customer ROAS rather than on platform reported return alone.
Where we sit against the published averages
Our numbers are from accounts we run, so it is fair to ask whether they are typical. Against the external benchmarks published for 2026, mostly yes, with two honest exceptions.
Our Google CPC of $1.26 sits close to the widely published ecommerce search average of about $1.16. Our Google CPA of $24.29 comes in under the published ecommerce median of $28.14. Our Google ROAS of 4.44 lands inside the 400% to 600% band commonly cited for ecommerce. Nothing exotic in the Google set.
Two places we differ, and why
Our Google conversion rate is high. We report 5.19% against a published ecommerce range of 2% to 3%. Some of that is genuine, and some is definitional. Conversion actions in these accounts are not all purchases, and accounts counting a wider set of actions will read higher. Read our Google conversion rate as directionally strong and not strictly comparable to a purchase only benchmark.
Our Meta CPA is high. We report $53.93 against published DTC beauty figures in the mid to high $30s. The likely reason is basket size. Published beauty data shows conversion rate falling hard as AOV rises, from about 2.78% for brands under $50 AOV to about 0.83% for brands over $200. Our Meta median AOV is $122.46, well into the premium band. Premium beauty costs more to acquire. That is the trade, not a failure.
Those external figures are published industry aggregates, not ours. We include them for context and they carry all the usual caveats about mixed methodology.
The vertical picture
Haircare is the standout. On Google it converts at 8.16% against the 5.19% beauty median, acquires at $17.07 against $24.29, and does it on a $141.63 basket against $118.65. It also carries the highest CPM of anything we buy at $42.17. Expensive to reach, cheap to convert, bigger order. We broke the category economics down in full in what it costs to grow a haircare brand in 2026.
The reason is structural. Haircare runs on replacement cycles, and people arrive searching for a specific shade, length, formula or refill. That is a fundamentally different search from someone browsing serum. The demand already exists and the search engine is just routing it.
Be careful with the top of this range, though. The middle 50% of haircare ROAS runs from 5.95 to 67.69, and that ceiling is one hair extensions account with an unusual conversion profile. The median of 9.00 holds up without it, but anyone quoting the top of that band is quoting an outlier.
Skincare and body is the volume. Thirteen Google accounts, eight on Meta, and it behaves close to the beauty median: $31.30 CPM, 4.35% conversion, $41.49 CPA, 3.58 ROAS on Google. If you sell skincare, the overall beauty numbers in this post are roughly your numbers, and our 2026 skincare advertising benchmarks break the category down by channel.
Makeup we are not publishing. We run one colour cosmetics account at meaningful spend, and one account is not a benchmark. Rather than dress up a single data point as an industry figure, we suppress any cut with fewer than five accounts. It stays suppressed until the sample supports it.
The 90 day view
Thirty day windows move around. Over 90 days and about $2.7M of media, the pattern holds and in places sharpens.
| Metric | Google 30d | Google 90d | Meta 30d | Meta 90d |
|---|---|---|---|---|
| CPM | $32.40 | $28.74 | $22.90 | $19.99 |
| CPC | $1.26 | $1.27 | $1.51 | $1.15 |
| CTR | 1.53% | 1.76% | 1.86% | 2.06% |
| CVR | 5.19% | 5.45% | 4.24% | 3.23% |
| CPA | $24.29 | $24.01 | $53.93 | $51.26 |
| ROAS | 4.44 | 5.48 | 2.10 | 1.77 |
| AOV | $118.65 | $123.17 | $122.46 | $111.29 |
The CPA gap is stable. The ROAS gap is wider over the longer window. AOV stays within a narrow band on both platforms. Nothing here suggests the 30 day read was a fluke.
What to do with this
Four moves
- If you are underweight on Google, that is probably the gap. Not because Google is cheap, it is not, but because the customers cost less at the end of the funnel. Test a budget shift and measure on contribution margin rather than platform reported ROAS.
- Do not read this as a case for abandoning Meta. Meta is where demand gets created for a brand nobody is searching for yet. The accounts running both are using Meta to build the category interest that Google then harvests. Kill Meta and you slowly starve your own branded search.
- Stop expecting channel to fix AOV. Three percent is the whole effect. Put that energy into the product page.
- Benchmark against your own vertical, not against beauty. The distance between haircare at $17.07 CPA and skincare at $41.49 is larger than the distance between platforms. A blended beauty number is a starting point, not a target.
Method
Figures come from paid media accounts Pennock manages directly, pulled from the Google Ads API and from account level Meta aggregates. Windows are the trailing 30 and 90 days ending September 14, 2026. All currency is USD.
Every published figure is a median across accounts. No single account value is shown, no client is named, and any cut with fewer than five accounts is suppressed rather than estimated. Ranges are the middle 50% of accounts.
Accounts whose conversion tracking does not report real ecommerce revenue are excluded from ROAS and average order value while remaining in cost and traffic metrics. One account with unreliable conversion tracking is excluded from all conversion metrics. Conversion rate is purchases per click on both platforms, but Meta credits view through purchases against a click denominator, so Meta conversion rate reads high relative to Google. External benchmark figures referenced in this post are published industry aggregates from third parties, not Pennock data.