Google Ads or Meta Ads is the wrong question. The two platforms don't do the same job — Meta creates demand, Google captures it. Comparing their ROAS in the same spreadsheet is comparing a door-to-door salesman to a cash register. Here's how to split the budget based on what your category actually allows.
Two platforms, two different jobs
Meta is an interruption machine. Nobody opens Instagram to shop for your products: the platform inserts your offer into a feed where no one was expecting you. Performance therefore depends almost entirely on creative — the algorithm finds the audience on its own.
Google is a capture machine. Someone typing "garage bike rack" or your brand name has already decided they want something. Search, Shopping and brand campaigns harvest demand that already exists. Google creates almost nothing: it intercepts.
The practical consequence is simple. Meta can grow your market — it's the only one of the two that puts your product in front of someone who wasn't looking for it. Google is capped at your category's search volume. You can double the Search budget: if 8,000 people a month search for your type of product in your market, you'll never capture 16,000.
That's why "Google or Meta" is the wrong question. The real question: how much demand already exists for my category, and how much do I need to create?
“Meta creates demand, Google captures it: judge each channel on its job, not on the same ROAS.”
The four inputs that decide the split
Before talking percentages, four variables settle the question.
- Your category's search volume. Open Keyword Planner and add up monthly non-brand searches for your core terms. Below a few thousand searches a month in your market, Google can never carry your growth — no matter how well the account is built.
- Average order value. A $2.50 CPC at a 2% conversion rate is a $125 acquisition cost. On a $60 AOV, non-brand Search will never work. On a $400 AOV, it's a bargain.
- Creative capacity. Meta burns creative like a furnace. What we see on the accounts we manage: without 8 to 12 new ads a month, creative fatigue erodes performance within 4 to 6 weeks. If you can't produce that volume, Meta will plateau fast.
- Gross margin. It sets the maximum CAC you can tolerate on cold acquisition. Meta prospecting costs more on the first purchase than brand Search — if your margin can't absorb acquiring a customer at a loss on order one, your split has to lean toward capture.
Nobody can give you a split without these four numbers. Anyone who does is selling you a recipe, not a strategy.
Judging both on the same ROAS distorts everything
The classic spreadsheet: Google shows 9x, Meta shows 2.8x. The manager's conclusion: cut Meta, double Google. It's almost always the wrong read.
Google's ROAS is inflated by brand. Your brand campaigns intercept people who already knew you — existing customers, word-of-mouth traffic, people your Meta ads warmed up. A good share of those conversions would have happened anyway, organically, for free.
A 10x ROAS on your own brand name isn't performance. It's your awareness ringing the cash register.
Meta's ROAS has its own distortions: view-through attribution credits sales to impressions that were never clicked, which inflates in-platform numbers. But the reverse effect exists too — Meta creates buyers who then convert... through your Google brand campaign. The channel doing the work doesn't get the credit.
The right reflex is incrementality thinking. Two simple tests: pause your brand campaign for two weeks and watch whether organic and direct conversions absorb the volume. Then compare Meta's in-platform ROAS with your blended MER (total revenue divided by total ad spend) over 60 days. If MER rises when Meta spend rises, Meta is working — whatever the attribution table says.
Starting splits by budget tier
These splits are starting points — what we see across the e-commerce accounts we manage — not dogma. Adjust them with your four inputs.
$5k/month: one channel only. Split it in half and neither channel accumulates enough signal to optimize. If your category has search volume and an AOV that supports the CPCs, go with Google (brand + Shopping + 2-3 tight non-brand ad groups). Otherwise, go with Meta. Win that channel before opening the second one.
$15k/month: cap capture, push creation. Fund Search fully first — brand, Shopping, the non-brand queries that convert. In most categories, that caps out between $3k and $5k/month. Everything else goes to Meta. The split often lands around 70/30 toward demand creation, because capture saturates first.
$50k+/month: full-funnel on both sides, plus a test. Meta prospecting with structured remarketing, Google across Search + Shopping + YouTube. Keep 5 to 10% to test a third channel. At this level, the challenge isn't the split anymore: it's scaling Meta without crushing efficiency — we documented our method in how to scale Meta past $10k/day.
The logic shared by all three tiers: capture gets funded first because it's finite. Creation gets every growth dollar because it's expandable.
The Performance Max trap
Performance Max is Google's best-disguised conflict of interest. The campaign blends Search, Shopping, Display and YouTube into a black box — and by default, it bids on your own brand name.
What we see in the accounts we audit: a PMax running at 6-8x ROAS where a significant share of conversions come from brand queries. The campaign is re-buying customers who were already typing your name. The report looks gorgeous. The money is being spent intercepting your own traffic.
Three fixes:
- Exclude your brand from PMax with brand exclusion lists — they're native in the interface.
- Run a separate brand Search campaign with its own capped budget. You want to see its real cost, isolated.
- Open PMax's search terms report every month. If brand queries dominate, your "growth engine" is a toll booth on your own awareness.
The same reflex applies to the whole account: every campaign has to prove it generates sales that wouldn't happen without it. If you're not sure what your account captures versus what it creates, that's exactly the kind of thing we break down in an audit.
When the split flips
Everything above leans toward Meta because most e-commerce brands sell discretionary products. But in some categories, the split flips completely.
High-intent categories — replacement parts, specialized consumables, products that solve an urgent problem — live on Google at 70-80%. Nobody discovers a heat pump filter on Instagram: you search for one when yours dies. Here, Meta plays a supporting role (remarketing, repeat purchase), not the engine.
At the other end, impulse and discovery categories — fashion, gifts, gadgets, new product formats — have almost no search volume to capture. Nobody searches for a product they don't know exists. These brands live on Meta at 80%+, and their Google boils down to brand and Shopping harvesting the demand Meta created.
The quick test: at the moment of purchase, was your customer searching for a solution, or did an ad just create the want? The answer dictates your primary channel.
On the accounts we manage, that read — not a magic percentage — is what sustains our 4.2x average ROAS. Both channels orchestrated as one system is exactly what we run in our e-commerce Funnel Growth program.