Splitting Budget Between Google and Meta
A worked, margin-based framework for deciding how to split Shopify ad budget between Google PMax/Shopping and Meta Advantage+.
Written by Mantas Jurgutis — Founder, Adsify — builds the Google & Meta automation merchants use daily
Editorially reviewed by Adsify Editorial on March 15, 2026 — Reviewed against Shopify, Google Ads and Meta official documentation.
Two platforms, two demand types
Google Shopping and Performance Max campaigns primarily capture existing demand — people actively searching or browsing with purchase intent. Meta Advantage+ campaigns primarily create and capture demand through interruption-based discovery in feeds and stories. This structural difference means the two platforms typically show different CVR and CPC profiles for the same product, and a fixed budget split (like the common but arbitrary '60/40 Google/Meta') ignores how those profiles interact with your specific margin.
Starting split with no data
For a Shopify store with no historical platform performance data, Google's and Meta's own account-setup guidance both point toward giving new campaigns a stable, unchanging budget for an initial learning window — commonly cited as roughly 1-2 weeks for Google Smart Bidding and until Meta's ad set exits 'learning' status, generally around 50 optimization events. A pragmatic starting split is 50/50 by budget for 4-6 weeks, sized so each platform's campaigns can individually reach roughly 15-30 conversions per week — enough to generate a usable read.
Worked example: initial split sizing
Target CPA estimate of $35 on both platforms initially. To get 20 conversions/week on each platform: weekly budget per platform ≈ 20 × $35 = $700, or $2,800/platform/month, $5,600 total monthly starting budget. This is a starting hypothesis, not a permanent allocation — it exists purely to generate enough data within 4-6 weeks to make an informed reallocation decision.
Reallocating based on blended CPA, not ROAS alone
After the initial test window, compare each platform's actual CPA (not just ROAS) against your break-even CPA. Worked example: break-even CPA of $40 (from a 54% contribution margin on a $74 AOV). Google delivered $32 CPA across 85 conversions; Meta delivered $46 CPA across 61 conversions. Google is comfortably under break-even with a solid sample size; Meta is above break-even. The next month's reallocation might shift the split to roughly 65/35 Google/Meta, while Meta's remaining budget funds a creative refresh test rather than being cut entirely.
Don't judge Meta on last-click alone
Meta's attribution and Google's attribution both undercount cross-platform assist effects — a customer who sees a Meta ad, doesn't click, then searches the brand name on Google and converts via a Google Shopping ad gets full credit assigned to Google in last-click models. Meta's own reporting documentation notes this is a known limitation of single-platform, single-touch attribution. Before drastically cutting Meta budget based on weak last-click CPA, check for a corresponding lift in branded search volume or direct traffic during the same period, which suggests Meta is contributing upper-funnel value not captured by its own reported CPA.
Category differences in platform performance
Category matters more than most generic advice acknowledges. Visually distinctive, impulse-friendly products (apparel, home décor, beauty) tend to perform relatively better on Meta's discovery-based placements; considered or search-driven purchases (electronics accessories, replacement parts, specific branded goods) tend to perform relatively better on Google's intent-based Shopping and Search inventory. These are general tendencies from how each platform's ad formats work, not guarantees — the only reliable way to confirm which applies to your specific catalog is the 4-6 week split test described above.
Worked example: category-driven reallocation
A home décor store selling visually strong products tests the 50/50 split and finds Meta CVR of 2.8% versus Google CVR of 1.6%, with similar CPC on both platforms (~$0.95 Meta, $1.10 Google). CPA on Meta = $0.95 ÷ 0.028 = $33.93; CPA on Google = $1.10 ÷ 0.016 = $68.75. Against a break-even CPA of $45, Meta is well under break-even and Google is over. This store should shift meaningfully toward Meta — perhaps 70/30 — while still maintaining a smaller Google Shopping presence to capture branded and high-intent search that Meta cannot replicate.
Accounting for full-funnel role, not just direct CPA
Google Search and Shopping campaigns targeting branded terms typically show artificially strong CPA because they're capturing demand generated elsewhere, including by Meta's upper-funnel activity. When splitting budget, separate branded Google spend (which should be judged more as a defensive, low-CPA capture mechanism) from non-branded Google Shopping/PMax spend (which competes more directly with Meta for new-customer acquisition), and do the platform-split comparison only on the non-branded portion for an apples-to-apples read.
Seasonal and inventory-driven adjustments
Some categories see meaningful seasonal shifts in relative platform performance — search intent on Google often rises sharply during defined shopping seasons (holiday gifting searches, back-to-school), while Meta's discovery-based demand generation is comparatively more stable year-round. If your break-even math shows both platforms are profitable during a seasonal peak, temporarily shifting the split toward Google to capture the intent spike (e.g., moving from 50/50 to 65/35 for a 4-6 week peak window) can capture incremental profitable volume that a static split would miss.
The risk of over-optimizing the split too frequently
Reallocating budget between platforms weekly based on short-term CPA swings undermines both platforms' learning phases — Google's Smart Bidding and Meta's delivery optimization both need a stable budget signal to build reliable conversion predictions, per each platform's own bidding documentation. A practical cadence is reviewing the split every 2-4 weeks with a minimum of 20-30 conversions per platform in the review window, rather than reacting to day-to-day or even week-to-week variance.
A three-tier budget structure
A workable structure many Shopify stores land on after several reallocation cycles: a 'core' tier (60-70% of budget) on whichever platform shows the lowest, most stable CPA relative to break-even; a 'growth' tier (20-30%) on the second platform, sized to keep it above the minimum viable test threshold so it keeps generating usable data; and a 'test' tier (5-10%) for new campaign types or audience experiments on either platform, evaluated on its own separate timeline rather than against the core tier's CPA target.
Where automation helps with rebalancing
Manually re-pulling CPA and POAS by platform every few weeks and rebalancing budgets is straightforward in principle but easy to neglect under day-to-day operational pressure. Adsify's optimizer runs every 6 hours across both Google and Meta campaigns launched from the same Shopify catalog and applies auto budget scaling rules based on performance thresholds, which handles the incremental, small adjustments day to day while larger strategic split decisions — like the tier structure above — remain a monthly or quarterly human call.