Meta Ads vs Marketplace Ads: How to Assign Channel Roles

Meta and a marketplace are both places to spend acquisition budget, and both return a revenue number. But they do different jobs, report on different ledgers, and their numbers are not directly comparable. The allocation question is not “which channel wins” — it is “what role does each channel play, and what does the marginal dollar buy in each one.” This guide answers that with contribution, customer ownership, and incrementality, pinning down the exact numerator and denominator behind every metric so two channels are never compared on mismatched math.

For the surrounding account decisions, compare Consolidate or Split? When to Pool Meta Campaigns for Liquidity and use Catalog Ads Not Delivering: A Feed-Health Checklist as the next diagnostic.

In short

The decision is a role assignment, not a ranking: give Meta a discovery role and a marketplace an intent-capture role, then judge each on marginal contribution after its own channel-specific costs — not on the revenue each platform reports for itself. The single metric that can adjudicate the split is marginal contribution per incremental dollar, and the only clean read on incrementality comes from a controlled test, not from platform-reported revenue.

Channel jobs (a role description, not a causal claim)

Describing a channel’s job is useful for planning. It is not the same as proving that channel caused a sale — that requires a test, covered below.

• Meta (discovery). People encounter products in feed without having searched for them. You are paying to create demand and to reach an audience, and Meta reports revenue it attributes to that exposure. That reported revenue is a steering signal, not settled proof that the purchase would not have happened anyway.
• Search (intent already expressed). Someone typed a query. The ad meets existing intent rather than creating it. Useful to name here because search-attributed revenue overlaps Meta-attributed revenue on the same owned-store orders — the reconciliation below shows the double count.
• Marketplace (intent capture inside the marketplace). A sponsored listing appears against competing products at the moment a marketplace shopper is deciding. The sale, the payment, and the reported revenue all live inside the marketplace ledger, separate from your owned store.

These are complementary roles for different parts of the buyer’s path. Treating them as substitutes and asking only “which is cheaper” is the error this guide is built to avoid, because “cheaper” depends entirely on which denominator you divide by.

Measurement dictionary (numerator ÷ denominator for every metric)

Every metric below states its own math. If a later section cites a number, this is where its definition lives.

• Meta paid ROAS = Meta-attributed revenue ÷ Meta ad spend. A platform steering measure. It is not proof of incrementality, and it is not comparable to a marketplace’s reported ROAS because the two are computed on different ledgers.
• Marketplace ads ROAS = marketplace-ads-attributed revenue ÷ marketplace ad spend. Attributed inside the marketplace, on marketplace sales. Marketplace referral fees and fulfilment charges are costs of the sale — they are not ad spend and never belong in this denominator.
• MER (marketing efficiency ratio) = total business revenue ÷ total paid-media spend. A whole-business ratio. Do not call this “blended ROAS”: its denominator is paid media only, not any single channel. MER measures overall paid-media dependence, not the efficiency of Meta or of the marketplace.
• Owned-store CAC = owned-store acquisition spend ÷ owned-store net-new customers. Align the populations: the numerator is only the spend that acquires owned-store customers (Meta + search here), and the denominator is only customers you can identify as first-purchasers in your own store. Never divide a numerator that includes marketplace spend by an owned-store-only customer count and call the result “blended CAC” — the two are different populations. Marketplace identity is limited, so report the marketplace’s paid-media spend allocated per total ledger order separately, not folded into a store CAC (see Customer identity limits).
• Contribution per order = AOV − (cost of goods + shipping + returns + channel fees + allocated paid-media spend). This is the number that decides allocation, because it is what each order actually leaves behind after that channel’s own costs. Compute both channels on this identical basis — gross contribution − channel fees − fulfilment − paid-media spend allocated per total ledger order — so the comparison is like-for-like.
• Marginal contribution per incremental dollar = incremental contribution generated ÷ incremental ad spend, measured from a controlled spend change. This is the deciding metric for the split: step a channel’s spend up or down and read the extra contribution the change produced against the extra spend it cost. It is a marginal read from a test, not an average of the account, and not a platform-reported figure.

Reconciliation table (one worked scenario)

The figures below are an assumption, not a benchmark — a single internally consistent scenario so you can see how the ledgers relate. Recompute against your own accounts. Two separate revenue ledgers are shown: your owned store, and the marketplace.

Illustrative scenario — an assumption, not a benchmark or an expected result.

Line	Owned store	Marketplace
Orders	1,800	700
AOV	$75	$60
Actual revenue (= orders × AOV)	$135,000	$42,000
Meta ad spend	$24,000	—
Search ad spend	$8,000	—
Marketplace ad spend	—	$6,000
Meta-attributed revenue (platform-reported)	$90,000	—
Search-attributed revenue (platform-reported)	$60,000	—
Marketplace-ads-attributed revenue (inside marketplace)	—	$21,000

Actual business revenue = $135,000 + $42,000 = $177,000. That is the only revenue the business banked.

Now watch the channel-reported totals against actual owned-store revenue. Meta-attributed ($90,000) + search-attributed ($60,000) = $150,000 of channel-reported revenue against $135,000 of actual owned-store revenue. The $15,000 gap (~11% of owned-store revenue) is evidence that the two channel reports cannot be summed — it exceeds the revenue the store actually banked. This scenario assumes that excess is attribution overlap (both platforms claiming credit for some of the same owned-store orders); it is not a measured count of exactly $15,000 of duplicated revenue, and pinning the true overlap needs order-level deduplication or a test. Either way, it is not additional business revenue, and you cannot sum channel-reported revenues and treat the total as sales.

From the table, the metrics derive directly:

• Meta paid ROAS = $90,000 ÷ $24,000 = 3.75× (Meta-attributed revenue ÷ Meta spend).
• Marketplace ads ROAS = $21,000 ÷ $6,000 = 3.5× (attributed inside the marketplace ledger).
• MER = $177,000 ÷ ($24,000 + $8,000 + $6,000) = $177,000 ÷ $38,000 ≈ 4.7× (total revenue ÷ total paid-media spend — not “blended ROAS”).
• Owned-store CAC = owned-store acquisition spend ÷ owned-store net-new customers = ($24,000 Meta + $8,000 search) ÷ 1,280 identifiable first-purchasers = $32,000 ÷ 1,280 = $25 (aligned populations — the numerator excludes the $6,000 marketplace spend, and the denominator counts only owned-store customers you can identify; 1,280 first-purchasers ≤ 1,800 owned-store orders, the rest being repeat orders). The $6,000 marketplace spend is not in this numerator and is reported separately below.
• Marketplace paid-media spend allocated per total ledger order = $6,000 marketplace ad spend ÷ 700 marketplace orders = $8.57 per marketplace order. This spreads marketplace ad spend across all marketplace ledger orders — it is not CAC per new customer: marketplace identity is limited, so the net-new customer count is unknown and cannot be deduplicated against the owned store — do not blend it into the owned-store CAC.

The 3.75× and 3.5× are on different ledgers, so a higher marketplace ROAS does not make the marketplace “more efficient” as a business — that comparison needs contribution, next.

Owned-store vs marketplace economics (contribution after channel-specific costs)

Compare what one order leaves behind on the same basis for both channels — gross contribution − channel fees − fulfilment − paid-media spend allocated per total ledger order — at 60% gross margin in this scenario. Here paid-media spend allocated per total ledger order = channel paid-media spend ÷ total channel-ledger orders (owned-store: $32,000 Meta + search over all 1,800 owned-store ledger orders; marketplace: $6,000 over all 700 marketplace ledger orders — these denominators are every order on each ledger, not the ad-attributed subset):

• Owned-store order (paid-media spend allocated across all ledger orders): $75 AOV × 60% = $45 gross contribution; no channel referral fee, no marketplace fulfilment; paid-media spend allocated per total ledger order = $32,000 ÷ 1,800 = $17.78. Average ledger contribution after allocating paid-media spend across all orders = $45 − $0 − $0 − $17.78 = $27.22 per order.
• Marketplace order (paid-media spend allocated across all ledger orders): $60 AOV × 60% = $36 gross contribution, minus a 15% referral fee ($9) and $4 fulfilment, minus paid-media spend allocated per total ledger order = $6,000 ÷ 700 = $8.57. Average ledger contribution after allocating paid-media spend across all orders = $36 − $9 − $4 − $8.57 = $14.43 per order. The referral fee and fulfilment are sale costs, not ad spend, so they stay out of the ROAS denominator and sit here in contribution.

On the same basis — average ledger contribution after allocating paid-media spend across all orders — the owned-store order leaves $27.22 and the marketplace order $14.43 in this scenario — but that gap is a scenario assumption, not a benchmark, and it is not yet an incrementality claim (that needs the controlled test below). Note also that the owned store may retain more of the first-party customer identity and relationship — email, on-site behaviour, the ability to remarket and to sell again — whereas the marketplace may limit the first-party customer identity and relationship available to the advertiser; record the actual access your marketplace provides rather than assuming either extreme.

Two honest caveats. First, the 15% referral figure is a scenario assumption; real marketplace commissions vary by marketplace and by product category, so use your own contract rate rather than any single number as universal. Second, the two paid-media-allocated-per-order figures above ($17.78 and $8.57) spread each channel’s paid-media spend across all of that ledger’s orders — they are not acquisition cost per new customer. The owned-store CAC per new customer is higher ($25, over 1,280 first-purchasers) because it divides by first-purchasers only, not all 1,800 orders — keep the per-total-ledger-order allocation basis when comparing the two channels’ after-allocation contribution, and the per-new-customer basis when judging acquisition efficiency.

Customer identity limits (do not assume marketplace buyers are new)

A recurring allocation mistake is asserting that a marketplace buyer is a new, incremental customer. There is no basis to assume that.

• On a marketplace, you may not receive the buyer’s email or a durable first-party identifier, so you cannot deduplicate a marketplace buyer against your owned-store customers. “Net-new” is unknown, not “yes.”
• Some marketplace buyers might have discovered the product through your Meta activity and completed the purchase inside the marketplace for its checkout, returns, or payment options. If so, the marketplace ledger books revenue that your Meta spend helped cause — but the reverse is equally possible, and neither is established without a test.
• Because you cannot see identity across the two ledgers, compute CAC only on the population you can identify — owned-store first purchases divided by owned-store acquisition spend (the aligned owned-store CAC above), plus any genuinely deduplicated marketplace identities you truly have. A figure is only “blended” if its denominator actually includes deduplicated marketplace identities; otherwise report the marketplace’s paid-media spend allocated per total ledger order separately. State that boundary rather than silently counting every marketplace order as a new customer.

The disciplined position: a marketplace order is real revenue on the marketplace ledger, of unknown incrementality to your Meta program, until a controlled test says otherwise.

Budget decision (marginal contribution and predeclared stop rules)

Allocate on the margin, not on a fixed split. For the next dollar, ask which channel returns more contribution after its own channel-specific costs, given your constraints — and predeclare the rules before you look at a good week.

• Marginal, not average. The next $1,000 into Meta or into marketplace ads is judged on the contribution it adds, using each channel’s own cost stack (fees and fulfilment for the marketplace; acquisition cost for the owned store). Average ROAS across the account hides the marginal picture.
• Constraints are real. Inventory, working capital, and operational capacity cap how fast either channel can absorb spend without eroding contribution.
• Predeclared stop rules. Write the thresholds in advance: a minimum contribution per incremental order, a spend-pacing limit that keeps learning intact, and a rule to pause a channel whose marginal contribution falls below break-even. No universal allocation percentage is prescribed — the split is whatever the marginal-contribution reads and your constraints produce.

Incrementality test (the only clean read)

Platform-reported revenue answers “what did the platform take credit for,” not “what would not have happened without this spend.” To read incrementality, run a controlled change and accept its limits.

• Holdout or geo test. Hold a defined audience or a set of regions out of a channel, or step spend up or down in matched regions, and measure the difference in total business revenue and orders across the two ledgers — not the platform’s own attributed number.
• Measure at the business level. Because Meta, search, and the marketplace report on overlapping or separate ledgers, the trustworthy readout is the change in banked revenue and order count, reconciled to your order records.
• State the limitations plainly. Holdouts need enough scale and duration to separate signal from noise; seasonality and external events can confound a single test; and one result is specific to the audience, period, and creative you ran. Treat a test as evidence to update on, and re-run it when conditions change.

Can software help?

Bach.ai audits your connected Meta account against 100+ checks, ranks what it finds by estimated impact, and proposes specific fixes. It stays read-only until you approve a change, then executes the approved change on Meta; connected Google Ads data is used for intelligence only. Think of it as an automated audit layer that surfaces issues and proposed fixes for your review — not a replacement for your team’s judgment, and it does not generate your creative.

Common mistakes

• Summing channel-reported revenues into “total sales.” Meta-attributed plus search-attributed revenue exceed actual owned-store revenue in this scenario, and marketplace-attributed revenue belongs to the separate marketplace ledger — none of these reports may be summed into business revenue. The owned-store excess shows the reports cannot be summed; overlap is a scenario assumption, not a measured duplicate amount.
• Calling MER “blended ROAS.” MER’s denominator is total paid media, not one channel — the label matters because it changes what the number means.
• Assuming marketplace buyers are incremental. Without identity resolution or a test, incrementality is unknown, not assumed.
• Comparing a marketplace’s ROAS to Meta’s ROAS as if they settle allocation. Different ledgers, different cost stacks — decide on contribution, confirmed by a test.
• Putting marketplace fees into ad-spend denominators. Referral and fulfilment charges are costs of the sale, so they belong in contribution, not in ROAS.

FAQ

Is Meta ROAS comparable to marketplace ROAS?

Not directly. Meta paid ROAS = Meta-attributed revenue ÷ Meta ad spend, computed on your owned-store ledger; marketplace ads ROAS = marketplace-ads-attributed revenue ÷ marketplace ad spend, computed inside the marketplace. They use different numerators, different ledgers, and different cost stacks. Convert each to contribution after that channel’s own costs per order, then compare — and confirm with a controlled test before you call either channel incremental.

What is the difference between MER and blended ROAS?

MER (marketing efficiency ratio) = total business revenue ÷ total paid-media spend. There is no separate “blended ROAS” here: if the denominator is paid media only, the correct name is MER. Calling it “blended ROAS” implies a single-channel efficiency read it does not provide — MER measures overall paid-media dependence across every channel at once.

Are customers from a marketplace new customers?

Treat that as unknown. Marketplaces limit the buyer identity you receive, so you may not be able to deduplicate a marketplace buyer against your owned-store customers, and you cannot see whether your Meta activity influenced the purchase. Count marketplace orders as revenue on the marketplace ledger and compute CAC on the population you can actually identify (aligned owned-store CAC = owned-store acquisition spend ÷ owned-store first-purchasers), rather than assuming every marketplace order is a new customer. Only call a figure “blended CAC” if its denominator genuinely includes deduplicated marketplace identities.

How do I know if my Meta spend is actually incremental?

Run a holdout or geo test: withhold a defined audience or matched regions from Meta, or step spend up and down in matched regions, and measure the change in total business revenue and orders across both ledgers, reconciled to your order records — not the platform’s attributed number. State the limits (scale, duration, seasonality, confounds), and re-run when conditions change.

Should marketplace fees count in my ROAS calculation?

No. Referral fees and fulfilment charges are costs of the sale, not advertising cost. Keep them out of any ROAS denominator (which is ad spend only) and account for them in contribution per order instead. Folding fees into ad spend would misstate both the ROAS and the comparison between channels.