Ads
Take a Shopee account moving its monthly ad budget from THB 40,000 to THB 100,000. In the accounts we model, revenue on that move commonly grows about 60% while net profit falls by roughly a third. Nothing in the sequence looks like a mistake while it is happening: the dashboard reports growth, the campaigns hit their ROAS targets, and the operator scales again next month.
The reason is in the definition. Shopee's own ads documentation defines ROAS as attributed gross revenue divided by ad spend [4]. That is mathematically clean and operationally misleading — it excludes the four largest variable costs of selling on Shopee: commission and transaction fees, seller-funded vouchers, shipping subsidies, and the cost of goods. A metric that omits most of where your margin goes cannot tell you when growth stopped paying.
Optimising toward platform ROAS is, on a scaling campaign, mathematically biased toward overspend.
Thesis: scaling Shopee ad budget is not a neutral volume decision — every input that platform ROAS ignores moves in the same direction as scale and against the seller's margin, so the metric's error grows precisely when the stakes do. What would falsify this is a scaling account where auction cost per click, voucher tier, and discount depth all held flat as budget rose; we do not observe that. This is the ad-side deep-dive of the increase profit on Shopee framework — the supply-side half is how to cut Shopee seller costs.
Why scaling Shopee ad budget breaks the platform metric
Shopee's Sale Transaction Fee varies by category — on the Thailand schedule effective 4 August 2026 (4 ส.ค. 2569) it runs 7.49–17.12% for non-Mall sellers and up to 19.26% for Shopee Mall, VAT-inclusive as published, with a 3.21% payment transaction fee and a THB 1 per-order infrastructure fee on top [5]. Free-shipping subsidies and voucher mechanics layer additional seller-funded costs on top, and several of those mechanics are toggled on by default during major campaign windows like Pay Day, 9.9, and 11.11 [5].
Aggregate those costs and the gap between platform-reported revenue and seller-realised revenue is structural — on the fee schedule effective 4 August 2026 it is typically 25–33 percentage points on a mainstream non-Mall category carrying a voucher and free shipping, before COGS or ads even enter the calculation, and about 19 points even on the 7.49% electronics floor [5]. A campaign reporting 4.0 platform ROAS on a category with 50% COGS lands at a true ROAS of about 0.9 once those fees and vouchers are deducted — below break-even, not barely above it, and reliably indistinguishable from a "winning" campaign in the platform dashboard. The same arithmetic on the superseded 1 June 2026 schedule produced 1.0.
The scaling trap, in baht
Baseline: THB 500,000 revenue · 35% gross margin · THB 40,000 ads · THB 60,000 net profit
Scaled up: THB 800,000 revenue · 25% gross margin · THB 100,000 ads · THB 40,000 net profit
Revenue grew 60%. Net profit fell 33%.
Marginal revenue: +THB 300,000 → marginal net profit: −THB 20,000.Each additional unit sold during the scale-up carries a higher ad cost (the auction is bid up by the seller's own additional spend) [3], a heavier voucher subsidy (Shopee's voucher tiers escalate with campaign size), and a marginally lower price (more orders at the discounted tier). The economics on the increment are negative even though the dashboard shows growth.
The asymmetry matters operationally. The dashboard reports the THB 800,000 month as a 60% revenue gain — a number worth celebrating in a board update. The operator who looks only at GMV scales the ad budget further the next month, the auction bids up further, and the marginal economics deteriorate again. Within two or three quarters the account has doubled in revenue and halved in net profit, with no single decision visibly wrong. The compounding is structural: every input the platform-reported metric ignores moves in the same direction as scale, against the seller's margin.
True ROAS, operationalised
The formula is straightforward. The hard part is the data plumbing — reconstructing fees, vouchers, COGS, and fulfillment per order across an account with hundreds of SKUs and dozens of campaigns. This is the work DataGlass automates. Once that data is unified, every campaign on a Shopee account can be ranked against its own break-even bar in production.
True ROAS = (attributed revenue − COGS − marketplace fees − vouchers − shipping subsidy − discounts) / ad spend
Break-even ROAS = 1 / (1 − total variable cost share)
Example: SKU with 50% COGS, 8% fees, 5% voucher, 4% shipping
break-even ROAS = 1 / (1 − 0.67) = 3.0
A 4.0 platform ROAS campaign on this SKU is fine.
A 3.5 platform ROAS campaign on this SKU is a loss.Three moves that make scaling Shopee ad budget safe
1. Cut the bottom-quartile keywords by true ROAS, not platform ROAS
Across the Shopee accounts we model, 20–30% of ad spend on a typical account runs below break-even on true ROAS terms. Pausing it lifts profit before any new sale. The cleanest implementation: rank every keyword by true ROAS, pause the bottom quartile, redistribute its budget proportionally to the top decile.
2. Refuse the campaigns whose voucher tier exceeds your category margin
Shopee's major campaigns escalate seller-funded voucher tiers — 5%, 10%, 15%, sometimes higher. A category running at 22% gross margin cannot absorb a 15% seller-funded voucher and still cover commission, fees, and ads. Decline the higher voucher tiers in those categories. Participation is not free; it is bought with margin you may not have.
3. Set a break-even ROAS per SKU and lock it into the campaign brief
A single account-wide ROAS target is, by construction, wrong for most of the catalog. Every SKU has a different break-even ROAS — derived from its own COGS, fees, and voucher exposure. The brief for every campaign should name the target SKUs, their break-even ROAS, and the rule for pausing.
A 4.0 platform ROAS feels profitable. A 1.3 true ROAS is barely break-even. The dashboard cannot distinguish between them.
Sensitivity — when the framework recovers more vs. less
The framework's margin recovery scales with the gap between platform ROAS and true ROAS. The table below stress-tests how much margin the per-SKU break-even rule recovers under different account profiles.
| Account profile | Typical platform-ROAS gap | First-quarter margin recovery |
|---|---|---|
| Mall-heavy account, high voucher participation | Platform 5.0 → True 1.4 | 5–8 percentage points |
| Non-Mall, moderate voucher tier | Platform 5.0 → True 2.0 | 4–6 percentage points |
| Mall-heavy, declines deep voucher tiers | Platform 5.0 → True 2.4 | 3–5 percentage points |
| Single-platform Shopee specialist (no Lazada/TikTok) | Platform 5.0 → True 1.8 | 4–6 percentage points |
| Cross-border seller (different fee stack) | Variable | Re-derive against negotiated rates |
| Top-decile enterprise (negotiated commission) | Platform 5.0 → True 2.6 | 2–3 percentage points |
The structural conclusion: accounts with higher Mall participation and deeper voucher exposure see larger margin recovery from the framework, because the platform-vs-true gap is larger to begin with. Top-decile sellers with negotiated rates have less to recover; below the size bound, simpler heuristics outperform the operational overhead.
Limitations and where this argument breaks
- Account-size lower bound. The per-SKU break-even ROAS framework requires order-line data clean enough to reconstruct fees, vouchers, and COGS per order. Below ~THB 200K monthly revenue, the operational overhead exceeds recovered margin; simpler heuristics (per-category margin assumption, fixed account-wide ROAS target with conservative voucher caps) outperform.
- COGS data quality. The per-SKU break-even ROAS is only as good as the COGS input. Long-tail SKUs without uploaded COGS default to category-mean inference; a 5-percentage-point COGS error shifts a SKU's break-even ROAS by ~0.5–1.0 — material on borderline-margin SKUs. Refresh COGS quarterly.
- Negotiated commission. Top-decile Shopee accounts operate with commission rates below the public-rate-card values used in the worked example. The framework still applies; the lookup table needs the negotiated terms.
- Cross-border sellers. Shopee International Platform sellers face a different commission and customs cost stack not modelled here.
- Time horizon. Sea Limited's documented AI investment in the ad auction is one-way pressure on cost-per-click [1][6]. The framework's margin recovery is durable but the input values (auction CPC, voucher tiers) drift upward over time. Re-run the audit quarterly.
- Internal-data scope. The 4–7 percentage-point margin recovery figure is aggregated across the SEA-6 Thai Shopee accounts we model directly. Not a population claim about all Shopee sellers; explicitly excludes accounts below the size bound.
Methodology
Public-data citations are taken from the Shopee Ads Help Center (ROAS definition, Target ROAS bidding mechanics, Keyword Ads documentation) [3][4], the Shopee general Help Center (commission, transaction fee, Shop Voucher mechanics) [5], Sea Limited's 4Q25 / 1Q26 investor disclosures [1], and the Bain e-Conomy SEA 2025 commentary on retail-media inflation in SEA marketplaces [6].
Internal-data claims — the 4–7 percentage-point margin recovery figure, the 20–30% loss-making ad-spend share, the typical platform-vs-true ROAS spread (4.0 → 1.3 on representative profiles), the 200%+ volume-lift requirement that 12% discounts impose on 18% contribution-margin SKUs, and the cohort distribution underneath the audit recommendations — are aggregated across the Thai SEA-6 Shopee accounts we model directly. The Shopee subset comprises approximately 280 active accounts across the DataGlass research methodology sample frame (Jan 2024 – Apr 2026, 28-month observation window). Worked examples in the post are illustrative composites, not the financials of any specific seller.