What is e-commerce optimization?

Read Shopee's own fee schedule and the numbers are concrete: non-Mall commission runs 1–6% by category and Shopee Mall 3–12%, with seller-funded vouchers and a Free Shipping Program cost-share stacked on top before a single baht of ad spend (Shopee Help Center). TikTok Shop's published affiliate schedule reaches roughly 20% before the platform's own commission applies (TikTok Seller University). None of these costs appears in the one number most sellers optimise against — units sold. That gap is the whole subject of this post.

More sales is not the goal. More true profit is. On SEA marketplaces those two numbers stopped moving together.

E-commerce optimization is the discipline of using data and mathematical optimization to make the next best decision for true profit — not for GMV. The claim this post defends: on Shopee, Lazada, and TikTok Shop in 2026, the decisions that maximise sales and the decisions that maximise profit diverge often enough that optimising for sales is a structurally biased proxy. What would falsify it is simple — if rising sales reliably produced rising profit, you would not need optimization at all; you could just chase the top line. The cost stack above is why they do not.

What "optimization" actually means

Optimization is not a marketing word. It comes from operations research — the field that uses data, mathematical models, and explicit constraints to find the best decision available, not merely a defensible one. INFORMS, the discipline's professional body, defines operations research as the application of advanced analytical methods to help make better decisions. Manufacturers use it to plan production at the lowest cost; logistics teams use it to route fleets; airlines use it to price seats. In every case the shape is the same: take the numbers, respect the constraints, and compute the decision that does best against an explicit objective.

Point that machinery at an online store on Shopee, Lazada, or TikTok Shop and you get e-commerce optimization. The objective is true profit. The constraints are budget, inventory, fee schedules, and competitor pricing. The decisions are which campaign to fund, which to stop, which SKU to push, and where to set price. The difference between this and a dashboard is the difference between a report of what happened and a computed answer to what to do next.

Why more sales does not mean more profit

Take a representative Shopee account scaling its ad budget into a campaign window.

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 +60%. Net profit -33%.
Marginal revenue +THB 300,000  ->  marginal net profit -THB 20,000.

Every incremental unit in the scale-up carries a higher ad cost — the seller bid the auction up with its own money — a heavier seller-funded voucher as the campaign tier rises, and a lower realised price as more orders come from discounted tiers. The dashboard reports growth; the bank account reports erosion. This is ad waste wearing a marketing-success label, and the platform's reported ROAS cannot see it because the numerator never subtracts marketplace fees, vouchers, or COGS. Bain's e-Conomy SEA 2025 documents retail-media inflation across SEA marketplaces — the marginal sale costs more to acquire every year — while Sea Limited's 4Q25 disclosure shows Shopee's commission-and-advertising take-rate climbing: the platform monetises seller activity, not seller profit.

The questions optimization actually answers

A store is not one decision; it is dozens, made weekly, under more variables than any operator can hold in their head — price, COGS, discounts, fees, ad spend, stock, competitors, and customer behaviour. Optimization combines those inputs into a single computed recommendation. Concretely, it answers:

• Which campaign should get more budget, and which should be stopped — ranked by true ROAS per SKU, not platform ROAS.
• Which promotions lift contribution margin, and which quietly eat it.
• Which SKU is worth pushing, and which loses more the harder you push it.
• Which price, on which product, clears the most profit at the demand you can actually expect.

None of these is a backward-looking report. Each is a forward decision with a number attached — the operating definition of optimization, and the reason a decision engine is a different category of tool from analytics. A dashboard tells you sales went up; optimization tells you to cut the third campaign and reprice two SKUs because that is what clears the most profit next week.

A dashboard tells you what happened. Optimization computes what to do next.

Operationally, this is what DataGlass computes for Shopee, Lazada, and TikTok Shop sellers: not another view of yesterday's numbers, but a ranked list of the next decisions that move true profit, with the arithmetic shown. How that looks under the hood is a separate post — see how e-commerce decision engines work.

Where this argument breaks

• Below ~THB 200,000 monthly revenue, the operational overhead of per-SKU optimization usually exceeds the recoverable margin — simpler category-level heuristics win.
• Optimization is only as good as the cost data feeding it. Without clean per-order reconstruction of fees, vouchers, and COGS, the "true profit" objective degrades to a guess.
• Sales-maximisation is not always wrong. For a launch buying market share, or for clearing aged stock, units can be the right objective for a defined window. Optimization means choosing the objective deliberately — not defaulting to sales.