20 percent more fulfilment space, added a quarter before the volume arrived, is the shortest version of MercadoLibre's third quarter. The company booked the cost early, cut shipping cost per unit in Brazil by 5% year on year on an FX-neutral basis, and still grew revenue 41% FX-neutral. Revenue rose 41% FX-neutral with USD 676 million of operating income, and management gave up 40 basis points of margin to fund logistics and affiliate marketing. As of November 12, 2026, the interesting question for e-commerce operators is not whether the bet worked, but which part of it is repeatable.
What the quarter added
| Metric | Reported | Why it matters |
|---|---|---|
| Revenue growth | 41% FX-neutral | Demand outran capacity additions |
| Operating income | USD 676 million | Absorbed the upfront logistics cost |
| Fulfilment space | Up almost 20% quarter on quarter | Capacity placed before peak, not after |
| Shipping cost per unit, Brazil | Down 5% year on year | Density gains offset the new cost |
The sequence matters. Adding space before peak means paying rent, labour and systems costs in a quarter that does not yet carry the volume. Most operators do the reverse: they wait for the volume, lose delivery promises in the peak weeks, and then buy capacity at a worse price. The trade-off is real, though, because a 40 basis point margin concession is only defensible if the space is used within a couple of quarters.
Why affiliate orders matter more than they look
The affiliate network passed 12 million members and orders through that channel grew 240% year on year. For a marketplace, this is a discovery channel that does not depend on search ranking or paid placement. It also changes the cost structure: commission paid to creators replaces part of the marketing budget, and the marginal cost of an extra order can be lower than a bid on a keyword.
The catch is measurement. Orders that arrive through a creator are harder to attribute across devices, and returns tend to be higher when the purchase decision is made from a video rather than a product page. Consumer demand shaped by creators also moves faster, which puts pressure on stock accuracy: a product that goes viral on a Monday can be out of stock by Thursday, and the review that follows is public.
Shipping cost per unit is the real test
Density, not distance, drives the curve
A 5% decline in unit shipping cost while capacity grows is a density result. More parcels per route and per picker spread the fixed cost of a delivery. If the decline had come from distance, it would have reversed as soon as fuel or tolls moved. Density gains persist as long as order concentration in the served postcodes holds.
The peak season is the stress test
Capacity added in the third quarter is tested in November and December, when basket mix shifts and delivery promises tighten. The relevant question for the next earnings call is not square metres but drops per route and on-time delivery in the busiest weeks. A marketplace that grows space without growing density ends up with lower cost per square metre and higher cost per order.
What to watch next quarter
Three numbers will show whether the bet is compounding: fulfilment utilisation, shipping cost per unit in Brazil, and the share of orders that arrive through the affiliate channel. If utilisation rises while unit shipping cost keeps falling, the capacity was placed correctly. If utilisation rises and the cost line flattens, the network is carrying space it cannot fill profitably.
FAQ
Is adding space before peak always right?
Only if the space is used within a couple of quarters. Paying for capacity that sits empty through the peak turns a logistics bet into a fixed cost problem.
Why does shipping cost per unit fall while capacity grows?
Because density improves faster than the new fixed cost arrives. More parcels per route spread the cost of each drop, which is why unit cost can fall even as total spending rises.
What is the risk in affiliate-driven demand?
Stock accuracy and returns. A product promoted by creators can sell out within days, and purchases made from video tend to come back more often than those made from a product page.
