The Localization Stack: what actually has to be rebuilt when you cross a border

Most "go global" decks price the wrong thing. They size the market and copy the product, then treat everything underneath — how goods move, how strangers trust each other, what the regulator allows, how money changes hands — as detail. It is not detail. It is the localization stack, and it is what actually breaks on a border. Here is the framework I use to decide whether a market is a launch or a decade-long infrastructure bet.

The framework: one portable layer on four local ones

Split every advantage you have into a portable product layer and the local substrate it rests on. The product — your app, ledger, brand, UX — ships in an afternoon. The substrate does not. It has four layers, and each has to be re-founded, not exported:

1. Distribution. The physical or human network that gets your thing to a person — agent density, last-mile logistics, addressing, retail shelves. You can't ship 40,000 trusted cash-in points or a parcel network; you build them.
2. Trust. What signals safety here — a telco's credibility, cash-on-delivery, a local number, a human on WhatsApp. Trust is market-specific and non-transferable.
3. Regulation. The license, the sandbox, the timing window that lets you operate. A gap that's open in one market is already shut in the next.
4. Money movement. The rails that let value flow — card penetration, real-time payments, remittance corridors. Where they're thin, you carry the cost the rail would.

The product is portable. The four substrate layers are local. That asymmetry is the whole game.

Worked examples

• M-Pesa — substrate that stayed home. Its moat was agent density, a telco trust anchor, and a one-time regulatory window — all local. Export the app without them and adoption collapsed; it was shut in South Africa in 2016. Substrate you can't ship.
• Jumia — product imported ahead of substrate. The "Amazon of Africa" copied the storefront into markets where logistics, digital payments, and remote-payment trust barely existed — then spent years, and a ~77% share-price fall, building the roads under the store.
• Nubank — substrate rebuilt, market by market. Rather than assume Brazilian trust and credit data would travel, Nubank re-founded the stack in each country. By late 2024 it had passed 100 million customers across Brazil, Mexico and Colombia — roughly 10 million in Mexico and about 2.5 million in Colombia — and became Latin America's most valuable bank. Slow, local, and it worked.
• UPI — a substrate rail that is actually traveling. India didn't export an app; it exported a payment rail, linking UPI to local fast-payment systems in about nine countries (Singapore, the UAE, France and more), with cross-border transactions growing roughly 20-fold in a year. When the thing you carry across the border is substrate — interoperable, not a walled app — it can travel.

The operator take

You don't "expand" a locally-won business; you re-found it, one substrate layer at a time. The number that decides a market isn't its size — it's how much of the four-layer stack you'll have to pour before the product has anything to stand on.

Pre-flight checklist

Before you enter a market, for each of the four layers:

• Is this layer portable (ships with the product) or local substrate (must be re-founded)?
• For every substrate layer, write the rebuild plan — who, how long, how much. No plan, no entry.
• Price the market by substrate-rebuild cost, not TAM. A high-inclusion market can be a worse bet for a substitution product, not a better one.
• Decide the honest verb: are you expanding (product travels) or re-founding (substrate must be built)? Budget accordingly.