A café owner in Manila asked us to sanity-check a quote: $28,000 for a menu-and-order mini-program. The same week, a freelancer offered to do it for $600. Both were real. Neither was lying. They were pricing different things, and the gap between them is the most useful thing to understand before you spend a dollar.

The honest answer to "what does a mini-program cost" is that the build is rarely the expensive part.

What you're actually paying for

Most quotes blur four different jobs into one number. Pull them apart and the range stops looking insane.

Line itemCheap versionExpensive versionWhat drives the gap
Screens & flow5–8 template screensCustom motion, many statesHow much you deviate from standard blocks
BackendReuses host servicesOwn CRM, ERP, inventoryWhether you already have systems to connect
DesignTemplateBrand system, IllustrationWhether a brand guide exists
Compliance & reviewStandard categoryPayments, health, finance docsRegulated categories need real paperwork

The $600 quote is the first column. The $28,000 quote is mostly the second and third columns plus the fourth. Both build a working mini-program. Only one of them also rebuilds the business behind it.

The part nobody quotes: the host already paid

Here's what makes mini-programs structurally cheaper than an app, and why a fair quote should reflect it. The host has already built and is maintaining identity, payments, messaging, distribution, and the runtime your code runs on. You're not buying those. A quote that charges you to "build login" or "integrate a payment gateway" from scratch is either padding or a misunderstanding — those are inherited, not constructed.

That's also the quiet risk. If a vendor prices as if they're building a standalone app, they'll build one inside a mini-program shell, and you'll pay app money for mini-program value.

Where budgets actually get eaten

In our experience the overrun never comes from the screens. It comes from three places, in this order.

First, integrations. The moment the mini-program has to talk to a legacy inventory system or a homegrown CRM, the estimate doubles and the timeline doubles with it. Second, content operations — someone has to keep prices, stock, and hours current, and that's an ongoing cost no build quote includes. Third, review cycles in regulated categories, where a rejected submission means another round of docs and waiting.

If you want a number that holds, scope the integration explicitly and treat content ops as a monthly line, not a launch line.

The runtime angle

A host-owned runtime is a cost item that rarely shows up on a quote because it's free to you — and that's the point. One engine, patched centrally, consistent across devices. A standalone app estimate carries device-fragmentation QA, OS version matrices, and a signing/distribution pipeline. Those aren't small. When you compare quotes, make sure you're comparing a mini-program against an app that includes all of that, not against a bare prototype.

The tradeoff to price in

Cheap builds lock you to a template and a vendor. Custom builds give control but cost real money and take real time. Neither is wrong — it depends on whether the mini-program is a storefront or the product. For a café it's a storefront; pay the first-column price and move on. For a business where the mini-program is the customer relationship, the second column is an investment, not padding.

The one thing I'd push back on in any quote is a line for "building the platform." You're not building a platform. You're renting a very good one.

Sources