Shopify app development means writing software that plugs into a Shopify store through Shopify's APIs to do something the platform doesn't do out of the box. That's the whole definition. An app can sit in the admin as a new page, inject a widget into your storefront, run quietly in the background syncing orders to a warehouse, or add a discount rule at checkout that no theme setting can produce. Most store owners never need to build one. The ones who do usually need it badly, and they usually find out the hard way after stacking five apps from the App Store to imitate one workflow.
We build and publish apps on the Shopify App Store, and we build private ones for single merchants. What follows is the part we end up explaining on almost every discovery call.
Public, custom, and private apps are not the same thing
Shopify's naming has shifted over the years, so this trips people up.
Public apps are the ones you install from the App Store. Anyone can install them. They go through Shopify's review process, they have to meet Built for Shopify performance and security standards, and they're built to serve thousands of stores with the same feature set. Klaviyo, Judge.me, Recharge. These are the apps you should reach for first.
Custom apps (created in your admin under Settings → Apps → Develop apps) are tied to one store. No review process, no listing, no distribution. You generate an access token, scope its permissions, and go. These are perfect for a script that pushes orders into your 3PL's API, or a small internal tool your ops team uses.
Private apps is the older term Shopify deprecated in 2022. If a developer still calls it a private app, they mean a custom app. Not a red flag, just vocabulary.
There's also a fourth category worth knowing: app extensions. Theme app extensions, checkout UI extensions, admin actions, and Shopify Functions all run inside Shopify's own infrastructure rather than on your server. For anything touching checkout on Shopify Plus, Functions and checkout UI extensions are now the only supported route since checkout.liquid was retired. If someone proposes a checkout hack using Liquid in 2025, walk away.
When a custom Shopify app is actually the right call
Our honest default answer is: not yet. If an App Store app does 80% of what you need for $49/month, pay the $49. A custom build starts at a few thousand dollars and never stops needing maintenance. Shopify ships two API versions a year and deprecates old ones on a rolling schedule, so your app is a small, permanent line item.
Build custom when one of these is true:
- Your process is genuinely unusual. Made-to-measure furniture with a per-square-inch price, B2B customers on net-30 terms with tiered catalogs, rental with return dates. No off-the-shelf app models this cleanly.
- You're paying for four apps to fake one workflow. We've seen merchants running a bundler, a discount app, a subscription app, and a custom-fields app that fight each other at checkout. Consolidating into one purpose-built app often costs less over 24 months and removes three sources of JavaScript from the storefront.
- You need to talk to a system Shopify has never heard of. An ERP on a private network, a legacy inventory database, a regional courier with a SOAP API. Integration work is the single most common reason merchants come to us for Shopify app development services.
- Storefront performance is dying under app scripts. Each app that injects a script tag adds render-blocking weight. Replacing three widgets with one theme app extension has taken LCP from 3.1s to under 1.8s on stores we've worked on. That's a measurable revenue difference on mobile.
- You want to sell the app. Some merchants solve their own problem, then publish it. That's a different project with a different budget, because App Store review, billing API, GDPR webhooks, and multi-tenant architecture all come into scope.
What Shopify app developers actually work with
The modern stack is fairly settled. Node or Ruby on the backend, React on the front end, Polaris for admin UI so your app looks native, and the Shopify CLI to scaffold and deploy. Data comes through the Shopify API — GraphQL Admin API for anything new, since Shopify has been clear that REST is legacy.
Two constraints shape almost every architectural decision. First, rate limits: GraphQL runs on a calculated-cost leaky bucket, and a naive bulk operation will hit the ceiling fast. Bulk operations and cursor-based pagination exist for a reason. Second, webhooks are at-least-once delivery, not exactly-once, so your app has to be idempotent. Skip that and you'll get duplicate fulfillments during a Black Friday traffic spike, which is the worst possible time to learn it.
Common use cases we build most often
- ERP, WMS, and 3PL sync — orders out, inventory and tracking back in, running on a schedule with a retry queue.
- Product configurators for build-your-own kits, made-to-order, engraving, or per-dimension pricing.
- B2B portals: customer-specific price lists, quote requests, PO uploads, approval flows.
- Ops dashboards inside the Shopify admin so warehouse or CS staff never leave one tab.
- Checkout logic using Shopify Functions — validation rules, shipping method filtering, discount stacking that the standard discount engine can't express.
- Marketplace and channel sync for Amazon, Walmart, or region-specific platforms.
How the build actually runs
A typical project with us has five phases. Scoping: two to five days mapping the workflow, listing every API call needed, and confirming Shopify can even do the thing (sometimes it can't, and better to know in week one). Architecture: where it's hosted, how auth works, what happens when the external system is down. Build: two to eight weeks for most single-purpose apps, in weekly increments you can see. Testing on a development store with real order volume replayed through it, including the failure cases. Deploy and watch: the first two weeks in production tell you more than the whole QA cycle.
If you're publishing to the App Store, add Shopify's review queue on top. Budget one to four weeks, and expect at least one round of feedback on your listing or your OAuth flow.
What it costs
Rough US-dollar ranges from projects we quote regularly:
- A single-purpose custom app — one integration, one clear job: $3,000–$9,000, roughly 60–140 hours.
- A mid-complexity app with admin UI, background jobs, and error handling: $10,000–$25,000.
- Deep ERP or multi-warehouse integration with two-way sync: $25,000–$60,000+.
- A public App Store app with billing and multi-tenant architecture: start at $30,000 and expect ongoing product work forever.
Then maintenance: 10–20% of build cost a year, more if you depend on a third-party API that changes. Offshore rates change these numbers materially — our clients in the US and UK work with our Bengaluru team for the same senior skill at a fraction of a US agency's hourly rate, which is why the ranges above sit well below what a New York shop would quote. If you want the full breakdown, we've published our Shopify development cost guide.
Best practices that save you money later
Use metafields and metaobjects for structured data instead of inventing your own database wherever you can — the data stays in Shopify, visible and portable. Request the narrowest API scopes that work; over-scoped tokens are a liability, especially if a contractor leaves. Ship a theme app extension rather than script tags. Log every failed webhook somewhere a human will look. Write down the API version you're on and diary the deprecation date, because the email from Shopify will arrive at the exact moment you're busiest.
And keep the app small. The most maintainable apps we've built do one thing well and hand off everything else.
FAQ
Do I need Shopify Plus to build a custom app?
No. Custom apps work on every plan. Plus matters for checkout-level customization through Shopify Functions and checkout UI extensions, plus higher API rate limits and Shopify Flow. If your app doesn't touch checkout, a Basic plan store can run it fine.
Will a custom app slow down my store?
A well-built one won't. Apps that run server-side never touch your storefront. Storefront-facing features should ship as theme app extensions, which load through Shopify's CDN and are far kinder to Core Web Vitals than injected script tags.
How long does Shopify take to approve a public app?
Usually one to four weeks, depending on queue depth and how many revision rounds your submission needs. Rejections are common on first submission and are almost always fixable within a few days.
Who owns the code?
You should. Get it in the contract, and get the repository in your own GitHub organization from day one. Any Shopify experts worth hiring will agree without argument.
Can a custom app replace several paid apps?
Often, yes, and that's one of the cleanest business cases. Add up your monthly app bill, multiply by 24 months, and compare it against a build quote. If you're spending $400/month across overlapping tools, consolidation pays for itself.
What happens when Shopify updates its API?
Shopify releases quarterly API versions and supports each for at least 12 months. Your app needs a version bump roughly once a year. It's usually a few hours of work, unless you've used something that's been removed.
Should I hire a freelancer or an agency?
A freelancer is fine for a small, well-defined integration if you can handle the spec yourself. For anything your revenue depends on, you want continuity — someone who'll still be reachable in eighteen months when the API version expires. If you're weighing options, our notes on hiring a Shopify developer cover the trade-offs.
A sensible first step
Before you commission anything, get someone to look at your current app stack, your API usage, and where your team is doing manual work that a script should handle. Half the time the answer isn't a new app at all — it's deleting two you're already paying for. We run a free Shopify store audit that flags exactly that: performance drag from installed apps, automation gaps, and the handful of places where a custom build would genuinely pay off. Send us your store URL and we'll tell you what we find, including when the answer is "don't build it."