The three questions that define an integration
First: what is the source of truth for each data type? Usually the ERP owns products, prices and stock, while Shopify owns orders and customers. Writing this down prevents the classic disaster of two systems overwriting each other.
Second: how fresh does data need to be? Stock might need near-real-time webhooks; prices might be fine syncing nightly. Real-time everything sounds nice but costs more and breaks more; most businesses need real-time for exactly one or two data types.
Third: what's the failure plan? Retries with backoff, idempotent operations so nothing duplicates, and alerts that reach a human. I build these in by default because integrations without them silently corrupt data.
Systems I've connected
Microsoft Dynamics 365 Business Central (B2B contract pricing synced into Shopify), Lightspeed POS (live repair pricing), Recharge (custom subscription portals), Klaviyo (events and profiles), Mollie and Stripe (payments), print-on-demand APIs, review platforms, and even legacy systems with no modern API at all, including one where I built a compatibility middleware so a factory's production-planning software survived a WooCommerce-to-Shopify migration unchanged.
If your system is obscure, that's rarely a problem. If it has documentation, I'll read it. If it doesn't, I'll reverse-engineer it carefully.
What it costs and how long it takes
A one-directional sync of one data type (say, nightly price updates) is often a one-to-two-week project. Full two-way integrations with multiple data types, reconciliation tooling and an admin panel are multi-week to multi-month. The scope conversation is free: bring what you know about your system's API and I'll tell you what's realistic.