Does the catalogue fit Shopify's product model?
Shopify now permits up to 2,048 variants on one product. That is a hard number, but it is not a recommendation to put every possible configuration into one product.
Shopify advertises unlimited products, so a large ordinary catalogue is not itself a reason to build custom commerce.
The answer flips when the catalogue is not really a catalogue. Made-to-measure products, dependent options, negotiated bundles, compatibility rules or marketplace inventory may need a model that fights Shopify's assumptions.
There is also an operational threshold. Shopify applies additional Admin API throttling after 500,000 product variants, limiting specified mutations to 10,000 new variants per day. Before that boundary matters, the better question is whether merchandisers can understand and maintain the data model without engineering help.
Can Shopify own the checkout?
This is the decisive question many stack comparisons hide. A standard store needs taxes, payments, fraud controls, discounts, shipping and refunds to work together. Shopify already owns that difficult surface and is PCI DSS Level 1 compliant.
Choose Shopify when the business can express its sale through Shopify's checkout and payment model. A custom storefront can still sit in front of it when the brand experience demands more freedom.
Move toward custom commerce when checkout itself is the product: a multi-party marketplace, account-specific contractual pricing, regulated approval before purchase, unusual settlement logic or a transaction that is not an ordinary cart and order. Forcing those models through discount codes and private apps creates a custom platform disguised as a Shopify store.
Country also matters. Shopify states that payment availability and pricing vary by location, while third-party payment providers can add plan-dependent transaction fees on top of their own processing charges. The official pricing page should be checked for the merchant's actual market before any revenue model is compared.
Is the storefront genuinely differentiated?
A brand does not need custom commerce because the design team dislikes templates. A well-engineered Shopify theme can be fast, distinctive and easy for a trading team to operate.
Custom storefront work becomes justified when content, product discovery and interaction are inseparable from the commercial advantage. Examples include editorial launches assembled from reusable stories, complex product finders, shared inventory across several digital surfaces or an experience whose interaction model cannot live inside a theme.
Even then, there are two different decisions. Headless Shopify replaces the storefront while keeping Shopify's commerce engine. Fully custom commerce replaces more of the engine itself. The second choice carries a much larger operational surface and should not be smuggled into a design conversation.
Who will own the difficult Tuesday after launch?
Shopify fits teams that want merchants to handle products, promotions and orders without waiting for developers. Its constraint is also its value: a vendor maintains the checkout, platform security and much of the operational machinery.
Custom commerce fits when the business has a continuing product and engineering function, not only a launch budget. Someone must own releases, incidents, payment changes, security updates, data repairs and the admin tools used by operations. An agency can provide that ownership, but it remains an ongoing product relationship.
This is where revenue becomes relevant, but there is no honest universal revenue threshold. The flip happens when the annual cost of Shopify-specific fees, workarounds, lost capability and constrained conversion exceeds the annual premium of owning custom software. High revenue makes small percentage costs meaningful, but it also raises the cost of downtime and mistakes.
Run the decision this week
Take a representative sample of products, orders, discounts, returns and international sales. Model them in Shopify without custom code. Mark every place where the team has to distort a product, bypass checkout or depend on a private integration.
Then compare annual totals, not launch quotes. Include platform fees, apps, payment differences, hosting, development, support and internal operating time. Assign a named owner to every custom component.
If the catalogue and checkout fit, start with Shopify. If only the experience does not fit, use a custom storefront over Shopify. Replace the commerce engine only when the transaction model itself demands it and the business is prepared to operate what it owns.