The Best-Served Problem in Business Software Still Gets Built by Hand

One problem holds two records in business demand at the same time.

When businesses ask for help with order management, they name a product they already want more often than in any other ask. Shopify above everything, then Stripe, then WooCommerce. Roughly one ask in three arrives with the tool already chosen.

The same problem is also the one businesses most often pay to have custom built. Nearly six in ten of those asks end as commissioned software.

The best-served problem in business software and the most hand-built problem in business software are the same problem.

A named product plus a custom build is a customer paying twice

The two records look like a contradiction. They are actually one sentence.

Shopify owns the storefront. Stripe owns the payment. QuickBooks owns the books. What none of them owns is the part in between: the order's actual journey through this particular business. The split shipment. The supplier who confirms by email. The wholesale customer on payment terms nobody else gets. The stock that lives in two warehouses and one garage.

Look at what the custom work actually is, and it is not rebellion against the products. It is scaffolding around them: API work, database work, inventory sync, the connective tissue between the tools the business already named. The asks come from exactly the businesses whose order flow refuses to look average: food distributors, manufacturers, retailers straddling wholesale and direct.

Buyers are not asking for a better Shopify. They are paying carpenters to make Shopify fit a building it was not drawn for.

The product covers the generic core, and operations is precisely where no two businesses are generic. The order flow is the business. So the buyer names the platform, keeps the platform, and then commissions the part that actually distinguishes their operation, by hand, one build at a time.

"It already exists" is the weakest objection in software

Every builder has killed an idea with that sentence. Somebody already does order management. Somebody already does inventory. The category is taken.

But existence and fit are different things, and demand measures the distance between them in commissioned custom work. A category can be saturated with products and starved of fit at the same time. That is not a crowded market. That is a crowd standing around an unsolved problem.

We wrote about the businesses walking out of Excel and finding no door. This is the stranger sequel: sometimes the door exists, the buyer walks through it, and then pays a carpenter anyway.

So the opening is not where no product exists. It is where products exist and keep failing to fit:

• Sell the fit, not the category. "Order management for food distributors" beats "order management" because the distributor's Friday standing orders and short-dated stock are the actual problem, and the average-shaped incumbent has already proven it will not bend that far.
• Build where the carpenters keep getting paid. Recurring custom work around Shopify, Stripe and QuickBooks is a map of exactly which edges the platforms leave sharp. Inventory across locations, terms-based invoicing, supplier coordination. Each edge is a product.
• Treat the named platform as the ground, not the competitor. The buyer has already chosen Shopify. Do not fight that choice. Be the thing that makes their Shopify behave like their business, and the platform's ubiquity becomes your distribution.

The lesson generalises past orders. Wherever a dominant product gets named constantly and custom builds follow it around like a shadow, the market is telling you the product won the category and lost the customer's specifics.

The specifics are for sale. They have been for sale, one custom build at a time, to whoever finally packages them.

This reading comes from millions of demand signals, collected continuously. Not surveys and not commentary, but real buying activity, the same intelligence that runs inside the product. It is one more way the market moves at the speed of AI: the categories are settled, and the fit never is.