Open the bookkeeping for a typical ten-person business on Vancouver Island and count the software subscriptions. Scheduling tool, invoicing tool, forms tool, a CRM nobody fully uses, two messaging products, a file-sharing plan, and something somebody signed up for in 2023 that still bills $49 a month. Twelve to twenty line items is normal. $800–2,000 a month is normal.
Now ask the harder question: does the work flow through any of it?
In most businesses, it doesn’t. The work, meaning the actual sequence of who-does-what that turns a phone call into a finished job into a paid invoice, lives in the gaps between the tools. It lives in the dispatcher retyping the booking into the scheduling app, the technician texting photos that someone later drags into a folder, the owner reconciling three systems every Sunday night, and the forty-tab spreadsheet that almost works because Karen built it over six years and Karen understands it.
The subscriptions are tools. The spreadsheet-and-retyping layer is the operating model. And the quiet shift of the past few years is that businesses have started turning that layer, the workflow itself, into software they own. Not another tool in the stack. The thing the stack was always failing to be.
Why custom applications became affordable: three things changed
Custom software has always been better than generic software at fitting a specific business. What changed is that it used to cost six figures and most of a year, which priced out everyone except enterprises. Three shifts broke that.
The foundations got industrial. The components that used to consume most of a custom build’s budget, including authentication, databases, hosting, and payments, are now mature, rentable infrastructure. The same foundations the SaaS giants ship on are available to a ten-person plumbing company, by the month, in a Canadian data centre. Nobody pours their own concrete anymore. You build on the slab.
One codebase now ships everywhere. A build that once meant three separate projects for iOS, Android, and web is now one project shipping all three. One cost, one maintenance surface. This alone roughly cut the price of “my crew needs it on their phones” by two-thirds.
AI compressed the labour. The repetitive majority of development work, the scaffolding, boilerplate, and test generation, got dramatically faster. The judgment work didn’t compress (we’ve written a whole piece on why it shouldn’t), but the typing did, and the typing was most of the bill.
Stack the three and the arithmetic flips. The workflow application that cost $150K in 2019 lands at $14–25K in 2026, built in 8–14 weeks instead of a year. That’s no longer an enterprise capital project. That’s a used truck: a serious purchase a small business makes when the case is clear, and the case has gotten very clear.
What faster-growing businesses do differently with software
The pattern among operators pulling ahead isn’t “more software.” Plenty of struggling businesses have twenty subscriptions. The pattern is more specific, and it’s worth stating as a rule:
They identified the one workflow where coordination was eating them alive, and they turned that workflow into an application.
Not the whole business. One workflow. The booking-to-dispatch-to-invoice chain. The member sign-up-to-renewal loop. The intake-form-to-approval pipeline. These workflows share a signature: high frequency, multiple hands, and information that has to be retyped or chased between steps. Every retype is minutes. Every chase is an interruption. Every handoff is a place where a job can quietly fall on the floor.
The arithmetic on coordination waste is unglamorous and decisive. A dispatcher spending 90 minutes a day shuttling information between a phone, a whiteboard, and an invoicing tool is 30+ hours a month. At $30/hr, that’s roughly $11,000 a year, every year, for one person, before counting the missed jobs and the double-bookings. A four-technician service business of the kind we describe in our reference builds cuts routine coordination calls by something like 40% when crews can see their day, capture photos and signatures, and close out jobs from the truck. The application doesn’t do anyone’s job. It removes the friction between jobs, and friction, unlike labour, removes cleanly.
There’s a second-order effect that matters more over time: when the workflow becomes an application, the business finally gets its own data. Not data scattered across five vendors’ dashboards. The actual operating record. Which jobs are profitable. Which customers come back. Where the Tuesdays go wrong. You can’t see any of that in the gaps between tools, because the gaps don’t keep records. And every AI capability worth having downstream depends on that record existing.
Own vs. rent: the software question nobody asks until it hurts
Here’s the 2026-specific part, because the calendar isn’t incidental to this argument.
The SaaS model spent two decades training businesses to rent their workflows. Renting was rational when owning cost $150K. But the rental terms have been deteriorating in plain sight: per-seat prices ratcheting annually, features moving behind higher tiers, products getting acquired and sunset, and the data, your operating record, the thing compounding in value, living in someone else’s system. Exportable in theory. Trapped in practice.
A subscription to your own workflow is a strange thing to hold once owning it costs less than two years of the rent.
That’s the real meaning of “applications matter more in 2026.” Not that software got more important. It’s been important for twenty years. What changed is which side of the rent-versus-own line the important workflows sit on. When we hand off a build, the client gets the repository, the hosting accounts, the documentation, and a recorded walkthrough. The application is an asset on their side of the table. Nobody can raise the rent on it, sunset it, or move its features to a higher tier. We build it. You own it.
When custom software is the wrong answer
Custom is not the answer to everything, and a piece like this owes you the boundary.
If your need is generic, meaning accounting, email, payroll, or a website, buy the SaaS. It’s excellent, it’s $40/mo, and a custom build would be a worse version of it at a thousand times the cost. The case for custom lives in one place only: the workflow that is specifically yours, where the generic tools keep almost-fitting and the gap is being filled by retyping, chasing, and Karen’s spreadsheet. Some of our Blueprint™ engagements end with “don’t build this; here’s the SaaS that already does it.” That answer costs $2,500 and saves $15,000, and we consider it a successful project.
But if you’ve duct-taped four tools together and the duct tape has a name and a salary, that’s the tell. That workflow is the business. In 2026, for the first time, it costs less to own it than to keep renting around it.
The companies growing faster aren’t smarter about software. They just did that arithmetic a year before their competitors.