Saas vs custom is reshaping how we work. The instinct is understandable. A new SaaS subscription costs £49 per user per month. A custom web application starts at five figures. The maths looks obvious until you actually run it.
We’ve now sat across the table from dozens of businesses who chose SaaS for the right reasons three years ago and are now trapped in workflows they cannot change, paying for seats they don’t need and watching competitors build digital products that compound while their tooling stagnates. The honest cost comparison is rarely the one your finance director starts with.
Where the SaaS vs custom calculation breaks
The headline subscription fee is the smallest cost in the equation. The real costs accumulate quietly:
Per-seat pricing scales linearly with growth. A platform that costs £15,000 per year at 25 users costs £60,000 at 100. There is no leverage. Every new hire is a new tax.
Workflow constraints become operational debt. SaaS tools are designed for the median customer. The further your business is from that median – and ambitious businesses always are – the more your team works around the tool. Spreadsheet exports. Manual reconciliation. Shadow processes. Each one is a hidden tax on every transaction you process.
Integration fees and middleware. The moment you have more than three SaaS tools, you need them to talk to each other. Zapier, Make, integration platforms – another £10,000 to £30,000 per year, plus the engineering time to maintain them.
Data lock-in. Your operational data lives inside someone else’s database, structured according to their model, exportable only on their terms. The strategic value of that data – the patterns, the trends, the AI training material – is harder to extract than you think.
Competitive parity, not advantage. Every competitor in your sector is using the same SaaS tools. Whatever workflow advantage they offer is, by definition, available to everyone. You cannot out-execute on infrastructure your rivals also have.
Where custom web applications win
a custom-built web application is not always the right answer. For genuinely commodity functions – email, video calls, basic accounting – SaaS is the rational choice. But for anything that touches your operational moat, custom changes the economics.
Cost stops scaling with users. A custom application’s running costs are infrastructure (hundreds to low thousands per month) plus maintenance – both flat or sub-linear with growth. At 100 users you are saving five-figure sums annually. At 500, six.
Workflows fit the business, not the other way around. Every step you remove from a process saves time across every transaction, every day, forever. Eliminating one minute from a workflow done 200 times a day is 1,200 hours a year reclaimed. That number alone often justifies the build.
Data becomes a strategic asset. When you own the data model, you can analyse it, train models on it, expose it to AI agents and build products around it. Your operational data is your competitive advantage – it should not live behind someone else’s API.
Integration is native, not bolted on. Custom applications are built around your specific integration needs. No middleware. No mapping fields between tools that were never designed to talk.
It compounds. Every feature you ship makes the next one easier. SaaS plateaus the day you sign the contract. Custom builds get more valuable every release.
The decision framework we use with clients
We don’t recommend custom builds reflexively. The honest test is four questions:
- Is this function central to your competitive position, or commoditised? Build the central. Buy the commodity.
- Will the cost of SaaS exceed the cost of custom within three years at projected growth? If yes, build.
- Are your workflows materially different from the SaaS market median? If yes, the workaround tax will eventually exceed the build cost.
- Does the data this system holds have strategic value beyond the function itself? If yes, you cannot afford to rent it.
If three of four answer yes, the case for custom is strong. If all four, it’s overwhelming.
What “custom” actually looks like in 2026
The custom web application of 2016 was a 12-month, six-figure undertaking. The custom web application of 2026 is something quite different.
modern frameworks (Next.js, Remix), serverless infrastructure (Vercel, Modal, Supabase), AI-assisted development, mature component libraries and design systems have collapsed both timelines and budgets. We routinely deliver MVP-grade custom applications in 8-12 weeks for sums that compare favourably with two years of mid-market SaaS.
The build itself is no longer the bottleneck. The bottleneck is clarity – knowing what you actually want, why and how it fits into the rest of your operation. Which is why our process starts with Discover and why we offer rapid prototyping for clients who need to test the idea before they commit to building it.
The strategic question
The SaaS-vs-custom debate gets framed as a cost question. It isn’t. It’s a question about whether your digital infrastructure is going to be a tax you pay to other businesses, or an asset that compounds inside your own.
Ambitious businesses build assets.
Wondering whether your next system should be built or bought? We help clients run that calculation honestly, including the parts SaaS vendors don’t put in their pitch decks. Start the conversation here.