Build vs Buy · Custom Software · Pulse Software Solutions
Jump To Section
I started building custom business software in the 1990s. CRMs, inventory systems, automation tools, all of it written from scratch for one company at a time. It was expensive, slow, and every system came with a lifetime commitment to maintenance, security patches, and upgrades. When SaaS arrived, I watched it eat that entire world, and honestly, it deserved to.
Now, 25 years later, I am watching the pendulum swing back. And this time the math is on the other side.
A Short History
In the 1990s and early 2000s, if a mid-sized business wanted a CRM that matched how they actually worked, they hired a firm like ours to build it. The software fit like a glove, but the total cost of ownership was brutal. You paid for the build, then you paid for the servers, the security, the upgrades, and the one developer who understood the codebase and could never be allowed to leave.
SaaS fixed all of that. One subscription, no servers, no upgrade projects, no key-person risk. Salesforce, HubSpot, Monday, Asana. You swiped a card and had working software the same afternoon. For about fifteen years, subscribing beat building almost every single time, and I told plenty of clients exactly that.
Then the subscriptions multiplied. The average company now runs 106 SaaS applications, and companies spend an average of $5,607 per employee per year on them, according to BetterCloud’s State of SaaSOps research. Worse, roughly half of those licenses sit unused. Zylo’s research puts unused licenses at 53% within 30 days of purchase.
The average organization now handles 211 SaaS renewals per year. That is four contract negotiations every week just to keep the tools you already have.
Pricing Pressure
Here is the part that changed the equation fastest. SaaS companies became attractive acquisition targets, and private equity firms noticed. PE software deals grew 28% in 2024, and the playbook after acquisition is predictable: raise prices on customers who are too embedded to leave. CIO Magazine reported price increases from PE-owned vendors running as high as 900% in extreme cases.
Even outside PE ownership, the trend is one-directional. Vertice recorded SaaS prices rising 11.4% year over year in early 2025, against 2.7% average G7 inflation. In 2024, 58% of software vendors raised list prices. Your software costs are compounding at four times the rate of everything else in your budget, and you have no control over it.
For years, my answer to frustrated clients was still to grit your teeth and pay it, because custom was worse. The overheads of building, the security burden, the long-term upgrade path. SaaS with a price hike was still cheaper than custom with all its baggage.
“That answer is no longer true.” – Manoj Manghnani, Pulse Software Solutions
The Turning Point
Three things happened at once, and together they rewired the build versus buy decision.
Tools like Claude Code, Cursor, and Lovable are not autocomplete gimmicks. They are how modern teams like ours actually ship. GitHub’s own controlled research found developers complete tasks 55% faster with AI assistance, and AI now generates roughly 46% of code written on the platform, a figure Gartner expects to hit 60% of all new code by the end of 2026. In the GoodFirms 2026 Custom Software Development Cost Survey, 90.6% of development firms reported using AI across the development lifecycle, and 61% expect it to cut project budgets by 10 to 25%. In our own work at Pulse Software Solutions, the reduction in man-hours on the right kind of project has been far larger than that, because AI compresses not just the coding but the testing, documentation, and iteration cycles.
The old fear was legitimate: a custom system meant you owned every vulnerability. Today we lock systems down behind platforms like Cloudflare Zero Trust, so the application is never exposed to the open internet at all. Only authenticated users on approved paths can even reach it. That architecture drastically shrinks the attack surface compared to the old days of a custom app sitting on a public server. Meanwhile, the SaaS side of the ledger has its own problem: 78% of organizations store sensitive data in third-party SaaS apps, and every vendor in your stack is a breach you cannot prevent, audit, or control.
A SaaS product is designed for millions of users, which means thousands of features you will never touch, settings screens nobody understands, and workflows you bend your business around. A custom system is designed for your 500 users. It does what your business does. No workarounds, no conversations that start with the tool not really supporting that, so we export to a spreadsheet. Simpler software is cheaper to build, easier to adopt, and far easier to maintain.
On The Ground
We have built replacements for standard project management systems and CRMs for clients where the off-the-shelf tool never quite fit. The pattern is always the same. The client was paying for a big platform, using maybe 20% of it, and running the rest of their process through workarounds, exports, and duct tape. The custom replacement did less on paper and more in reality.
And the ownership terms are completely different:
The software is yours. Not licensed, owned.
Your data sits where you decide, under your policies, not scattered across vendor clouds.
Nobody can acquire your software and triple the price, because there is no subscription to reprice.
Hosting and maintenance still exist, but they are small, stable line items instead of compounding per-seat fees.
Custom software is the expensive bespoke option, so buy unless you absolutely cannot.
At 2026 development costs, purpose-built systems are often the simpler, cheaper, safer choice for the tools that almost fit.
I want to be honest about the other side, because I have been doing this too long to pretend there is not one. Custom software is not zero-maintenance, and it is not the right answer for everything. If a SaaS tool fits your workflow cleanly and the pricing is sane, keep it. Nobody should custom-build email or accounting. The comeback is happening in the middle: the CRMs, project trackers, quoting tools, and operational systems where the standard product almost fits, and almost costs you real money every day.
Run The Numbers
For two decades the decision tree was simple: buy unless you absolutely cannot. The new version is more interesting. Add up what you actually pay for the tools that almost fit. The subscriptions, the unused seats, the integration glue, the hours lost to workarounds, and the compounding renewals. Then price a purpose-built system at 2026 development costs, not 2015 costs.
The tool fits your workflow cleanly, pricing is sane, and nobody is building spreadsheets around it. Email and accounting live here permanently.
Seats are going unused, renewals keep climbing, and integration glue is growing. Price the alternative before the next renewal lands.
You use a fraction of the platform, your process runs on workarounds, and the vendor has you too embedded to leave. This is where custom wins.
For a growing number of businesses, that comparison now lands on the side nobody expected. Custom is back, not as the expensive bespoke option, but as the simpler, cheaper, safer one.
If you have run those numbers for your own stack and they surprised you, I am happy to compare notes. After 25 years on both sides of this cycle, the least I can offer is an honest read on whether building makes sense for your situation.
Manoj Manghnani builds AI-powered software and automation at Pulse Software Solutions, serving clients across the US, UK, Australia, and Canada. Schedule a consultation to talk through your build vs buy math.
Send us your current stack and we will price a purpose-built alternative at 2026 development costs, with an honest answer either way.
© 1998-