Five-year cost: SaaS vs custom (without the spreadsheet fantasy)
Monthly SaaS feels cheap until workarounds, seats, and lock-in show up on the bill. Here is how to think in five-year totals without pretending the spreadsheet is precise.
Vinerals TechnologiesWorkshop notes
The SaaS pitch is monthly. The custom quote is upfront. So the meeting ends with someone multiplying the subscription by twelve, comparing it to the build price, and declaring custom insane before lunch is over.
That math leaves out most of what you will actually pay. SaaS stacks grow seats, add-ons, connectors, and workarounds. Custom software has a bill after launch too. Five years is where the picture gets honest, if you count everything and stop pretending one decimal place means certainty.
This piece is a TCO frame for finance-aware operators: what to include, how SaaS creeps, what custom really costs after go-live, a worked example in ranges, and when the line that looks expensive on day one is cheaper by year five.
What to count (and what people forget)
Total cost of ownership is not a vendor invoice. It is every dollar and every hour the choice costs you over the period you care about. For most SMEs comparing SaaS and custom, five years is a useful horizon: long enough for subscriptions to compound, short enough that you are not forecasting fiction.
On the SaaS side, count all of this
Core licences.
Per seat, per site, per module. Use your realistic headcount in year three, not today’s.
Annual increases.
Plan for renewal bumps. Five to eight percent per year is a common band unless you have a fixed contract.
Adjacent tools.
The reporting add-on, the integration platform, the form builder, the BI layer. Stacks rarely stay one product.
Implementation and migration.
Setup fees, data migration, consultant hours, the first year of tuning.
Internal labour.
Admin time, training new hires, managing vendor tickets, building exports because the API costs extra.
Workarounds.
Spreadsheets, manual re-entry, duplicate data entry because the SaaS does not fit the process. Labour has a cost even when it is not on the vendor’s invoice.
Exit cost.
Export fees, migration to the next tool, double-running systems during cutover. Lock-in is a line item, just often paid late.
On the custom side, count all of this
Initial build.
The quote for design, development, testing, and launch.
Hosting and infrastructure.
Cloud, backups, monitoring, certificates. Often CAD $500 to $3,000 a month depending on scale and compliance.
Maintenance and change.
Security patches, dependency updates, small feature requests. Budget fifteen to twenty-five percent of the original build per year as a planning band, not a promise.
Internal product ownership.
Someone steers priorities, accepts work, tests releases. It can be a fraction of a role, but zero is how systems rot.
Major enhancements.
New modules, new integrations, regulatory changes. Phase two is real money. Decide upfront whether it is funded or deferred.
If an item is unknown, use a range and note the assumption. Ranges are honest. False precision is how bad decisions get stamped with a spreadsheet.
How SaaS cost creep actually happens
SaaS rarely jumps in one dramatic line item. It creeps through a dozen small yeses.
Seat growth.
You hired. Every new user is another monthly tick. Features tied to headcount scale even when usage does not.
Tier upgrades.
The report you need sits in the next plan. So does the API access. So does SSO.
Module sprawl.
Inventory module, field service module, advanced analytics. Each reasonable alone. Together they are a second software budget.
Integration tax.
Middleware, consultant hours, brittle Zapier chains that become production infrastructure by accident.
Storage and usage fees.
Documents, photos, API calls, automated jobs. Fine at low volume. Surprising at operational scale.
Professional services.
Annual health checks, reimplementation after you outgrow the config, training every time the UI changes.
Creep is not a moral failure. It is the business model. Your defence is to model year three and year five explicitly, not just year one’s introductory price.
What custom costs after launch
Custom software is not free once it ships. The difference is where the money goes: you are paying for change you choose, not for seats on someone else’s roadmap.
Keep-the-lights-on work.
Hosting, monitoring, security updates, bug fixes. Predictable if someone owns it.
Dependency drift.
Libraries age, browsers change, APIs deprecate. Ignored maintenance becomes an emergency rewrite.
Evolving the product.
New rules, new buyer requirements, new internal process. Custom shines when the change is yours. It still costs labour.
Knowledge continuity.
Documentation and handover matter. If only one contractor knows the deployment, you pay a risk premium every year.
A useful planning split: separate “run” from “improve.” Run is non-negotiable. Improve is where ROI lives. If you cannot fund run, custom is a bad bet no matter how good the build.
A worked example in ranges, not false precision
Imagine a company with about fifty staff running one core operational workflow: scheduling, field data, approvals, and billing export. They are comparing a SaaS vertical platform plus surrounding tools against a custom build focused on the same job.
These are directional CAD bands for 2026 planning. Your numbers will move with sector, compliance, and how much of the stack is already in place.
SaaS path, five years
Years 1 to 2.
Core platform plus two adjacent tools: roughly CAD $6,000 to $9,000 per month all in once seats and modules land. Implementation and migration: CAD $40,000 to $80,000 one-time.
Years 3 to 5.
Renewal increases and headcount: drift toward CAD $8,000 to $12,000 per month. Integration maintenance and internal admin: another CAD $30,000 to $50,000 per year in labour or consultants.
Five-year total (illustrative).
Roughly CAD $550,000 to $850,000, plus exit migration if you leave.
Custom path, five years
Year 1 build.
CAD $150,000 to $220,000 for a focused system with integrations and launch support.
Years 2 to 5 run.
Hosting CAD $12,000 to $36,000 per year. Maintenance at fifteen to twenty-five percent of build: CAD $22,000 to $55,000 per year.
Enhancements.
Plan for one meaningful phase: CAD $40,000 to $80,000 spread across the period, or explicitly defer it.
Five-year total (illustrative).
Roughly CAD $320,000 to $480,000, with more variance tied to how much you improve versus merely run.
In this shape, custom looks higher in year one and often lower by year five. SaaS looks cheaper early and climbs with seats and modules. The crossover is usually somewhere between year three and year four for teams in this size band, assuming the custom build stays focused.
Change one assumption and the picture moves. Thirty staff instead of fifty narrows the gap. Heavy compliance widens custom’s run costs. A SaaS product that fits eighty percent out of the box flattens the SaaS curve. That is why the ranges matter more than the example.
When the “expensive” option is cheaper
Upfront price is a storytelling tool. Five-year TCO is closer to economics. Custom tends to win on total cost when several of these are true.
The workflow is your advantage.
Software that encodes how you win in the market, not how you file expenses.
Seat count climbs but value per seat does not.
You are paying for users who only touch the system twice a week.
The SaaS stack is already crowded.
Three subscriptions and middleware to do one job is a tax.
Integrations are fragile and frequent.
Every vendor update breaks your chain. Custom can own the connector properly.
Data ownership and exit matter.
Migration and export fees are part of the SaaS bill, whether you pay them now or later.
The process is stable enough to amortize a build.
You will use the same core workflow for years, not pivot quarterly.
SaaS tends to win on total cost when the job is commodity, the team is small, requirements change constantly, or a mature product already fits without heroic workarounds. That is a good outcome. TCO thinking should make SaaS the honest choice sometimes, not always push custom.
How to build the comparison without fantasy
You do not need a forty-tab model. You need the same horizon, the same scope, and visible assumptions.
1. Name the job once.
Same workflow, same integrations, same user count trajectory.
2. Use three columns.
SaaS, custom, hybrid. Hybrid is often the real answer.
3. Model low and high bands.
Not one number per line. Show what moves the total.
4. Separate one-time and recurring.
So year-one cash flow is visible, not buried.
5. Include internal hours.
Even rough estimates. They often decide the winner.
6. Write the assumptions on the page.
Headcount, renewal rate, maintenance percent, phase two yes or no.
7. Ask what has to be true for each path to win.
If the SaaS win requires a product that does not exist yet, that is a risk flag, not a plan.
A spreadsheet that survives contact with leadership is one where anyone can point at a line and say “we assumed X.” If nobody can, the number is decoration.
A short decision lens after the math
TCO tells you cost. It does not tell you fit. After the bands, run these questions.
Can we fund run costs for the winner?
SaaS or custom, the recurring line has to be real.
Does either path depend on heroics?
One person’s spreadsheet, one brittle integration, one vendor feature on the roadmap.
What is the cost of being wrong for two years?
Switching tax, lost data, team fatigue. Sometimes that favours SaaS speed. Sometimes it favours owning the core.
Are we comparing a product to a project?
SaaS is renting a product. Custom is commissioning one. The economics differ because the asset differs.
Five-year TCO will not give you a single correct answer. It will give you a clearer argument: what you are paying for, what creeps, what you own, and where the crossover might live if your assumptions hold.
We build software by hand for SMEs, and we spend a fair amount of time telling people to keep the SaaS stack when the math and the fit say so. If you are weighing custom against subscriptions and want someone to stress-test the assumptions without selling you a build, we are glad to walk through the bands with you. Bring the monthly invoices and the quote. We will start there.
