For a Singapore SME comparing SaaS to custom software for one workflow, the right answer is not the cheaper licence. It is the cheaper three-year total cost of ownership (TCO) once you add implementation, admin time, workarounds, change fees, hosting and security, maintenance, and exit or export cost — plus who controls the roadmap and how reversible the choice is. SaaS usually wins when the workflow is commodity and seats stay stable. Custom usually wins when the workflow is how you compete, change requests never stop, or you must own the data path and exit terms.

Educational only; not financial, tax or procurement advice. Any Singapore-dollar figures below are labelled illustrative placeholders, not quotes or benchmarks.

Who this is for, and who should skip it

For: operations, finance and IT leads at Singapore SMEs who must choose SaaS subscription versus a custom build (or a thin custom layer) for one named workflow — order intake, field scheduling, partner onboarding, inventory exceptions, claims — and need a board-ready TCO sheet, not a vendor pitch.

Skip if: you are choosing between a low-code platform and hand-written code (read low-code vs custom for Singapore SMEs first — that is platform versus code, not subscription versus build); you are already locked into rebuilding a full ERP or payroll core; or you need a grant-eligibility walkthrough. This article does one job: help you compare SaaS TCO Singapore-style against a custom option using control and reversibility, not sticker price.

The decision is not price versus price

Licence price sits on one line. Three-year TCO sits on ten. A SaaS quote that looks light in month one can still lose once seat growth, integration work, workaround labour, and a painful export land in years two and three. A custom quote that looks heavy up front can still win if change volume is high, the workflow is unique, and you keep clean ownership of data and code.

Treat the decision as a portfolio choice for one workflow: buy commodity systems of record when the market already solved the problem well; build (or commission) a thin workflow layer when the process is how you differentiate. That hybrid pattern shows up again after the worksheet.

Three-year TCO worksheet: SaaS versus custom

Copy this table into your next internal memo. Fill categories with your own numbers; do not paste a vendor’s marketing total as TCO. Categories matter more than invented amounts.

TCO line item (3 years)	SaaS subscription	Custom / build
Licences / seats / modules	Recurring per seat or usage; often rises with headcount and add-ons	Usually none after go-live; optional third-party libraries
Implementation / configuration	Vendor or partner setup, data mapping, training	Discovery, build, UAT, cutover; larger up-front slice
Integrations	Connector fees, middleware, brittle sync jobs	Built APIs and jobs you own; still costs engineering time
Admin / power-user time	Ongoing configuration, ticket triage, user provisioning	Lower day-to-day config if UX fits; still needs product owner time
Workarounds / shadow process	Spreadsheets and side tools when the product cannot flex	Lower if scope matched reality; high if you under-scoped
Change / request fees	Paid change packs, professional services, delayed roadmap	Sprint cost you control; backlog prioritisation is yours
Hosting / security / compliance ops	Mostly in subscription; still your DPAs, access reviews, PDPA duties	Cloud, monitoring, patching, pen-test budget you carry
AMS / maintenance / support	Tiered support; premium SLAs cost extra	Application management or in-house AMS retainers
Exit / export / parallel run	Export quality, format, and notice period decide the real cost	Repo, docs, and runbooks you already own if contracted well
Parallel-run / dual-system period	Often months of SaaS plus old process	Often months of old process plus new build before cutover

Sum every row for three years. Then ask two non-money questions: who sets the dependency terms (roadmap, pricing changes, data residency options), and how reversible is the choice if the workflow changes?

Illustrative worked example: order-exception workflow

Constructed example for illustration, not a client case. SGD figures are placeholders to show the arithmetic — replace with your quotes.

A Singapore wholesale distributor needs better handling of order exceptions (partial ship, credit hold, special pricing). Two options on the table:

SaaS path (illustrative): S$480 per seat per year × 25 seats = S$12,000/year in licences (S$36,000 over three years). Implementation and training S$18,000. Two integrations S$22,000. Estimated admin time S$15,000 over three years. Change packs S$12,000. Workarounds (one FTE-day per week equivalent, illustrative) S$30,000. Exit/export rehearsal and parallel run S$10,000. Illustrative three-year SaaS TCO ≈ S$143,000.
Custom thin layer (illustrative): Build a workflow app that sits on top of the existing ERP/API: discovery and build S$95,000. Hosting and security ops S$18,000 over three years. AMS retainer S$36,000 over three years. Change capacity already inside AMS/sprints (no separate change packs). Exit cost near zero for data you already store. Parallel run S$8,000. Illustrative three-year custom TCO ≈ S$157,000.

On sticker price alone, SaaS looks cheaper. On three-year TCO they are close. The deciding factors become control and reversibility: if exception rules change every quarter and sales leadership owns the edge cases, custom (or a hybrid) often wins even when the totals are similar. If exceptions are rare and the vendor’s standard rules fit, SaaS wins and you keep engineering time for something else.

Control, reversibility, and risk

TCO without control is incomplete. Ask who can raise prices, retire a feature, change an API, or move data residency. With SaaS, the vendor sets most of those terms; your leverage is contract length, exit clauses, and whether a credible alternative exists. With custom, you set the roadmap, but you also carry delivery and security risk — which is why a clear statement of work and a vendor scorecard matter as much as the build estimate (see SOW red flags for Singapore procurement and the vendor scorecard).

Reversibility is the other half. Can you export complete records in a usable format within a defined notice period? Can you run reporting without the vendor UI? For personal data involving Singapore individuals, the PDPC’s summary of the data protection obligations (including Transfer Limitation when data leaves Singapore) still applies whether you buy or build — the difference is whether safeguards sit in a vendor DPA or in your own systems and contracts.

The hybrid pattern that often wins

Many Singapore SMEs do not need an all-or-nothing choice. A durable pattern is:

Buy commodity systems of record (accounting, payroll, standard CRM, email) where the market is mature and switching cost is understood.
Build a thin workflow layer for the process that creates margin or customer experience — the exception engine, the partner portal, the scheduling brain.
Integrate carefully so the custom layer reads and writes through APIs instead of becoming a second system of record.

That is often cheaper over three years than forcing a unique process into a SaaS configuration maze, and safer than rebuilding accounting or payroll from scratch. If you need product shape before a full build, an MVP development path can prove the workflow with a bounded budget. Teams that later productise an internal tool sometimes move toward SaaS product development — a different decision from buying someone else’s SaaS.

Six-question decision scorecard

Score each question Yes / Partial / No. More Yes answers on questions 1–3 tilt SaaS; more Yes on questions 4–6 tilt custom or hybrid.

Is this workflow commodity in your industry? Yes → SaaS lean. No → custom lean.
Will seat count and module needs stay fairly stable for three years? Yes → SaaS lean. Rapid growth or many add-ons inflate SaaS TCO.
Does the vendor’s standard object model fit without heavy workarounds? Yes → SaaS lean. If you already run shadow spreadsheets, budget them as TCO.
Is this workflow how you compete or protect margin? Yes → custom or hybrid lean.
Do you expect frequent rule changes that cannot wait for a vendor roadmap? Yes → custom lean.
Do you need contractual control of export, data residency options, and IP for this layer? Yes → custom lean (with written assignment and exit terms).

If questions 1–3 say SaaS and 4–6 say custom, prefer hybrid: keep the system of record in SaaS, build the thin differentiating layer, and write export and API access into both contracts.

Mistakes and traps

Comparing a SaaS licence to a project quote. Put both on the same three-year worksheet first.
Ignoring seat growth. A headcount plan that doubles field staff doubles many SaaS lines.
No export clause. Exit cost is not theoretical; rehearse a sample export before you sign.
Building accounting or payroll. Those are regulated, mature categories — buy unless you have a rare, counsel-backed reason.
Confusing low-code with buy-vs-build. Low-code is how you might build; SaaS vs custom is whether you subscribe to someone else’s product. Use the low-code decision frame only after you know you are building.
Skipping AMS. Custom without maintenance budget becomes shadow IT. Plan application management before go-live.

FAQ

What does SaaS TCO Singapore usually miss on the first spreadsheet?

Admin time, workarounds, integration upkeep, change packs, and exit/export. Licence and implementation alone understate three-year cost.

When does custom software beat SaaS on three-year TCO?

When change volume is high, the workflow is unique to how you sell or operate, seat-based pricing would scale painfully, or you need ownership of the data path and exit terms. Run the worksheet; do not assume.

Should we ever mix SaaS and custom?

Yes — often. Buy systems of record; build thin workflow layers. That hybrid is the default for many Singapore SMEs that want both speed and control.

How does PDPA affect buy vs build?

Obligations follow the personal data, not the commercial model. SaaS needs a sound DPA and transfer safeguards if data leaves Singapore; custom needs your own controls, contracts and logging. Start from the PDPC’s data protection obligations overview.

What engagement model fits a custom thin layer?

Fixed-scope for a bounded MVP, then time-and-materials or a retained squad for change — see engagement models and custom software in Singapore.

Next step

Pick one workflow. Fill the ten-line TCO worksheet with real quotes and honest admin-time estimates. Score the six questions with ops and finance in the same room. If the sheet points to a custom or hybrid path, talk through scope and exit terms with a software development partner in Singapore, or contact WC1 for a structured build-vs-buy workshop on that single workflow.

Build vs Buy SaaS TCO Framework for Singapore SMEs