The wrong decision here costs years and millions of euros—either building software that duplicates mature products, or bending operations around tools that will never fit your strategy. As a Sweden-based development company, we bias toward pragmatism: buy commodity capabilities, build where you differentiate, and be ruthless about what “differentiation” actually means to customers and investors. This article gives CTOs and VPs a repeatable framework, numeric guardrails, and the political language to align engineering, finance, and GTM.
We also assume you operate in environments where trust matters—European buyers ask sharper questions about data residency, subprocessors, and long-term viability than a decade ago. The framework below still applies globally, but evidence and documentation are part of the cost of both paths.
Start with strategy, not features
Ask three questions before you evaluate vendors or sprint plans:
- Will this system create a durable competitive advantage? If yes, custom or heavily configured software may be justified. If no, prefer SaaS and integration.
- Is the workflow standardized across your industry? If yes, a mature off-the-shelf product likely exists—your risk is implementation, not invention.
- What is the cost of being wrong? Regulatory exposure, brand damage, and revenue leakage belong in the same spreadsheet as license fees.
If leadership cannot answer these succinctly, pause procurement. You are about to fund either bespoke mediocrity or SaaS spaghetti.
The total cost model most decks miss
Off-the-shelf TCO = licenses + implementation + integrations + training + annual uplift + internal admin time. Enterprise SaaS commonly rises 4–9% yearly; multi-year deals may cap increases—negotiate early.
Custom TCO = build + maintenance (often 15–25% of build annually) + cloud + security + team retention + opportunity cost of delayed features.
Rough planning anchors from EU projects: a focused custom B2B web app might be EUR 80,000–200,000 to reach production; a mid-market SaaS suite might charge EUR 40–150 per user/month list for advanced modules—before consultants reshape processes to fit the tool.
Compare five-year TCO, not sticker price. A EUR 120,000 custom module with EUR 25,000/year maintenance can beat a EUR 90,000/year license stack if the custom module tightly automates margin-driving work—but only if maintenance and staffing are honest.
When off-the-shelf wins
Choose packaged software when:
- The domain is mature (CRM, HRIS, core accounting rails) and vendors compete on breadth.
- Time-to-value is critical—8–16 week implementations beat 9–15 month builds.
- Security and compliance certifications are table stakes your team does not want to own (yet).
- Integrations to your ecosystem are first-class and actively maintained.
Red flags for SaaS: rigid data models that break your unique workflow, poor API coverage, vendor lock-in with punitive export terms, and roadmap uncertainty if the vendor is PE-roll-up fragile.
Quantify switching costs before you sign: migration tooling, historical data retention, and contractual minimums can turn a EUR 80,000/year line item into a multi-year anchor even when the product stagnates. Ask vendors for reference customers who left—what broke, and what the exit truly cost.
When custom build wins
Build when:
- Workflow is your moat—operations, pricing engines, routing, or risk models that map to how you win deals.
- Off-the-shelf forces harmful compromises that show up in conversion rates or unit economics.
- You need deep integration with proprietary data platforms where glue code becomes the product.
- Latency, privacy, or data residency constraints make multi-tenant SaaS unacceptable.
Red flags for build: unclear product ownership, second-time founder overconfidence without discovery, and “not invented here” culture that ignores commodity layers.
If you build, fund product ownership explicitly—a single accountable owner with budget authority beats a committee that “aligns weekly” while engineers ship conflicting priorities.
The hybrid path most enterprises actually take
The best outcomes are often buy the core, build the edge—ERP or CRM as system of record, custom services for customer-facing differentiation, analytics, and automation. Budget EUR 40,000–120,000 for a solid integration layer (eventing, canonical models, retries, observability) bridging SaaS APIs to your stack—this layer is where reliability lives.
For AI features in 2026, buy inference and vector infra where possible; build orchestration, evaluation, and UX where your brand and compliance posture matter.
Decision worksheet (executive version)
Score each option 1–5 on:
- Strategic fit (does it reinforce differentiation?)
- Time-to-value (calendar months to pilot value)
- TCO (5-year) (all-in, internal and external)
- Risk (vendor viability, execution risk, security)
- Flexibility (how painful are future pivots?)
- Talent (can you hire/maintain it?)
Weight strategic fit and TCO highest unless you are in a regulated or safety-critical domain—then risk and auditability dominate.
Organizational politics: how to get alignment
Finance will push Opex and predictable line items—SaaS fits neatly. Engineering may push clean architecture—custom fits neatly. GTM wants features next quarter. Procurement wants three competitive bids and standard terms. Your job is to translate trade-offs into revenue and risk:
- “Off-the-shelf CRM gets us 90% in 12 weeks, but we accept X process change and Y integration tax.”
- “Custom pricing engine ships in 7 months, targets +Z bps margin—requires two FTEs ongoing.”
Numbers end debates; adjectives extend them. Add one slide with reversibility: what it takes to undo the decision in twelve months if assumptions fail—often the most sobering exercise in the room.
Common European procurement realities
GDPR, Schrems II follow-ons, and AI Act-adjacent obligations increase scrutiny on subprocessors and data flows. SaaS can accelerate compliance if vendors present credible artifacts; custom build shifts more evidence burden to you—budget InfoSec and legal time accordingly.
Public sector buyers may require accessibility, local hosting, or specific contract clauses—these can disqualify otherwise attractive SaaS or add six-figure customization.
Mistakes we see repeatedly
Customizing SaaS to death—you pay consulting at build rates for vendor constraints. Building commodity—auth, basic billing, generic admin—because a team wants a greenfield repo. Ignoring maintenance—custom software rots without owners. Underinvesting in change management—great tools with untrained users fail in adoption metrics.
Practical recommendation
Default to buy for standard domains; build where your P&L or customer promise depends on unique workflows; integrate deliberately with a first-class internal platform mindset. Revisit the decision every 12–18 months—the line moves as vendors add AI and APIs improve.
If you need an external build partner, favor teams that challenge your scope and document assumptions—confidence without curiosity is expensive. The right decision is rarely pure custom or pure SaaS; it is a portfolio that matches your risk appetite and time horizon.
Deeper dive: integration tax and “second system” risk
Every SaaS tool introduces integration tax: identity mapping, duplicate detection, eventual consistency between systems, and operational runbooks when APIs change. A common mistake is to assume iPaaS solves this cheaply—connectors accelerate happy paths, but edge cases still consume senior engineering time. Budget EUR 25,000–80,000 for a serious first-phase integration program when three or more business-critical systems must agree on customer truth.
Custom build carries second-system risk: teams underestimate how long parity features take—search, permissions, audit, reporting—because employees compare against mature SaaS unconsciously. Mitigate with explicit non-goals in v1 and parity roadmaps tied to revenue milestones, not vanity.
Metrics that justify custom spend
Executives should expect leading indicators within two quarters of launch: cycle time down 15–35% on targeted workflows, error rates down measurably, or conversion up on a funnel step tied to the new system. If you cannot define before/after measurement, you are funding technology theater.
For off-the-shelf, track adoption, time-on-tool, support volume per user, and renewal health by cohort. A cheaper license with 40% lower active usage is not cheaper—it’s waste with better optics.
Vendor diligence checklist (short)
For SaaS: API SLAs, export formats, uptime history, security artifacts (SOC 2 Type II, pentest summaries), roadmap cadence, and exit clauses. For custom partners: reference calls, code access expectations, warranty window, and documentation deliverables. Time invested here returns 10× during implementation.
Nordic labor market context
Hiring senior product engineers in Stockholm, Gothenburg, or Copenhagen remains competitive—total compensation pressure and long notice periods slow internal builds. That does not automatically mean “outsource everything”; it means portfolio decisions should include calendar time as a first-class variable. Sometimes six extra months to market costs more than EUR 100,000 in margin—model it explicitly.
Summary for the board
Present decisions as bets with costs, timelines, and kill criteria. Show why packaged software cannot meet a specific workflow without unacceptable compromise, or why custom build is unnecessary because vendor depth and integrations cover your differentiation elsewhere. Clarity beats ideology—your job is to make the portfolio rational, measurable, and revisitable.
Finally, write down who owns the decision six months later. SaaS ownership often sits with IT and business ops; custom ownership sits with engineering and product. Misaligned ownership produces orphaned systems—technically fine, organizationally abandoned—so assign executive sponsors and review cadences the same way you would for any material operational change.
