K
KnowMBAAdvisory
Digital TransformationIntermediate8 min read

Build vs Buy Framework

Build vs Buy is the recurring decision: write it ourselves, or pay a vendor? The wrong default has cost companies billions in both directions — Building commodity capability you should buy, or Buying differentiating capability you should build. The right framework asks four questions: (1) Is this capability part of our competitive moat? (2) Is the market for this category mature with credible vendors? (3) What is the True 5-Year TCO of build vs buy (not just year-1 cost)? (4) What is the time-to-value gap, and can we afford that gap? KnowMBA POV: Default to BUY for non-differentiating commodity capability. Default to BUILD only when the capability is core to your customer-perceived value AND no vendor solution lets you express your distinct logic.

Also known asBuild or BuyMake vs BuyBuild Buy PartnerBBP DecisionStrategic Sourcing Decision

The Trap

The two failure modes: (1) 'Build everything' — engineering teams love to build, NIH (Not Invented Here) syndrome runs rampant, and you end up with an in-house auth system, in-house data warehouse, in-house CMS — none of which are your differentiator, all of which now require dedicated teams forever. (2) 'Buy everything' — the company assembles 90 SaaS tools, 60% of which overlap, integration becomes a nightmare, and you've outsourced the actual logic of your business to vendors who can change pricing or get acquired. The trap in BOTH directions is comparing year-1 license cost to year-1 build cost. Always model 5-year TCO including ongoing maintenance, skill retention, and switching cost.

What to Do

Apply a 4-question framework to every significant capability: (1) Differentiation Test: Does this directly create competitive advantage? Score 1-5. If 1-2, lean buy. If 4-5, lean build. (2) Market Maturity Test: Are there 3+ credible vendors with proven scale? If yes, lean buy. (3) 5-Year TCO Model: Build = (initial dev cost) + (5 × annual maintenance ≈ 25% of dev cost) + (opportunity cost of engineers). Buy = (5 × license) + (integration cost) + (switching cost). Compare honestly. (4) Time-to-Value: How much would launch delay cost in revenue/competitive position? Buy is almost always faster. Decide based on the framework, not on whether engineering 'finds it interesting.'

Formula

5-Year Build TCO = Initial Build Cost + (5 × Annual Maintenance) + Opportunity Cost. 5-Year Buy TCO = (5 × Annual License) + Integration Cost + Switching Cost

In Practice

Stripe famously builds its core payment processing logic but buys nearly everything else (Slack for chat, Notion for docs, Workday for HR, Salesforce for CRM, Zendesk for support). The reasoning is explicit: payments IS Stripe's product — anything that touches the payment processing path is built in-house with extreme care. Everything else is commodity, and engineering time spent on it is engineering time NOT spent on payments. Result: a $95B+ company with engineering largely focused on the differentiating capability.

Pro Tips

  • 01

    Calculate the 'real' build cost: take engineering's estimate and multiply by 2.5x. Software estimates are wrong by ~150% on average for greenfield builds. The honest comparison uses the realistic number.

  • 02

    The 'we'll customize the open-source version' option is almost always the worst of both worlds. You inherit the maintenance burden of build with the constraints of buy. Pure build or pure buy — avoid the middle path unless you have a dedicated open-source contributing team.

  • 03

    When you buy, negotiate for exit — data export rights, API access, and contract terms that don't lock you in. The biggest hidden cost of 'buy' is the cost of leaving 5 years later when the vendor raises prices 40%.

Myth vs Reality

Myth

Building is always cheaper because we don't pay license fees

Reality

License fees are typically 10-20% of total cost. Engineering time, ongoing maintenance, security patching, scalability work, and the cost of NOT working on differentiating features dominate the math. Build is almost always more expensive over 5 years for commodity capability.

Myth

Buying means we lose control

Reality

You lose control over the implementation, not the outcome. Most enterprise SaaS lets you configure deeply enough to express your business logic. The real loss-of-control risk is being locked into a vendor who can change pricing or stop investing in features — mitigated by negotiating exit terms upfront.

Try it

Run the numbers.

Pressure-test the concept against your own knowledge — answer the challenge or try the live scenario.

🧪

Knowledge Check

Your B2B SaaS company needs identity / authentication. Two options: build with open-source libraries (engineering estimates 4 months, $400K) or buy Auth0 ($60K/year for your scale). Which is the right call?

Industry benchmarks

Is your number good?

Calibrate against real-world tiers. Use these ranges as targets — not absolutes.

Build vs Buy Default by Capability Type

Default recommendations for B2B SaaS and enterprise software companies

Always Buy (commodity)

HR, Payroll, Email, Helpdesk, Auth, CRM

Usually Buy

Analytics, BI, Marketing automation, Support

Depends

Data platform, Workflow engine, Integration

Usually Build

Customer-facing UX, Pricing logic, Recommendation

Always Build

Core product algorithm, Proprietary IP

Source: Gartner Build vs Buy Decision Framework / Stripe Engineering Blog

Real-world cases

Companies that lived this.

Verified narratives with the numbers that prove (or break) the concept.

💳

Stripe

2010-Present

success

Stripe explicitly applies a build-only-what-differentiates philosophy. Payment processing logic, fraud models, and developer APIs are built in-house with extreme rigor. Everything else — internal communication (Slack), documentation (Notion), HR (Workday), CRM (Salesforce), support (Zendesk), data warehouse (Snowflake) — is bought. Engineering org of ~3,500 is concentrated almost entirely on the payments product. The result: industry-leading API quality, fraud detection, and developer experience — because that's where the focus went.

Engineering Headcount on Core

~80%

Engineering on Internal Tools

~5% (vs 20-30% typical)

Company Valuation

$95B+

Build/Buy Discipline

Explicit org policy

Concentrate engineering on the capability customers actually pay for. Buying everything else is not a cost — it's a strategic choice to multiply the impact of your engineering team.

Source ↗

Decision scenario

The Order Management Build vs Buy

You're CTO of a $400M GMV ecommerce retailer. Your current order management system (OMS) is a 12-year-old in-house build that breaks weekly. Engineering proposes a 14-month rebuild for $4.5M. Procurement found Manhattan Active Omni at $1.2M/year subscription + $1.8M implementation. The OMS handles inventory, fulfillment, returns, and exposes APIs to the storefront.

Current OMS Age

12 years

Engineering Estimate (Build)

$4.5M / 14 months

Buy Option (Manhattan)

$1.2M/yr + $1.8M impl

Current Outage Cost

$80K/incident, ~weekly

Engineering OMS Headcount

11 engineers

01

Decision 1

First decision: Is OMS differentiating? Argument FOR build: 'Our fulfillment workflow is unique to our business.' Argument AGAINST: 'Manhattan, Oracle, IBM Sterling all serve $1B+ retailers without those retailers feeling constrained.' What's your call?

Build — our workflow is unique and a vendor will constrain usReveal
Reality at month 18 (4 months over): engineering shipped 70% of features. The 'unique workflow' turned out to be 90% standard with one custom rule that Manhattan would have configured. Total spend: $7.2M (60% over estimate, industry standard). 11 engineers tied up for 18 months — opportunity cost of features not shipped on the storefront ≈ $12M in lost revenue. Outages stopped, but at 3-4x the buy cost.
Build TCO (3yr): $4.5M → $9.6MEngineering Time Lost: 11 engineers × 18 monthsStorefront Features Delayed: ~24 features deferred
Buy Manhattan — OMS is commodity at our scale; redirect 8 of the 11 engineers to storefront and personalization (the actual differentiator)Reveal
Manhattan implementation took 11 months (close to estimate). Total 3-year TCO: $1.8M + 3×$1.2M = $5.4M (vs $9.6M for realistic build). 8 engineers redirected to storefront shipped a personalization engine that lifted conversion 14%, generating ~$56M in incremental annual GMV. The 'unique workflow' was handled by Manhattan's configuration in 2 weeks. Outages dropped to 1/quarter.
3yr TCO: $9.6M → $5.4MEngineers Freed: 8 to storefrontIncremental GMV: +$56M from personalization

Related concepts

Keep connecting.

The concepts that orbit this one — each one sharpens the others.

Beyond the concept

Turn Build vs Buy Framework into a live operating decision.

Use this concept as the framing layer, then move into a diagnostic if it maps directly to a current bottleneck.

Typical response time: 24h · No retainer required

Turn Build vs Buy Framework into a live operating decision.

Use Build vs Buy Framework as the framing layer, then move into diagnostics or advisory if this maps directly to a current business bottleneck.