Change ManagementAdvanced7 min read

Capability Build Strategy

Capability build strategy is the deliberate choice between BUILDING new capabilities internally (training existing employees), BUYING them through hiring, BORROWING them through contractors and partners, or BLOCKING them through automation and tooling that removes the need. Most transformations default to one mode (usually 'training') without considering the others, which is why they consistently underdeliver. The right capability strategy mixes all four for the same target capability: e.g., for a data-engineering capability, you might HIRE 4 senior engineers, BUILD 30 existing employees through a 6-month bootcamp, BORROW expert contractors for the first 12 months while internal capability ramps, and BLOCK certain low-value work via low-code tools. The strategy choice depends on capability scarcity, urgency, switching cost, and the strategic centrality of the capability — and gets revisited annually as the capability matures.

Also known asCapability Build vs BuyWorkforce Capability StrategySkill Build Strategy

Challenge a friend Browse library

The Trap

The trap is defaulting to 'we'll train our way there' for every capability. Training is the slowest mode (6-18 months for most non-trivial capabilities), has variable success rates (50-70%), and produces 'taught' but not always 'expert' capability. For genuinely scarce, urgent, or strategically central capabilities, build-only strategies are catastrophically slow. The other trap is the opposite — pure 'buy' strategies that hire externally without building any internal capability. Hire-only strategies create dependency on a small group of senior new hires, who become single points of failure and often leave within 18-24 months when their original mandate is complete. The third trap is treating capability strategy as a one-time decision; in fact, the right mix changes as the capability matures — early stage favors buy/borrow, mature stage favors build/block.

What to Do

Run a structured capability strategy by mapping each target capability against four dimensions: (1) Strategic centrality — is this a core competitive differentiator (build long-term), enabling capability (buy/borrow), or commodity (block via automation/outsource)? (2) Urgency — needed in 3 months (buy/borrow), 12 months (build aggressive), or 36 months (build steady)? (3) Scarcity in market — abundant talent (build cheaper than buy) or genuinely scarce (must buy/borrow at premium)? (4) Switching cost — high (build for retention), low (buy is fine). Plot the capability on these dimensions and choose the mix. Revisit annually. The output is a B/B/B/B mix per capability with explicit % allocation: e.g., 'AI engineering: 40% build, 35% buy, 20% borrow, 5% block (via tooling).' Track build-vs-buy ratios at the portfolio level to ensure no single capability over-relies on one mode.

Formula

Capability Strategy Diversity Index = 1 - Σ (mode_share²) for each of the 4 modes (build/buy/borrow/block). Range 0-0.75; higher = more diversified mix. Single-mode strategies score 0; balanced quarter splits score ~0.75.

In Practice

Amazon's approach to building its AWS engineering capability (early 2000s through 2010s) is a textbook B/B/B/B mix: BUY for the most senior cloud-platform leaders (hiring AWS founder roles externally and from inside Amazon retail), BUILD for the bulk of engineers through a structured rotation program where retail engineers spent 6-18 months in AWS to learn cloud fundamentals, BORROW expert consultants for early customer implementations while internal expertise ramped, BLOCK the need for additional engineers in some workflow areas through aggressive automation and self-serve tooling. The mix changed over time — heavy buy/borrow early, shifting to heavy build/block as the capability matured. Amazon's ability to scale AWS from $0 to $100B+ revenue in ~17 years required this disciplined capability mix; a build-only strategy would have been too slow, a buy-only strategy would have been too expensive and culturally fragile. (Sources: 'The Everything Store,' Brad Stone, 2013; 'Working Backwards,' Bryar/Carr, 2021; AWS public origin stories from Andy Jassy interviews.) A parallel case: McKinsey's Organizational Health Index (OHI) research consistently identifies 'capability building' as one of the four management practices most strongly correlated with organizational performance — and the highest-scoring organizations on capability building use a mix of build/buy/borrow/block, not a single mode.

Pro Tips

01
For genuinely scarce capabilities (e.g., specialized AI, novel regulatory expertise), buy-or-borrow first to seed the team, then build around the seed. Pure build strategies for scarce capabilities take 24-36 months and have 50%+ failure rates because there's no one internally who can teach the new skill at the depth required.
02
Track build-vs-buy ratios at the portfolio level. If 80%+ of your capability investment is going to one mode (usually build), you're under-using the other modes. The portfolio mix should look closer to 30/30/25/15 for build/buy/borrow/block on a healthy capability portfolio at scale.
03
Borrowing (contractors, consultants, partners) is undervalued. It's the only mode that can deliver expert capability in 4-8 weeks. Use it as the bridging mechanism while build matures — this is where most transformations drop the ball, treating contractors as 'temporary' rather than as deliberate capability strategy.

Myth vs Reality

Myth

“Building internal capability is always better than hiring or contracting”

Reality

Building is slower (6-18 months), riskier (50-70% success rate at expert level), and only economically optimal for capabilities the org will use long-term and at scale. For short-term needs, scarce skills, or commodity capabilities, building is the WRONG mode. The 'always build' bias is one of the most expensive defaults in HR strategy.

Myth

“Buying capability via senior hires creates instant capability”

Reality

External hires take 6-12 months to become productive in a new organizational context (Watkins, 'The First 90 Days'); for senior leadership roles, productive impact often takes 12-18 months. The 'just hire someone' instinct understates the assimilation cost. Buy is fast relative to build, but not instant — and assimilation must be planned.

Try it

Run the numbers.

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

🧪

Knowledge Check

You need to build AI engineering capability across a 4,000-person enterprise. Talent is genuinely scarce in the market, the capability is strategically central, and you need meaningful capability within 12 months. What's the right strategic mix?

Industry benchmarks

Is your number good?

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

Healthy B/B/B/B Capability Mix at Portfolio Level

Portfolio-level capability investment in mid-large enterprises pursuing major capability shifts

Build (existing employees)

30-40%

Buy (external hires)

25-35%

Borrow (contractors/partners)

20-30%

Block (tooling/automation)

10-20%

Single-mode dominance (>60% one mode)

Warning Signal

Source: Synthesized from McKinsey Organizational Health Index research and Bain workforce strategy frameworks

Real-world cases

Companies that lived this.

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

📦

Amazon (AWS Capability Build)

2003-2015

success

AWS was built through a deliberate B/B/B/B mix that evolved as the capability matured. Early years (2003-2008): heavy BUY of cloud platform leaders externally, heavy BORROW of consulting expertise for early customer implementations, lighter BUILD as internal expertise was being created. Middle years (2008-2012): shifted toward BUILD as Amazon internalized cloud expertise via the famous internal-rotation programs (retail engineers rotating into AWS), reducing BUY share, leaning on BORROW only for novel capability gaps. Mature years (2012-2015+): heavy BUILD as internal capability could now teach itself, BLOCK rising as self-serve tooling reduced the engineer headcount needed per customer. The mix evolved deliberately — Amazon's leadership tracked the capability stage and adjusted the mode allocation accordingly. AWS revenue grew from ~$0 in 2003 to $7.9B in 2015 to $80B+ by 2023.

Capability Strategy Mix

Evolving B/B/B/B (heavy buy/borrow early, heavy build/block mature)

Time to $1B Revenue

~7 years

Time to $80B Revenue

~20 years

Internal Mobility Mechanism

Retail-to-AWS rotation programs

Capability strategy mix is not static — the optimal B/B/B/B allocation changes as the capability matures. Early stage favors buy/borrow (need expertise fast); mature stage favors build/block (have expertise to teach, can automate). Companies that lock in one mix and don't evolve it underperform.

Source ↗

📊

McKinsey OHI Research on Capability Building

Ongoing (2003-present)

success

McKinsey's Organizational Health Index (OHI), built on data from over 2,000 organizations and 5 million respondents over 20+ years, identifies 'capability building' as one of the four management practices most strongly correlated with sustained organizational performance. The OHI research goes further: organizations in the top quartile of capability building outperform bottom-quartile peers on EBITDA growth by 2-3x. Critically, the OHI shows that high-performing organizations on capability use mixed strategies (build/buy/borrow/block), not single-mode strategies. The single-mode 'we just train people' or 'we just hire' patterns are statistically associated with bottom-quartile performance.

OHI Sample Size

2,000+ organizations

Top vs Bottom Quartile EBITDA Growth Multiple

2-3x

Capability Building Rank Among Practices

Top 4 (of all measured practices)

Capability building is one of the highest-ROI management practices in the OHI dataset, but the high-ROI version is mixed-mode capability building, not single-mode. Organizations that train without hiring, or hire without training, or rely entirely on contractors, all show worse performance than organizations that mix the four modes deliberately.

Source ↗

Decision scenario

Allocating the Capability Build Budget for an Enterprise AI Initiative

You're the CHRO of a 10,000-person financial services company. The CEO has authorized a $25M capability build budget over 24 months for enterprise AI capability across data science, ML engineering, and AI product management. Current internal capability is minimal (~15 ML engineers). External AI talent is genuinely scarce — every bank, fintech, and consulting firm is hiring. You have to allocate the $25M across build/buy/borrow/block.

Capability Budget

$25M over 24 months

Current Internal Capability

~15 ML engineers

Target Capability

200+ AI/ML practitioners across 3 functions

Talent Market

Highly scarce (top quartile of historical scarcity)

Strategic Centrality

CEO-declared top-3 priority

Decision 1

First decision: how do you allocate the $25M across the four modes?

BUILD-heavy allocation: $18M (72%) into a comprehensive 12-month internal AI bootcamp covering 250 employees, $4M (16%) into hiring 8 senior leaders, $2M (8%) into contractors, $1M (4%) into AI toolingReveal

By month 12: bootcamp completion is 230/250, but only ~80 graduates are operating at meaningfully expert level (32% expert rate is typical for from-scratch internal bootcamps in scarce capability areas). The 8 senior hires are stretched across too many mentoring responsibilities. By month 24, internal capability is real but lagging competitors who used heavier buy/borrow. Total expert-level practitioners: ~110 (vs target 200). Strategy diversity index: 0.46 (mode concentration). The CEO is disappointed in the visible pace.

Expert Practitioners at 24mo: 0 → 110 (target: 200)Time-to-Impact: Slow — first material output at month 14+Strategy Diversity Index: 0.46 (mode concentrated)

Mixed B/B/B/B allocation: BUY $9M (36%) for ~25 senior AI leaders/engineers at premium comp, BORROW $6M (24%) for ~20 contractor seats over the first 12-15 months while internal capability ramps, BUILD $7M (28%) for selective 9-month program covering ~120 strong internal candidates (paired with buys/borrows for hands-on mentorship), BLOCK $3M (12%) for AI tooling and platform automation that reduces total practitioner headcount neededReveal

Within 6 months, the buy + borrow combination is producing real AI/ML output — model development, infrastructure setup, early business value. The build program benefits enormously from the mentorship of senior buy/borrow practitioners; the structured pairing produces ~70% expert rate vs the 32% rate of standalone bootcamps. Within 18 months: total expert-level capability is ~210 (25 from buy + 65 from borrow-converted-to-employees + 95 from build + tooling reduces total need). Strategy diversity index: 0.71. Total spend $25M; capability is on target AND visibly impactful from month 6 onward, which sustains CEO and board confidence in the longer build component.

Expert Practitioners at 18mo: 0 → 210 (vs 200 target)Time-to-Impact: Fast — material output by month 6Strategy Diversity Index: 0.71 (well-diversified)Build Program Expert Rate: 32% standalone → 70% with mentorship

Decision 2

Six months in, the contractor (BORROW) component is delivering material output. Your VP Talent suggests cutting the contractor budget by 50% to save money — 'we have enough internal capability now.' Your VP Engineering says contractors are still essential for the next 6 months. What do you do?

Cut contractor budget by 50% to demonstrate cost discipline; reallocate the savings to additional internal trainingReveal

The contractor reduction creates immediate capacity shortages on 4 of the 7 active AI initiatives. Three projects slip by 3-4 months because internal capability is not yet senior enough to operate independently on advanced model architecture. The build program loses the mentorship momentum that was producing the 70% expert rate. By month 12, the early-momentum advantage erodes; by month 18, you're behind plan. Premature borrow-shedding is one of the most common capability strategy errors — the borrow component is what BRIDGES the time gap until build matures.

Active Project Slip: 3 of 7 projects delayed 3-4 monthsBuild Program Expert Rate: 70% → 55% (mentorship loss)Cost Saved: $1.5M (vs $4-5M project value lost)

Maintain contractor budget through month 15, then begin a planned 6-month wind-down as senior internal capability reaches independent-operation threshold; redirect any savings to additional BUY for senior-level talent that's hard to develop internallyReveal

The disciplined wind-down preserves project momentum and the build program's mentorship dynamic. By month 15-18, internal capability is genuinely ready to operate independently on most workstreams. The contractor wind-down releases ~$3M, redirected to hire 6 additional senior data scientists at premium comp — backfilling the senior-level strategic role contractors had been playing. By month 24, internal capability is at full target level and the contractor dependency has been planfully removed. This is the textbook borrow-as-bridge pattern.

Project Slip: 0 (all timelines preserved)Senior Hires Added: +6 (from redirected contractor savings)Final Internal Capability: On target, sustainable

Related concepts