Biology of Business

Door 3: DECIDE 3.3

Infrastructure Decision Framework

"You need to decide whether to build your own infrastructure or keep paying someone else's fees, what type of distribution network to design, and how to scale it without building too early (Webvan) or too late (losing to competitors who invested first). How do you make infrastructure decisions that match your actual volume and growth trajectory?"

What you'll get

A build-vs-rent decision with ROI calculation, an active/passive infrastructure type selection, a Murray's Law network design (if building), and a staged investment plan that matches infrastructure spending to actual volume thresholds.

When to use this

When 3PL or platform fees exceed $5M annually and you're questioning whether to build in-house. When designing a distribution network from scratch. When scaling requires choosing between active and passive transport modes. When evaluating whether infrastructure is a competitive advantage or a commodity best rented.

The process

1

Calculate Your Infrastructure Tax

How to do this
Mycorrhizal fungi extract 10-30% of a plant's photosynthate as payment for nutrient access through the underground network. The plant pays this tax because building its own root network to reach distant nutrients would cost more than the tax. Network operators — Stripe (2.9% per transaction), AWS (cloud markup), Visa (interchange fees), FedEx (shipping premiums), Shopify (platform fees) — extract the same kind of infrastructure tax. The tax is worth paying when individual participants can't replicate the network's value. Sum every fee you pay to infrastructure providers: payment processing, cloud hosting, logistics and fulfillment, platform fees, SaaS tools. Calculate total as a percentage of revenue. This is your infrastructure tax rate. Below 5% of revenue: the tax is negligible — don't waste engineering resources building alternatives. At 5-15%: monitor and model build scenarios, but the tax is likely still worth paying for speed and flexibility. Above 15%: seriously evaluate whether owned infrastructure would cost less — you may be paying premium fees at scale where bulk economics favor ownership. The critical question isn't the tax rate alone — it's whether you could actually replicate the network's value. Visa's payment network has acceptance at 80M+ merchants. AWS has global data center infrastructure. You can't replicate these at any cost. But a regional fulfillment network? A payment processing layer for a specific vertical? These are replicable at sufficient scale.
  • Infrastructure tax inventory: every provider, annual cost, and percentage of revenue
  • Tax rate assessment: negligible (<5%), monitoring zone (5-15%), or evaluation threshold (>15%)
  • Replicability analysis: which infrastructure could you build vs. which is irreplaceable
  • Growth trajectory: are infrastructure costs growing faster or slower than revenue?
If your infrastructure tax is below 5% of revenue, stop here — keep renting and focus resources on your core business. If 5-15%, proceed to Step 2 for specific build-vs-rent analysis on the highest-cost items. If above 15%, proceed urgently — you're funding someone else's margins at scale.
2

Run the Build vs. Rent ROI Calculator

How to do this
Don't grow arteries before you have blood flow to justify them. Webvan built $1 billion in fulfillment infrastructure before order volume justified it — they went bankrupt in 2001. Amazon waited until volume made 3PL mathematically impossible before building fulfillment centers. The difference? Murray's Law: blood vessels branch with diameters optimized for the trade-off between pumping cost and infrastructure cost. Below certain flow volumes, capillaries (rented infrastructure) are metabolically cheaper than arteries (owned infrastructure). Above certain thresholds, permanent infrastructure pays for itself. For each infrastructure category identified in Step 1, run the ROI calculation. Step A: Calculate current rental cost. Annual volume × cost per unit (typically $4-8 per order for 3PL, 2.9% per transaction for payment processing, per-unit cloud costs). Step B: Estimate owned facility total cost. Upfront investment + annual maintenance. For fulfillment: facility lease + equipment ($500K-2M per year) + labor for 20-50 people ($800K-2M per year) + operating costs ($200K-500K per year). For technology: engineering team + infrastructure + compliance + operations. Step C: Calculate payback period. Payback = Upfront investment ÷ (Current annual rental cost - Projected annual owned cost). Decision rules: Build if payback is under 3 years AND volume is stable or growing. Stay with provider if payback exceeds 4 years OR volume is uncertain. In the 3-4 year zone, assess strategic value — is this infrastructure core to competitive differentiation? Even with a long payback, build if infrastructure is your moat. Even with a short payback, rent if network effects make replication impossible.
  • ROI calculation for each infrastructure category: current rental cost, estimated owned cost, payback period
  • Build-vs-rent recommendation for each category with confidence level
  • Sensitivity analysis: how payback changes if volume drops 20-30%
  • Strategic assessment: competitive advantage vs. commodity for each category
If payback is under 3 years with stable volume, build. If over 4 years or volume is uncertain, rent. If 3-4 years, let strategic value break the tie. For categories where you decide to build, proceed to Step 3. For categories where you decide to rent, revisit annually as volume grows.
3

Choose Active or Passive Infrastructure

How to do this
Plants use two fundamentally different transport systems. Xylem is passive — water is pulled upward by evaporation from leaves, requiring no metabolic energy. It's cheap but uncontrolled; the plant can't direct water to specific tissues. Phloem is active — sugar is pushed to specific locations by ATP-driven pumps. It's expensive but precisely controlled; the plant sends energy exactly where it's needed. Your infrastructure choice follows the same logic. Score your product on four characteristics (0-3 each): Product value greater than $100 per unit? Product perishes quickly (less than 30 days)? Customers pay premium for fast delivery? Delivery speed is competitive differentiation? Score 0-4: Passive infrastructure. Ocean freight, ground shipping, 7-14 day delivery. Cheap, uncontrolled, high-volume. This is xylem — adequate for low-value, non-perishable goods where speed isn't competitive. Maersk's container shipping is the ultimate passive infrastructure. Score 5-8: Hybrid. Passive long-haul with active last-mile. 3-5 day delivery. Most e-commerce operates here — products travel cheaply across long distances, then switch to expensive active delivery for the final miles. Score 9-12: Active infrastructure. Air freight, expedited delivery, 1-2 day or same-day. Expensive, precisely controlled. This is phloem — necessary for high-value, time-sensitive, or perishable goods where speed is the product. FedEx overnight delivery, Zara's fast-fashion supply chain, and grocery delivery all require active infrastructure. Validate economics: active transport cost should be less than 10% of product value. Shipping a $5 item via $15 overnight delivery is the equivalent of building phloem to transport water — metabolically wasteful.
  • Product characteristic score (0-12) with infrastructure type recommendation
  • Active/Passive/Hybrid infrastructure selection with rationale
  • Economic validation: transport cost as percentage of product value
  • Competitive assessment: does your infrastructure choice match or beat competitors?
If you score 0-4, invest in passive infrastructure and compete on price or selection, not speed. If 5-8, build hybrid — this is where most optimization opportunities live. If 9-12, invest heavily in active infrastructure and make speed your competitive weapon.
4

Design the Network with Murray's Law

How to do this
Cecil Murray proved in 1926 that blood vessel branching follows an optimal ratio: parent vessel radius cubed equals the sum of daughter vessel radii cubed (r₀³ = r₁³ + r₂³). This minimizes total cost — the sum of pumping cost (favors wide vessels) and infrastructure cost (favors narrow vessels). Distribution networks follow the same optimization. Define four facility tiers following vascular branching. Tier 1 — Trunk (like the aorta): few large regional hubs. High capacity, low count. These are your primary fulfillment centers or central warehouses. Tier 2 — Major branches (like arteries): more medium metro hubs. Moderate capacity, moderate count. These serve metro areas and handle regional sorting. Tier 3 — Minor branches (like arterioles): many small local hubs. Lower capacity, higher count. These handle last-mile staging and local inventory. Tier 4 — Twigs (like capillaries): thousands of pickup points, lockers, or micro-fulfillment locations. Minimal capacity, very high count. Calculate optimal facility count per tier. Tier 1 = Daily throughput ÷ Large facility capacity (approximately 75,000 orders per day). Tier 2 = Tier 1 count × 3. Tier 3 = Tier 2 count × 4. Tier 4 = Tier 3 count × 15. Validate with Murray's Law: trunk capacity cubed should approximately equal the sum of all branch capacities cubed. Each tier should handle 110-120% of the previous tier's volume (accounting for returns, transfers, and buffer). Walmart's distribution network closely follows this pattern: ~40 regional distribution centers (Tier 1), ~150 fulfillment centers (Tier 2), and thousands of stores serving as local pickup points (Tier 4).
  • Four-tier network architecture with facility count and capacity at each tier
  • Murray's Law validation: does branching ratio minimize total cost?
  • Geographic placement map: facility locations matched to customer density
  • Network health scorecard: delivery distance, utilization, last-mile cost share, 2-day delivery rate
This step only applies if you're building owned infrastructure at scale (Series D+, $500M+ revenue, 16,500+ orders per day). Earlier-stage companies should use 3PLs (Step 2 confirmed this) and revisit Murray's Law network design when volume thresholds are crossed.
5

Stage the Investment

How to do this
Organisms don't build their full vascular network at birth. Blood vessels grow in response to demand — tissues that need more oxygen signal for new vessel growth (angiogenesis), while underused vessels are pruned. Infrastructure investment should follow the same demand-responsive pattern. Stage 1 — Seed (use 3PLs entirely): order volume under 500 per day. Zero owned infrastructure. Rent everything. Focus all capital on product-market fit. This is embryonic circulation — minimal, adequate, and cheap. Stage 2 — Growth (selective build): 500-5,000 orders per day. Build your first owned facility only where concentration justifies it (60%+ of orders in one region). Keep 3PL for all other regions. This is early childhood growth — building the trunk while relying on external circulation for the periphery. Stage 3 — Scale (multi-tier build): 5,000-50,000 orders per day. Build Tier 1 and Tier 2 facilities. Maintain 3PL for Tier 3 and 4 in low-density regions. This is adolescent growth — the major vascular branches are in place. Stage 4 — Optimization (full network): 50,000+ orders per day. Build all four tiers. Optimize with Murray's Law. Continuous capacity adjustment. This is adult vascular maturity — the full network is built and being fine-tuned. At each stage, set the threshold that triggers the next stage. Don't advance because of ambition — advance because volume demands it. The threshold is the point where the build-vs-rent payback from Step 2 crosses below 3 years for the next tier of infrastructure.
  • Current stage assessment with evidence
  • Stage-appropriate infrastructure plan: what to build now, what to rent, what to plan for
  • Volume thresholds for stage transitions: the specific order volumes that trigger next investment
  • Transition timeline: when you expect to cross each threshold based on growth trajectory
If your infrastructure matches your stage, maintain the plan and monitor thresholds. If you're under-invested (volume has crossed thresholds but infrastructure hasn't been built), accelerate — you're losing margin to 3PL fees at scale. If you're over-invested (infrastructure built ahead of volume), freeze investment and focus on volume growth to fill existing capacity.
6

Monitor Network Health Continuously

How to do this
Vascular networks maintain health through constant monitoring — blood pressure sensors in vessel walls detect blockages, flow sensors identify underperforming branches, and angiogenesis signals trigger new vessel growth where demand increases. Build continuous monitoring around four network health metrics. Average delivery distance: target under 150 miles. If over 150, the network is under-branched — add facilities closer to customer concentrations. Facility utilization: target 75-85%. Below 60% means over-built (freeze expansion). Above 90% means bottleneck risk (expand or add tier). Last-mile cost share: target under 40% of total shipping cost. If last-mile exceeds 40%, Tier 3 and Tier 4 infrastructure is insufficient — add local hubs or pickup points. Two-day delivery rate: target above 90% of orders. If below 90%, geographic coverage gaps exist — map underserved regions and plan facility additions. Review quarterly. Compare actual metrics to targets. Identify bottlenecks (facilities above 90% utilization), dead zones (regions with long delivery times), and waste (facilities below 60% utilization). Adjust the network continuously — infrastructure is not a one-time decision but an evolving organism that grows, prunes, and adapts to changing demand patterns.
  • Network health dashboard: delivery distance, utilization, last-mile cost, 2-day delivery rate
  • Bottleneck alerts: facilities approaching capacity limits
  • Expansion candidates: regions where demand has outgrown infrastructure
  • Pruning candidates: facilities with consistently low utilization
If all four metrics are at target, maintain current network. If delivery distance or 2-day rate is off target, expand (add facilities). If utilization is below 60%, contract (close or consolidate facilities). If last-mile costs are climbing, add Tier 3-4 infrastructure closer to customers.
✓ Framework complete

See it in action: Amazon

Adapt to your context

industryVariations

[object Object][object Object][object Object][object Object]

companySizeVariations

[object Object][object Object][object Object][object Object]