Procurement teams compare engagement models on rate cards. Engineering teams know the rate card hides most of the cost. Finance teams know both groups are partly right and want a model that surfaces the gap. This piece is that model.
Total Cost of Delivery (TCD) is a seven-component framework that produces an apples-to-apples comparison across the six engagement models enterprises actually use: in-house hiring, traditional outsourcing, staff augmentation, freelance marketplaces, captive offshore (ODC), and Virtual Delivery Center. Each component is finger-able with numbers your existing systems already produce. The output is a TCD-per-shipped-feature ratio that compares cleanly across all six.
Why a TCD framework matters
Most procurement comparisons fail at the same point: they treat headline rates as comparable when the cost structures are fundamentally different. An in-house engineer's $150K loaded salary isn't comparable to a contractor's $80/hour rate isn't comparable to a VDC's milestone-based pricing. They're priced on different bases, denominated in different units, and absorb different costs across the engagement layer.
TCD normalizes the comparison. Every cost — visible or hidden, vendor-invoiced or internal-payroll — gets pulled into a single per-shipped-unit number. Then the six models can be compared honestly, and the model selection becomes deliberate instead of defaulting to whichever was used last time.
The seven components of TCD
1. Direct labor cost
The headline number. For internal staff: loaded salary × engagement duration × allocation %. For contractors: hourly rate × hours billed. For VDCs and outsourcing: milestone-based or fixed-bid contractual cost.
This is where rate-card comparison stops. TCD continues.
2. Management overhead
The internal time required to manage the engagement. For staff augmentation, this typically runs 20–40% of one engineering manager per 4-contractor team. For in-house hiring, it's the standard 1:6 to 1:8 manager-to-engineer ratio. For a VDC, it's near zero — the platform's embedded delivery manager owns it.
Calculate as: burdened EM time × percent allocation × engagement duration.
3. Ramp tax
Productivity loss during onboarding. Every model has it; the magnitude varies widely.
- In-house hire: 8–12 weeks at 50% productivity, plus the recruiting cycle that preceded.
- Contractor: 4–8 weeks at 35% productivity.
- Outsourcing engagement: 60–90 days for the team to ramp on context and tooling.
- ODC/GCC: 90–120 days for the captive-style operation to stand up.
- VDC pod: 5–10 days because the bench is pre-vetted; ramp tax is 1–2 sprints, not 8.
Calculate as: (engagement_cost × ramp_weeks / total_weeks) × (1 - average_ramp_productivity).
4. Bench tax
Cost of paid-but-unproductive time when the next task isn't ready. Highest for staff augmentation (you pay regardless). Lower for outsourcing (built into the SOW). Near-zero for VDCs (platform absorbs). Zero for in-house (you redirect the engineer to other work).
Calculate as: estimated idle hours × hourly rate.
5. Coordination overhead
Time spent integrating across vendors, contracts, tools, and reporting. Highest when running multi-vendor staff augmentation. Lower for single-vendor outsourcing. Near-zero for VDC (platform-default tooling and reporting).
Calculate as: DM time × hours per week × duration × number of vendor relationships.
6. Knowledge attrition cost
Cost of lost institutional knowledge when contractors rotate or staff turn over. Highest for short-term contractors. Moderate for in-house hires (every 2–3 years). Lower for stable VDC pods (rotation is platform-managed, knowledge is pod-retained, not individual-retained).
Calculate as: expected rotations × (lost senior time + next-contractor ramp tax).
7. Termination / pivot cost
Cost to unwind the engagement when scope, business, or vendor relationship changes.
- In-house: severance, unused PTO, knowledge handoff to remaining team.
- Outsourcing: contractual wind-down clauses (often 90 days).
- ODC: multi-quarter decommissioning (facilities, employees, equipment).
- Staff aug / Freelance: per-contractor notice, typically 30 days.
- VDC: milestone-bounded; next-milestone exit, no decommissioning.
Calculate as: termination notice × current burn rate, plus any contractual penalties.
The TCD formula
TCD = (1) Direct labor + (2) Management overhead + (3) Ramp tax + (4) Bench tax + (5) Coordination overhead + (6) Knowledge attrition + (7) Termination / pivot cost
Then normalize: TCD-per-shipped-feature = TCD / shipped features over engagement duration.
This second normalization is the one that breaks ties. Two models with similar TCD can have very different shipped-feature output, and the per-feature number is what matters for ROI conversations.
Worked example: 6-month engagement, 4-engineer team
Assume the same scope across all six models: 6 months of effort, equivalent to 4 mid-senior engineers. Compare TCD across the six models, with realistic mid-case numbers:
| Component | In-house | Outsourcing | Staff Aug | Freelance MP | ODC | VDC |
|---|---|---|---|---|---|---|
| Direct labor | $400K | $310K | $330K | $280K | $280K | $280K |
| Mgmt overhead | $50K | $30K | $100K | $130K | $40K | $5K |
| Ramp tax | $80K | $60K | $50K | $45K | $90K | $0 |
| Bench tax | $0 | $15K | $25K | $10K | $30K | $0 |
| Coordination | $10K | $25K | $50K | $80K | $30K | $5K |
| Knowledge attrition | $0 | $5K | $30K | $40K | $5K | $0K |
| Termination cost | $50K | $50K | $10K | $5K | $70K | $0K |
| TCD total | $590K | $495K | $595K | $590K | $545K | $290K |
Some immediate observations:
- The "cheap" headline rates of staff aug and freelance marketplaces produce TCDs higher than outsourcing once the hidden costs are included.
- In-house has high direct labor but low coordination/attrition costs, landing near outsourcing on TCD.
- VDC's TCD advantage is structural — it absorbs categories 2, 4, 5, and 7 at the platform layer, leaving direct labor as the dominant cost.
- The numbers in this example are mid-case. Worst-case (high turnover, multi-vendor staff aug) pushes that column to $700K+. Best-case (well-bounded outsourcing engagement, single vendor) drops to $440K.
When the framework breaks (and what to do about it)
TCD assumes you can quantify each component. Three scenarios where that gets hard:
- You don't have historical data on management overhead. Most enterprises don't track EM time per engagement. Solution: have your EMs track time for two weeks across one current engagement, extrapolate.
- You don't know the bench tax. Hard to measure if it's been hidden as busywork. Solution: estimate at 5% of contractor billing as a starting point; refine based on observed gaps in deliverable production.
- The engagement is novel — no comparable history. Solution: model with industry benchmarks (we've published mid-case numbers above; adjust for your context). Better imperfect than no model at all.
The framework is approximate by design. The point isn't precision; it's surfacing categories of cost that the rate card hides. Even rough estimates produce comparisons that beat rate-card-only by a wide margin.
Procurement-grade scorecard
To operationalize the framework in a procurement comparison, structure as:
- Define the engagement scope and duration.
- For each candidate model, fill in the seven TCD components.
- Calculate TCD total.
- Estimate shipped features over the engagement (ratio per model based on observed productivity).
- Calculate TCD-per-shipped-feature.
- Pick the model with the lowest TCD-per-shipped-feature, weighted by strategic factors (regulatory needs, internal capacity, etc.).
The scorecard is a single spreadsheet your procurement team can build in an hour. The hard part is getting the EM time and bench tax numbers — once those exist, the comparison is mechanical.
Frequently asked questions
Doesn't this framework favor VDCs?
It favors models that absorb hidden costs at the platform layer rather than pushing them onto the buyer. VDC happens to be one such model; outsourcing partially absorbs them too (the framework gives outsourcing the second-best score in the example above). It penalizes models that pretend rate-card cost equals total cost, which is staff aug and unmanaged marketplaces.
How accurate are the numbers in the worked example?
Within ±20% for a typical mid-market enterprise. The relative ordering across models is more stable than the absolute numbers. If your numbers move 10–15% in any direction, the ordering rarely flips.
Can this framework compare in-house hiring vs. external?
Yes. In-house's strength is low termination cost (per engagement), low coordination overhead, low attrition. Its weakness is high ramp tax (recruiting cycle plus 8–12 week ramp) and limited elasticity. For sustained, predictable workloads it competes well with outsourcing on TCD; for variable or burst workloads it loses to a VDC.
What if procurement won't accept "soft" cost categories?
Convert them to hard ones. Management overhead = burdened EM hours × billing rate. Ramp tax = lost output × output value. Bench tax = invoiced hours × percentage idle. Each of these is defensible to procurement once you stop calling them "soft." They're hard internal costs hiding in salary lines.
Where to start
Run the framework against one in-flight engagement. Pick the engagement where you have the best instrumentation — usually a long-running staff-aug or in-house team. Calculate the TCD components from your actual data, not estimates. The number will likely surprise you.
If you want help building the model for your specific engagements, schedule a 30-minute call. We'll plug in your numbers and produce a procurement-defensible TCD comparison across the relevant models.
For deeper dives on each component, see hidden costs of staff augmentation, VDC vs Outsourcing, and VDC vs ODC.