We calculate capacity by applying a 30% productivity discount to external agencies, modeling their throughput at 0.7 FTE. While staff augmentation contractors embedded in our teams reach 85-95% efficiency, external pods face inherent overhead from handoffs and system friction. We use these blended ratios to ensure roadmaps reflect actual output rather than optimistic headcount.
Why we apply a productivity discount to external vendors
External teams lack immediate access to tribal knowledge and undocumented architectural history. When we staff an initiative with an outside agency, we are not just buying engineering hours; we are introducing a communication tax. Communication overhead increases linearly with the number of handoff points between internal owners and vendor pods. Even with a clean API contract, the "why" behind an architectural decision often lives in Slack threads or defunct Jira tickets that external partners cannot see.
Security protocols and tooling restrictions create a secondary tax on vendor velocity. Internal staff bypass these hurdles through established permissions, while agencies often wait days for VPN access or sandbox environments. We model agencies at 0.7 FTE to prevent over-committing the roadmap. If we plan 30 concurrent initiatives based on raw headcount, we inevitably face a 20-30% delivery gap by mid-quarter. Modeling at 70% capacity accounts for this friction before the first sprint begins.
What is the standard ramp-up time for agency teams in R&D?
We do not expect peak modeled capacity on day one. For an R&D organization running 20+ initiatives, the intake process must account for the following ramp-up curve:
- Month 1 (40% capacity): Teams navigate environment setup, codebase familiarity, and local development quirks. Most of this time is spent in "read-only" mode.
- Month 2 (60% capacity): The pod begins contributing to non-critical path features. They are functional but require heavy code review and frequent architectural guidance.
- Month 3+ (70% capacity): Steady state for agencies. If the resource is an embedded individual (staff augmentation), we move this up to 90%.
We bake this curve into our quarterly intake process. If a project requires ten engineers to hit a deadline in three months, and we use an external agency, we cannot simply hire ten people. We must over-provision or extend the timeline to account for the 12-week journey to "effective" headcount.
How do you account for knowledge transfer time in capacity models?
Knowledge transfer is a work item in the backlog, not a background task. We deduct 10-15% of internal lead engineer capacity when they are assigned to oversee a vendor pod. This is a non-negotiable "management tax." If a Lead is managing two external pods, their own individual contribution capacity drops to 70% or lower.
To mitigate future handoff friction, we include documentation requirements in the vendor's Definition of Done. We also track "shadow time" where FTEs must review 100% of external code contributions. This ensures that when the vendor contract ends, the intellectual property and the logic behind it remain within the building.
Should contractors be included in core team velocity calculations?
We differentiate between embedded individuals and external pods. Embedded contractors are integrated into team velocity after their first two sprints. Because they participate in the same standups and use the same CI/CD pipelines as FTEs, their output eventually mirrors the rest of the team.
External pods, however, are measured on a separate burn-up chart. Mixing their velocity with internal teams skews performance benchmarks and makes it difficult to identify where systemic bottlenecks exist. We use a "blended FTE equivalent" for portfolio-level planning to normalize output across diverse team structures. While velocity is measured at the team level, capacity planning happens at the resource-type level. This distinction allows us to see that an initiative staffed by 10 agency engineers has the same "carrying capacity" as a team of 7 internal FTEs.
What is the maximum ratio of contractors to FTEs?
We maintain a maximum 3:1 contractor-to-FTE ratio within any single functional area. Exceeding this ratio leads to a dilution of intent. When the majority of a team has an execution-only mindset, the long-term product vision becomes blurred.
High-ratio teams show a measurable increase in technical debt. External incentives often prioritize speed and "meeting the SOW" over long-term maintainability. To counter this, core architectural decisions and infrastructure ownership must remain with 100% internal staff. This prevents vendor lock-in and ensures that the foundation of our R&D remains under our direct control.
| Resource Type | Modeled Capacity | Utilization Target | Integration Level | | :--- | :--- | :--- | :--- | | Internal FTE | 1.0 (Baseline) | 80% | Native | | Embedded Contractor | 0.9 | 95% | Deep (Staff Aug) | | External Agency Pod | 0.7 | 95% | Low (SOW-based) | | New Hire (Month 1) | 0.2 | N/A | Onboarding |
How does the cost-per-point vary between FTEs and staff augmentation?
FTEs carry a 1.4x to 1.8x burden for benefits, equity, and office space. This makes their hourly rate comparable to premium agencies. However, the cost-per-point is often lower for agencies in the short term because they can scale rapidly without the 90-day recruitment cycle.
We value agencies for operational flexibility. The ability to spin down a 10-person pod in 30 days is a strategic hedge against shifting priorities or budget cuts. Long-term cost efficiency always favors FTEs once an initiative moves from the "build" phase to the "run" phase. At that point, the 30% productivity discount becomes an expensive permanent tax that outweighs the benefit of flexibility.
The Blended Capacity Playbook for R&D Heads
To implement this model, we follow a specific sequence during the quarterly planning cycle:
- Audit current vendor depth: Classify every external head as either "Embedded" (0.9) or "External Pod" (0.7).
- Set utilization targets: We plan internal R&D teams for 80% utilization to allow for innovation and unplanned maintenance. Contractors are planned at 95% against their specific SOW.
- Update RMO templates: Use "Effective FTEs" instead of raw headcount in all resource management office templates.
- Review vendor velocity: If an agency consistently exceeds its 0.7 baseline over two quarters, we adjust their modeled capacity upward to 0.8.
- Apply the internal tax: Mandate a 10% reduction in the capacity of any internal lead engineer managing an external workstream.
Honest Tradeoff: The "Us vs. Them" Risk
Treating all team members as 100% equivalent simplifies administrative overhead and fosters a unified culture. Our discount model can inadvertently signal that external partners are "second-class" contributors. This can hurt morale and reduce the very engagement we need to make the partnership successful. If we focus too much on individual discounts, we risk ignoring systemic issues—like poor documentation or slow CI/CD pipelines—that cause productivity drag for everyone, not just contractors.
In one breath
We model agency capacity at 0.7 FTE to account for the communication and system friction inherent in external partnerships. By applying a 10% management tax to internal leads and maintaining a 3:1 staffing ratio, we protect the roadmap from optimistic over-commitment. This blended approach prioritizes operational flexibility while ensuring that core architectural intent remains internal.