When a first external engineering pod works well, the obvious next step is to add a second one. The assumption is that output doubles. In practice, a second team adds capacity and a coordination problem at the same time, and if the work is split badly, the coordination cost absorbs most of the gain. The decision to scale should therefore rest on three questions: is the first pod genuinely at capacity, is there a separable area of work, and is the client side ready to steer two teams? If all three answers are yes, the split should follow domain boundaries rather than technical layers, and coordination should rely on a small set of shared standards rather than on more meetings. This guide sets out the decision criteria, the split options and a step sequence.
The numbers: demand for capacity is not going away
The case for a second pod usually rests on sustained demand, and the labour market data supports that assumption. Bitkom counted around 79,000 unfilled IT positions in Germany in 2026, and 76% of companies in that survey expect the shortage to worsen. Its 2025 study put the average time to fill an IT vacancy at 7.7 months. The broader picture is tighter still: the IW Köln projected in 2026 that Germany could be short of about 723,000 skilled workers by 2029, 47% more than in 2024, and of 4.3 million workers overall by 2036 as the baby boom cohorts retire. Supply on the delivery side keeps growing. NASSCOM reported that India's technology industry added about 135,000 net jobs in FY2026, and GitHub's Octoverse 2025 counted about 21.9 million developers in India. Capacity is available; the constraint is whether your organisation can steer two teams without losing the gains in coordination.
Signals that a second pod makes sense
- Sustained demand above capacity. The backlog of accepted, well-specified work has grown over several sprints, not just in a single peak.
- A stable first pod. The first team has consistent sprint acceptance, stable delivery metrics and documentation that someone new could follow. Scaling an unstable team multiplies its problems.
- A separable area. There is a product area, service or domain that can be developed with limited day-to-day dependency on the first pod's work.
- Client-side steering capacity. Someone on your side can own priorities and acceptance for the second area. Two pods reporting to one overloaded product owner will both slow down.
Signals that it does not
- The first pod is slow because it waits for decisions, reviews or acceptance on the client side. A second pod will wait as well.
- The backlog is large but poorly specified. The bottleneck is product definition, not engineering capacity.
- The only work available is tightly coupled to what the first pod is doing. Two teams in one codebase area tend to block each other.
- Quality problems in the first pod are unresolved.
If the first set of signals is clear and the second absent, proceed. If not, fix the bottleneck first. It is usually cheaper.
How to split the work
| Split model | Description | Assessment |
|---|---|---|
| By domain or service | Each pod owns a bounded area end to end, from interface to data store | Usually the best option: clear ownership, fewer handoffs, separate acceptance |
| By product stream | Each pod owns a customer-facing capability, possibly touching shared components | Workable if shared components have a clear owner and interface contracts |
| By technical layer | One pod does frontend, the other backend | Generally weak: every feature needs both pods, handoffs dominate |
| Shared pool | Both pods pull from one backlog as one team | Weak for external teams: ownership blurs, and the client tends to allocate individuals |
Domain-based splits fit the way most modern systems are structured, and they map cleanly to separate sprint Leistungsbeschreibungen and acceptance processes. That last point matters contractually: each pod then delivers a defined result in its own area, which is easier to specify, accept and attribute.
The shared-pool model deserves a specific warning. When two external teams merge into one backlog and the client decides each morning who works on what, the engagement starts to look like the provision of individual workers under the client's direction. In terms of §1 Abs. 1 Satz 3 AÜG, that is the pattern a Werkvertrag or Dienstvertrag structure should avoid.
Coordination: standards over meetings
Two pods need some coordination. The goal is to keep it small and structural.
- Shared engineering standards: one Definition of Done, one set of quality gates, one coding standard, maintained in a repository both teams contribute to.
- Interface contracts: APIs and events between the two areas are specified, versioned and tested with contract tests, so that changes on one side do not break the other silently.
- A shared ADR log: architecture decisions that affect both areas are recorded in one place and reviewed by both.
- A short architecture forum: one session every sprint or every two sprints with the senior engineers of both pods and your architecture owner, focused on cross-cutting decisions.
- Cross-pod review for shared code: changes to shared libraries require review from the other pod.
Avoid creating a coordination role that merely relays information. If two pods need a dedicated person to pass messages between them, the split is probably wrong.
A step sequence for adding the second pod
- Define the area. Write a one-page description of the second pod's domain: responsibilities, interfaces to the first pod's area, known dependencies.
- Prepare the interfaces. Before the second pod starts, document and, if necessary, stabilise the interfaces between the two areas.
- Name the client-side owner. A product owner and an acceptance authority for the new area.
- Extend the framework, add a sprint document. The second pod receives its own sprint Leistungsbeschreibung and its own acceptance process under the existing framework agreement.
- Seed knowledge deliberately. The first pod's documentation, system map and ADRs are the onboarding material. A short handover session from the first pod's senior engineer to the second pod helps, as long as the first pod's own delivery is planned around it.
- Run a 30-day onboarding. Follow the same week-by-week structure as for the first pod, including a 30-day review.
- Review after one quarter. Compare combined delivery metrics before and after. If throughput has not increased materially, examine coordination costs and the split before scaling further.
An illustrative cost view
Illustrative calculation: if one pod is priced around the cost of one German engineering hire per month, a second pod roughly doubles the external cost. Whether that pays off depends on how much additional accepted output it produces after coordination overhead, and on how much additional client-side time it consumes for steering and acceptance. Estimate both before deciding, and measure both after one quarter. If the second pod mainly converts waiting time into more waiting time, the money is better spent on the client-side bottleneck.
Summary
Scale when demand is sustained, the first pod is stable, a separable area exists and your side can steer it. Split by domain, coordinate through standards and contracts, and keep each pod's scope and acceptance separate. Give each pod its own lead and its own sprint documents, so that ownership stays clear on both sides.
The remarks on the AÜG and contract structures are general information, not legal advice.
A second pod pays off only when ownership is as clear as it was with the first. At niogod™, a second pod is set up with its own Projektleiter, Senior Consultant and Pilot Engineer and its own sprint documents, so each team carries separate scope and acceptance. If a second team is on your roadmap for next year, it is a conversation worth having before the backlog forces it.
Talk to niogod™
Planning engineering capacity in DACH? Book a 30-minute consultation with the niogod™ team.
Book consultation