There is a number that keeps coming up when we talk to engineering leaders at DER software companies: 40 percent. That is roughly what teams estimate they spend on integration work — building and maintaining connections to hardware vendor APIs — rather than on the features that actually differentiate their product.
Forty percent is a conservative estimate. We have talked to teams where the real number is closer to 60 percent during active vendor expansion phases. And unlike most engineering costs, integration sprawl tends to get worse over time rather than better. Every new hardware vendor you support adds to the maintenance burden, compounding across the whole integration surface.
This piece puts numbers to the cost so you can make an informed decision about how to invest your engineering capacity.
The initial integration cost
Building a production-quality integration with a DER hardware vendor API takes longer than most teams estimate. A realistic breakdown:
| Activity | Estimated Time |
|---|---|
| API documentation review and protocol research | 1-2 weeks |
| Authentication and credential management implementation | 3-5 days |
| Endpoint mapping and data model normalization | 1-2 weeks |
| Error handling and retry logic | 3-5 days |
| Rate limit management | 2-3 days |
| Test coverage | 1 week |
| Vendor sandbox testing and debugging | 1-2 weeks |
Total: 5 to 9 weeks for a first integration with an unfamiliar vendor API. Teams with prior experience in the specific protocol (OCPP, OpenADR, IEEE 2030.5) can compress the protocol research phase, but the implementation and testing phases do not compress much.
At a fully loaded engineering cost of $150,000 to $200,000 per year for a senior engineer, five to nine weeks of their time runs $14,000 to $35,000 per vendor integration. For a product that needs to support 8 hardware vendors, that is $112,000 to $280,000 in one-time integration build costs before you write a line of product code.
The ongoing maintenance burden
The initial cost is a one-time expense. Maintenance is permanent. Sources of integration maintenance work:
- Vendor API updates — DER vendors update their APIs. Sometimes with deprecation notices, sometimes without. A field that returns watts in one API version returns kilowatts in the next. A new required authentication header appears. Your integration breaks in production.
- Rate limit changes — Vendors adjust rate limits without much ceremony. If your application was designed around a 1,000 requests per day limit and the vendor drops it to 300, you need to redesign your request scheduling logic.
- OAuth token rotation — Token expiry and refresh handling for vendor OAuth APIs requires active monitoring. A token that silently expires takes down your integration until someone notices.
- Edge case bug reports — Real-world device deployments surface edge cases that test environments do not. The charger firmware version that returns null for a required field. The inverter model that reports a different timezone than expected.
We estimate maintenance at roughly one week per vendor per quarter for a mature integration — meaning a product with 8 vendor integrations spends 8 weeks per quarter, or roughly 30 percent of one engineer's annual capacity, just on keeping existing integrations working.
The opportunity cost: what you are not building
The hardest part of integration sprawl to quantify is the opportunity cost — what the team did not build because they were maintaining integrations instead.
DER software products compete on features: better analytics, smarter dispatch logic, a more intuitive customer dashboard, deeper demand response participation capabilities. Every engineering week spent on integration maintenance is a week not spent on those features. In a market where the competition is active and customer expectations are rising, falling behind on product velocity has real revenue consequences.
We have seen teams where a product manager wanted to ship a new demand response optimization feature in a quarter, but two of the available engineers were pulled into a three-week integration maintenance firefight when a major vendor changed their authentication model. The feature shipped a quarter late. Whether that caused a churn event or a lost deal is impossible to isolate, but the cost is real.
The cost of integration sprawl is not just the engineering time you can measure. It is also every product decision you made differently because your team was not free to build what customers actually needed.
The team structure implications
Integration sprawl also shapes how teams hire and organize. When integration maintenance requires deep vendor-specific knowledge — who knows the OCPP 2.0 quirks of ChargePoint specifically, who understands the SolarEdge API's pagination behavior — teams accumulate implicit expertise that becomes a single point of failure. The engineer who built the Enphase integration leaves, and their knowledge of why certain edge-case handling logic exists walks out the door with them.
An abstraction layer that centralizes vendor-specific complexity reduces this key-person dependency. Integration knowledge is documented in the adapter layer, not in the tribal knowledge of whoever built it.
What the math says
Let's put the numbers together for a mid-scale DER software product supporting 8 hardware vendors:
- Initial build: $150,000 to $250,000 (one-time)
- Annual maintenance: $80,000 to $120,000 per year (ongoing)
- Opportunity cost (conservative estimate): $100,000 to $200,000 per year in delayed feature value
Over a three-year product horizon, integration sprawl costs $400,000 to $800,000 in engineering investment and forgone product velocity. That is before accounting for the increased fragility of a product that has no abstraction layer between its business logic and eight different vendor APIs.
These numbers help explain why the teams we talk to who have invested in normalization infrastructure early tend to ship faster, with smaller engineering headcounts, than teams who built their integrations in the conventional ad-hoc way. The infrastructure investment pays back, usually within the first year.
