How to Use Build vs Buy Decision Engine
The Build vs Buy Decision Engine evaluates each infrastructure component on development cost, maintenance burden, opportunity cost, and managed service pricing to recommend build or buy per component.
What It Does
Use the calculator with intent
The Build vs Buy Decision Engine evaluates each infrastructure component on development cost, maintenance burden, opportunity cost, and managed service pricing to recommend build or buy per component.
Technical founders deciding which parts of their stack to build custom and which to outsource to managed services.
Interpreting Results
Compare annual build cost against annual buy cost per component, not just the upfront build hours. The recurring maintenance burden, valued at your hourly rate, is what usually tips commodity infrastructure toward buy. Reserve building for the component that is your actual differentiation; outsource the rest so your scarce hours go to the product, not to undifferentiated plumbing.
Input Steps
Field by field
- 1
Set your hourly value
Enter what an hour of your time is worth, ideally your billable rate or the value of an hour spent on product and customers instead. This is what turns build hours into a real dollar cost, so a higher hourly value pushes more components toward buy.
- 2
List each component
For each stack component (auth, database, payments, search), enter the hours to build it, the ongoing maintenance hours per month, and the managed-service monthly cost. Enable only the components you actually need to decide on.
- 3
Be honest about maintenance
Maintenance hours per month is the input founders most often understate. Self-built infrastructure carries patching, monitoring, and incident time forever, while a managed service absorbs it. Enter the realistic recurring burden, not the best case.
- 4
Read per-component verdicts
Read the annual build cost (build hours plus maintenance, valued at your rate) against the annual buy cost for each component, plus the overall verdict and annual savings. A component where building costs more than buying over a year is almost always a buy, especially early when your hours are worth more on product.
- 5
Re-run as your time value changes
Re-run when your hourly value rises (it usually does as the business grows) or when a managed service changes price. The build case weakens as your time becomes more valuable, so a component worth building at launch is often worth buying a year later.
Common Scenarios
Use realistic starting points
Buy commodity infrastructure
Hourly value
100
Auth: build / maintain / buy
40h / 4h-mo / 25
Database: build / maintain / buy
20h / 2h-mo / 25
At 100 per hour, 40 build hours plus 4 maintenance hours a month dwarfs a 25-per-month managed service. Watch how clearly auth and database come out as buy when your time is priced honestly.
Low hourly value flips the math
Hourly value
25
Component build hours
40
Managed service cost
50
When your time is cheap and the managed service is expensive, building can win on pure dollars. Watch whether the build verdict holds once you add the maintenance hours, which often erase the apparent saving.
Try These Tools
Run the numbers next
AI Stack Cost Calculator
Estimate your full AI app stack cost at different user scales — hosting, DB, auth, AI API, and services.
LLM Vendor Lock-In Cost
Engineering, downtime, and payback when migrating between LLM providers.
Embeddings DB Cost
Pinecone, Postgres+pgvector, LanceDB, or Turbopuffer — cheapest for your workload.
FAQ
Questions people ask next
The short answers readers usually want after the first pass.
Sources & References
- Choose Boring Technology — Dan McKinley
- Choose Boring Technology (talk) — Dan McKinley
Related Content
Keep the topic connected
How to Use LLM Vendor Lock-In Cost
Estimate the all-in cost of switching LLM providers — engineering hours, downtime opportunity cost, and payback months at a new vendor's price. Free.
How to Use AI Stack Cost Calculator
Estimate your full AI app stack cost at different user scales including hosting, database, auth, AI API, and third-party services.
How to Use Embeddings DB Cost
Embeddings DB cost walkthrough: how to enter vector count, dimensions, and query volume to read the Pinecone vs pgvector vs LanceDB vs Turbopuffer comparison.