In this case study, we analyzed five comparable teams from three agencies working on similar projects for four clients. The performance gaps were striking.
Note: This data reflects the performance of the team based in that country for this case study, not the country's overall performance.
What’s driving these differences? That’s a question for future research.
We also found that Net Promoter Score (NPS) doesn’t seem to capture the full picture of team performance.
NPS reflects client expectations: teams that promise more than they can deliver may see lower scores, while those who underpromise but exceed expectations often receive higher ratings.
While useful, NPS alone may not reveal whether a team is truly delivering on value for the investment.
For client organizations, this raises an interesting question: are you getting the value you’re paying for, or are your expectations simply too low? NPS alone can’t answer this, nor can intuition.
Our data isn’t yet sufficient for broad conclusions on performance versus price for outsourced teams, but basic economics suggests that, in general, higher prices should indicate higher performance.
But where’s the limit? Is there a “sweet spot” that maximizes value? We’re digging deeper to find out.
Methodology:
-Analyzed 5 software teams from 3 outsourcing agencies, working on projects for 4 clients.
-Team sizes ranged from 10 to 20 members, handling both back-end and front-end components.
-Selected client companies and project scopes were matched for size, scope, and maturity.
-Data on NPS, team tenure, and hourly costs were provided by clients (not independently verified).
-Hourly costs were rounded & converted to USD using the average exchange rate from Oct 2024.
- Software engineering output was measured using an algorithmic model developed from research at Stanford University. Output is measured by quantifying the impact of each commit using a language-agnostic model on tokenized source code extracted from Git.
-Our dataset spans >50,000 engineers and >2 billion lines of code from millions of commits.
About Our Mission:
Our goal is to help engineering teams make decisions based on transparent and objective data, not intuition and politics.
Organizations use our data to make decisions about work methods, team setup, outsourcing, tool use, etc.
