Google — Engineering Performance
Avg. perf / dev / mo (ETV)
+19.0%
1.35 → 1.60
Active contributors
−16.4%
128.0 → 107.0
Growth
−0.1pp
34.7% → 34.6%
Fixes
+1.6pp
11.6% → 13.2%
Google vs. 500 OSS Performance Index
Per-engineer ETV for Google plotted against the pooled 500 OSS Performance Index. Both series are 90-day trailing rolling averages scaled to a 30-day month, so the curves sit on the same scale (ETV / dev / mo) and can be compared point-for-point. Latest reading: Google is 19% below the index (1.60 vs 1.97 ETV/dev/mo). Baseline gap was 4% below.
Monthly reports
Highlights
- The *ADK Go V2* release ([893e4a40 · wolo]) marked a major overhaul of the workflow engine, introducing a robust scheduler-based runtime, Human-in-the-Loop (HITL) interactions, and dynamic node orchestration.
- Significant enhancements were made to the *Perfetto UI*, including new trace summary billboards ([8b1b64d0 · Steve Golton]), a process information strip ([e427da76 · Steve Golton]), a barebones memory overview page ([aa839152 · Steve Golton]), and a complete multi-trace merging dialog ([969ed94e · Lalit Maganti]).
- The *BigQuery DataFrames (BigFrames) UDF transpiler* was enhanced to handle control flow constructs like `if`/`else` statements ([a8cbde39 · TrevorBergeron]), greatly expanding UDF expressiveness.
- The *ADK workflow system* gained new capabilities, such as exposing `max_parallel_workers` for concurrency control ([199d9548 · Shangjie Chen]) and allowing `ToolNode` to accept JSON strings or `types.Content` objects ([4e446324 · João Westerberg]).
- *AI skills* documentation and workflows were standardized with an explicit skill root, and *Android memory analysis* workflows were enhanced to process multiple heap dumps from a single trace ([2302ea01 · Lalit Maganti]).
- Reusable components and utility functions were extracted from the *Perfetto UI's Memscope* plugin ([d9c6f174 · Steve Golton]), improving UI consistency and reusability.
- New *UI chart components*, `ProportionBar` and `FlamegraphChart`, were introduced to expand visualization capabilities within the *charting subsystem* ([7cd2217c · Steve Golton]).
Observations
- Total output increased 36% (174 current vs 127 5-month average), Grow score increased 37% (58 current vs 42 5-month average), and Maintenance score increased 37% (95 current vs 69 5-month average) compared to the 5-month average, indicating a highly productive period focused on both new development and system upkeep.
- The Waste score increased 30% (21 current vs 16 5-month average), driven by several critical bug fixes and correctness issues, such as resolving incorrect machine labels in *Trace Processor* UI ([0765201a · Lalit Maganti]), fixing *DeepSeek-V3* tool-call parsing in *LiteLLM* ([c5b2caad · ChunFu]), and addressing a *Firestore client library* precision issue ([61471b86 · shollyman]).
- Commit volume decreased 13% (657 current vs 752 5-month average), suggesting that the increased output and score metrics were achieved with fewer, potentially larger or more impactful, individual commits.
- A significant pattern of maintenance activity involved addressing `ST1020` and `ST1021` linter violations across numerous Go packages, including `vm`, `tools`, `prog`, `syz-cluster`, `dashboard`, and `pkg` (e.g., [3c9e31db · Alexander Potapenko], [58bf2096 · Alexander Potapenko], [commit/73432ca2], [d0f790ac · Alexander Potapenko], [84d041ce · Alexander Potapenko], [49618c54 · Alexander Potapenko], [ac883038 · Alexander Potapenko], [1d86acc5 · Alexander Potapenko], [2836eca3 · Alexander Potapenko], [87d702b7 · Alexander Potapenko], [d64e1eb8 · Alexander Potapenko], [00c9c07a · Alexander Potapenko], [0e6b3f96 · Alexander Potapenko], [2b3dde56 · Alexander Potapenko]), indicating a focused effort on GoDoc comment quality and code consistency.
- The *Perfetto UI* was a major area of development, with multiple commits dedicated to enhancing memory analysis features, improving UI components, and fixing related issues (e.g., [8b1b64d0 · Steve Golton], [e427da76 · Steve Golton], [aa839152 · Steve Golton], [d9c6f174 · Steve Golton], [969ed94e · Lalit Maganti], [0765201a · Lalit Maganti], [b4bc3287 · Lalit Maganti], [6c61f7f0 · Steve Golton]).
- Extensive refactoring was observed within the *ADK workflow* system, including extracting the agent transfer loop ([7329f7d3 · Shangjie Chen]) and unifying node execution logic ([51e2f52b · Shangjie Chen]), aimed at improving consistency and maintainability.
Repositories
Active repositories ranked by average performance per developer per month (over the last 90 days). The chart shows monthly performance composition — each repo as a stacked layer, with the top of the stack representing total org performance per month. Top 9 repos shown; the remainder is aggregated as “Other”.
| Repository | ||||
|---|---|---|---|---|
| guava | 3 | 44 | 4.9 | +233%since Q2 2025 |
| perfetto | 22 | 167 | 2.5 | +102%since Q2 2025 |
| syzkaller | 6 | 45 | 2.5 | +173%since Q2 2025 |
| google-cloud-python | 19 | 84 | 1.5 | +34%since Q2 2025 |
| skia | 13 | 44 | 1.1 | −3%since Q2 2025 |
| zerocopy | 2 | 7 | 1.1 | +584%since Q2 2025 |
| adk-python | 14 | 44 | 1.0 | +27%since Q2 2025 |
| adk-go | 3 | 8 | 0.9 | +616%since Q2 2025 |
| google-cloud-go | 19 | 50 | 0.9 | +148%since Q2 2025 |
| go-github | 4 | 8 | 0.7 | +270%since Q2 2025 |
| flatbuffers | 2 | 4 | 0.6 | −92%since Q2 2025 |
| python-genai | 12 | 9 | 0.2 | −59%since Q2 2025 |
Company total12 repositories | 107unique devs | 515ETV total | 1.60ETV / dev / mo | +65%since Q2 2025 |
| Performance (ETV) is the sum of every repository above. Active devs at the company level counts unique contributors across all repos, so a contributor working in multiple repos is counted once here but appears in each repo's row (the per-repo column will sum higher). ETV / dev / mo = Company ETV ÷ unique devs ÷ 3 mo. The "Since start" column compares each repo's Q1 2026 quarterly performance to the first quarter it had any activity — for repos that existed in Q2 2025 (when this index began), that's Q2 2025; for younger repos it's the quarter they actually started. The company row uses Q2 2025 as the baseline since the index itself began then. | ||||
Performance Growth vs Active Contributors
Shows how engineering performance scales relative to team growth. Left axis shows total performance score, right axis shows active contributor count. The gap between curves represents productivity gains — more delivered per person, not just more people. Unit: Engineering Throughput Value (ETV).
Cost per Performance Unit
−47%
If performance per engineer grew 89%, each unit of engineering performance now costs approximately 47% less than at the baseline 90-day window (ending 2025-06-29). This is a directional estimate — the exact figure depends on fully-loaded engineer cost, but the direction is unambiguous.
Effective Capacity Added
+95 engineers
At today's productivity, the current 107-person team delivers the performance equivalent of 202 engineers at the baseline 90-day rolling window (ending 2025-06-29). That's roughly 95 engineers worth of capacity added through productivity gains, not hiring.
Performance Composition
Stacked bars show total complexity performance split into Growth (new value), Maintenance (sustaining systems), and Fixes (rework). The yellow line overlays performance per contributor — rising line means each engineer is delivering more, regardless of team size changes. Unit: Engineering Throughput Value (ETV).
CapEx vs OpEx
Monthly CapEx vs OpEx split. CapEx (capitalizable investment) is Growth — new features and capabilities. OpEx (operating expense) is Maintenance plus Fixes — keeping the lights on and reworking what's already shipped. The yellow line is the CapEx share, a quick read on how much of the month went into building new vs sustaining existing. Unit: Engineering Throughput Value (ETV).
Hours per Repository
Trailing 90-day window (64 working days). Org-level capacity is allocated to each repo by its share of org performance, then split CapEx / OpEx by that repo's own Growth vs Maintenance + Fixes mix.
| zerocopy | 61.1% | 38.9% |
| adk-go | 58.0% | 42.0% |
| perfetto | 51.8% | 48.2% |
| python-genai | 48.1% | 51.9% |
| google-cloud-go | 38.5% | 61.5% |
| syzkaller | 35.8% | 64.2% |
| skia | 25.4% | 74.6% |
| google-cloud-python | 24.9% | 75.1% |
| adk-python | 21.7% | 78.3% |
| go-github | 15.2% | 84.8% |
| flatbuffers | 2.4% | 97.6% |
| guava | 0.7% | 99.3% |
| Total | 34.6% | 65.4% |
Fix Burden Distribution
Monthly rework volume broken down by who did it. Top contributors carry their named slice; everyone else is rolled into Others. Use this to spot whether fix work is concentrated on a small group (bus-factor risk) or distributed across the team.
Quarterly Summary
The raw numbers behind the charts: commits analyzed, active contributors, total performance, performance per developer, and the Growth / Maintenance / Fixes split for each quarter.
| Quarter | ||||||||
|---|---|---|---|---|---|---|---|---|
| Q2'25 | 2,407 | 119 | 305.04 | 0.9 | 35.6% | 54.5% | 9.8% | — |
| Q3'25 | 2,916 | 126 | 351.99 | 0.9 | 32.4% | 57.8% | 9.8% | +15% |
| Q4'25 | 2,830 | 136 | 411.27 | 1 | 33.8% | 55% | 11.1% | +17% |
| Q1'26 | 3,763 | 130 | 503.95 | 1.3 | 34.5% | 53.3% | 12.1% | +23% |
Top Contributors
Contributors ranked by performance per month (Growth + Maintenance + Fixes), over the last 90 days normalized to a 30-day calendar month.
The best way to measure AI efficiency
SampleA preview of the Navigara engine running on a sample organization. The numbers below are illustrative, not part of the OSS500 benchmark above.
Measure
Score every commit by depth
GitHub commits are weighted by what it took to write them, not by lines of code. The result is ETV per developer per month.
SourceGitHub
Spend
Tie ETV to cost
AI token bills and seat costs are pulled per team and divided by the ETV produced. The result is your true cost per unit of work.
SourceToken usage + finance
Map
Tie work to objectives
Each ETV is mapped to your Jira epics and labels, so you can see what's key-aligned, aligned, or unmapped capacity.
SourceJira
Performance
ETV delivered per developer / month
9.4ETV / dev / month
5.6 below target- Non-AI 5.8
- AI 3.6
AI Efficiency
AI spend per ETV unit delivered
$4.20
$0.60 over target- Cost / ETV $4.20
Objective Alignment
Share of work mapped to key objectives
51%
24 pts below target- Key-aligned 28%
- Aligned 23%