Technical Product Management Course · by Stanislav Belyaev
EN RU

PRs Completed per Week

5 outgoing · 17 incoming · 22 total connections

Map Detail
Developer Experience AMPLIFIED IN MONOREPO AMPLIFIED DISTRIBUTED

PRs Completed per Week

PRs Completed per Week tracks the number of pull requests each developer merges in a given week, serving as a throughput indicator. It provides a rough proxy for individual and team output velocity when viewed alongside PR size and complexity. This metric is most useful in trend analysis rather than absolute comparison, as it must be interpreted in the context of work type and team norms.

PRs merged/closed per developer per week. Elite teams: 15-25 PRs/week with small PRs. Low performers: 2-5 PRs/week. This is THE velocity metric — directly measures how fast developers ship working code.

MONOREPO CONTEXT

CRITICALLY AMPLIFIED: Without smart CI tooling (Bazel, Nx, Turborepo), every PR triggers full-repo builds and tests, creating massive bottlenecks. PR throughput can drop 60-80% vs polyrepo without affected-project detection + remote caching. Google's investment in build tools is specifically to maintain PR velocity at scale.

DISTRIBUTED CONTEXT

CRITICALLY AMPLIFIED: THE #1 distributed-team productivity killer. Each PR requires cross-TZ review handoffs (12-24h each), merge queue failures during sleep hours, and batching to minimize round-trips. PR completion rate drops 50-70% in distributed teams vs colocated. A colocated dev completing 20 PRs/week may drop to 6-8 PRs/week distributed across 3 TZs.

Scale Impact
👤 Solo / Pair (1–3)
0.3
👥 Team (4–15)
0.6
🏢 Department (15–100)
0.8
🏛️ Organization (100+)
1

PR throughput drops 60-80% without smart CI at monorepo scale. Cross-TZ review handoffs add 12-24h per round-trip, crushing velocity.

5
Influences
17
Influenced by

→ Influences

Developer Satisfaction

Completing and shipping work is deeply satisfying. High throughput = visible progress = motivation.

Velocity is core to satisfaction
Developer satisfaction / HashiCorp DevEx
Distributed: Perpetually blocked PRs in distributed teams create chronic frustration. Low PR completion rate is a major attrition driver.
Deployment Frequency

More completed PRs = more deployments. Direct relationship in CD environments.

1:1 in trunk-based dev
DORA / Trunk-based development
Change Lead Time

Higher PR throughput reduces WIP and queue times, cutting lead time.

Throughput-latency trade-off
Queueing Theory / DORA
High CriticalDIST
Attrition

Developers leave when they feel unproductive. Low PR completion rate signals throttled velocity.

Velocity frustration → attrition
Developer retention research
Distributed: Distributed devs feeling artificially throttled by TZ gaps leave fastest. PR velocity is highly visible frustration metric.
Context Switching

Completing PRs reduces WIP count. Lower WIP = fewer context switches between stalled items.

WIP limit reduces switching
WIP limits / Kanban research

← Influenced by

Code Review Turnaround

Each additional hour of review latency directly reduces weekly PR throughput. 24h review → devs batch changes → fewer PRs.

Linear relationship
Axify, Code Climate, Haystack
Distributed: Cross-TZ review latency (12-24h) is the dominant bottleneck. Distributed teams complete 50-70% fewer PRs than colocated teams purely from review delays.
Merge Queue Wait

Queue failures reset PR progress. Each failure = hours or days of delay before retry completes.

Expedia: 20+ PRs competing
Trunk.io / Merge queue scalability
Monorepo: Monorepo merge queues are especially brutal without queue partitioning. A single failure invalidates all queued PRs.
Distributed: Queue failures overnight mean devs wake up to failed merges, resubmit, and wait another 12-24h cycle.
CI/CD Pipeline Speed

Slow CI creates iteration bottleneck. 45-min pipeline = max 10-12 iterations/day. Fast 5-min pipeline = 50+ iterations/day.

Direct throughput limit
CI/CD pipeline optimization
Monorepo: Without affected-project detection, monorepo CI becomes the primary PR velocity killer. Every PR waits 45+ minutes.
Test Flakiness

Each flaky failure requires re-run. At 16% flakiness (Google), most PRs hit at least one flake.

Google: 84% pass→fail are flaky; reruns add pipeline cost
Google flaky test analysis
Monorepo: Broader dependency graphs in monorepos mean more tests run per PR = more flake exposure. Exponentially impacts throughput.
Distributed: Flaky failures overnight create 12-24h delays instead of immediate retries. Amplifies impact dramatically.
Context Switching

Each context switch costs 23 minutes to recover. High-interrupt environments prevent completing PRs.

Only 2.3 hrs deep work/day
UC Irvine (Gloria Mark), Uplevel
Distributed: Morning context loading from overnight notifications destroys the first work block, reducing available PR completion time by 1-2 hours daily.
Flow State

Flow state enables 500% productivity bursts. Deep work sessions are when PRs get completed.

Up to 500% productivity in flow
Csikszentmihalyi 10-year study
Distributed: Distributed teams lose morning flow blocks to context loading. Fewer flow hours = fewer completed PRs.
High CriticalMONODIST
PR Size

Larger PRs take longer to write, review, and merge. 200-line PR = 1 day. 2000-line PR = multiple days.

Optimal: 200-400 lines
Graphite, LinearB, PropelCode
Monorepo: Cross-cutting monorepo changes create massive PRs (15+ services) unless using stacked PRs. Massive PRs kill throughput.
Distributed: Batching to minimize cross-TZ round-trips creates larger PRs, creating vicious cycle: large PR → more review rounds → more batching.
High CriticalMONO
Test Suite Exec Time

Test execution is often the CI bottleneck. Long test suites limit iteration speed.

Primary CI bottleneck
CI/CD bottleneck analysis
Monorepo: Full-repo test suite without selective testing destroys PR velocity. Must have smart test selection.
High CriticalMONO
Build Times

Each PR requires multiple builds (dev build + CI builds + review iterations). 10-min build × 10 iterations = 100 min lost.

2-min build × 200/day = 100+ min
DevOps workflow analysis
Monorepo: 2B-line monorepos without distributed builds create catastrophic build bottlenecks that make PR completion nearly impossible.
High CriticalDIST
Cognitive Load

High cognitive load slows down all work. Overloaded devs make less progress per hour.

76% cite stress
Developer burnout and cognitive load research
Distributed: Morning cognitive overload from overnight context loading reduces effective work hours, directly cutting PR throughput.
High CriticalDIST
Technical Debt

Debt makes every change take longer. 25-50% slower delivery in high-debt codebases.

23-33% time on debt
Stripe 2023, Academic research 2024
Distributed: Tech debt slowdowns are worse when each iteration costs 12-24h. A change that takes 3 iterations instead of 1 costs 2-4 days extra in distributed teams.
High CriticalDIST
Documentation Quality

Good docs eliminate blocking questions. Self-service answers = uninterrupted PR work.

30% senior time saved
Stack Overflow / Developer productivity
Distributed: Can't ask sleeping colleagues. Without docs, devs get blocked for 12-24h waiting for answers, destroying PR throughput.
Medium LowMONO
Tool Fragmentation

App-switching overhead adds up. 1,200+ toggles/day = 4 hrs/week lost.

~4 hrs/week
Harvard Business Review 2022
Monorepo: Monorepos reduce tool fragmentation, weakening this dependency.
Medium LowDIST
Meeting Load

Meetings fragment the day. 11 hrs/week in meetings leaves less time for PR completion.

~11 hrs/week
NetworkPerspective, Uplevel, Reclaim
Distributed: Distributed teams have fewer synchronous meetings by necessity, reducing impact.
Handoff Latency

Each PR requires multiple handoffs (submit → review → address feedback → merge). 12-24h per handoff means 2-3 day minimum per PR vs same-day in colocated teams.

THE dominant distributed bottleneck
DORA, DevOps throughput research
AI Tools Adoption Rate

Individual productivity boost: +21-98% more PRs per developer. However, organizational impact is UNCHANGED or slightly NEGATIVE due to review bottlenecks and stability risks.

DORA 2024: 25% AI adoption correlates with 1.5% decrease in team throughput.
DORA 2024 State of DevOps
AI Learning Curve

19% slower initially for experienced devs. 4-6 weeks to break even, 8+ weeks for full productivity.

Initial productivity dip
Microsoft Research
Metrics map by Stanislav Belyaev · Analysis powered by Anthropic Claude Opus 4.6 · All data validated by human experts