Throughput

Throughput measures how much work flows through a system or team in a given time period. For product teams, this typically means how many items - stories, tasks, or features - are completed per week or sprint. Unlike velocity, which measures estimated effort, throughput counts completed items regardless of their size, providing a simpler and often more useful metric for understanding delivery capability.

Why it matters

Understanding throughput helps teams answer practical questions: How much work can we realistically complete? Are we getting faster or slower over time? When might we finish the current backlog? These questions matter for planning, stakeholder communication, and continuous improvement.

Throughput is particularly valuable because it's based on actual completion, not estimates. When teams measure velocity using story points, they're measuring their estimates of work. When they measure throughput, they're measuring work that's actually done. This makes throughput less susceptible to gaming and estimation inflation.

Throughput vs. velocity

Both metrics describe team productivity but differ importantly:

Velocity measures story points (or other estimates) completed per sprint. It relies on the accuracy of estimates and is specific to each team's estimation scale.

Throughput measures items completed per time period, regardless of size. It's simpler, more objective, and easier to compare across teams.

Velocity can mislead when estimates are inconsistent or inflated over time. A team that increases story point estimates without changing actual productivity will show "improving" velocity. Throughput avoids this - you either finished the item or you didn't.

That said, throughput works best when work items are roughly similar in size. If items vary dramatically, throughput alone doesn't capture the full picture.

Measuring throughput

Effective throughput measurement requires:

Consistent item definition. Count items at a consistent level of granularity. If sometimes you count epics and sometimes you count tasks, throughput becomes meaningless.

Clear completion criteria. Define what "done" means for items being counted. Is it merged to main branch? Deployed to production? Verified by QA? Consistency matters.

Regular time periods. Track throughput over consistent periods - weekly, bi-weekly, or per sprint. This enables trend analysis and forecasting.

Historical data. Throughput's value grows with historical data. A single data point says little; months of data reveal patterns.

Throughput in kanban

Throughput is particularly central to Kanban systems, which optimize flow rather than time-boxed sprints:

Flow metrics. Kanban emphasizes continuous flow. Throughput shows whether flow is healthy and improving.

WIP limits connection. Work-in-progress limits improve throughput by reducing context switching and helping items complete faster. Changes to WIP limits should reflect in throughput changes.

Cumulative flow diagrams. These visualizations show throughput as the rate at which items move from "doing" to "done" - the gap between lines on the diagram.

Using throughput for forecasting

Throughput enables probabilistic forecasting using Monte Carlo simulation:

This approach acknowledges uncertainty honestly. Instead of promising "we'll finish by March 15," you can say "there's a 85% chance we'll finish by March 15 and a 50% chance by March 1."

Factors affecting throughput

Understanding what influences throughput helps teams improve:

Work item size. Smaller items complete faster and more predictably. Breaking down large items improves throughput and reduces variability.

Work in progress. Too much concurrent work slows everything down. Counter-intuitively, starting fewer things can increase completion rate.

Dependencies. Waiting for external teams, reviews, or resources blocks completion. Reducing dependencies improves throughput.

Process bottlenecks. Stages where items accumulate constrain the whole system. Identifying and addressing bottlenecks lifts overall throughput.

Team stability. Team changes disrupt throughput as members onboard and relationships adjust. Stable teams generally have higher, more predictable throughput.

Throughput patterns

Healthy and unhealthy patterns appear in throughput data:

Steady throughput. Consistent completion rate week over week indicates a stable, predictable team. This supports reliable forecasting.

Improving trend. Gradually increasing throughput suggests successful process improvements or team maturation.

High variability. Large swings in throughput make forecasting difficult and may indicate process problems, unclear item definitions, or irregular prioritization.

Declining trend. Decreasing throughput warrants investigation. Causes might include growing technical debt, team changes, or increasingly complex work.

Batch completion. Items completing in clusters rather than continuously suggests process issues - perhaps too much WIP or waiting for batch deployment.

Common pitfalls

Several mistakes undermine throughput as a metric:

Gaming the metric. Splitting items artificially to increase throughput numbers without delivering more value. Guard against this by watching whether customers notice any change.

Ignoring size variation. If throughput is 10 items per week, that means different things if items are consistently sized versus wildly variable. Consider tracking item size distribution alongside throughput.

Comparing across teams. Different teams work on different types of items. Comparing their throughput directly can be misleading.

Sacrificing quality. Pushing to increase throughput at the expense of quality creates technical debt that eventually slows throughput down.

Short-term focus. Optimizing throughput this week at the expense of sustainable pace harms long-term throughput. Sustainable delivery beats bursts followed by burnout.

Throughput and system performance

Beyond team productivity, throughput applies to systems:

API throughput. Requests processed per second, measuring system capacity.

Transaction throughput. Orders, payments, or other transactions processed per time unit.

Feature throughput. Rate at which product capabilities reach customers, connecting delivery to value.

These system metrics help product teams understand whether infrastructure supports growth and whether bottlenecks exist in value delivery.

The product manager's perspective

Product managers benefit from throughput understanding in several ways:

Realistic planning. Knowing historical throughput helps set realistic expectations for how much can be accomplished in a given timeframe.

Stakeholder communication. Throughput provides concrete data for conversations about timelines. "Based on our throughput, this backlog will take 8-12 weeks" is more defensible than intuition-based estimates.

Prioritization consequences. Understanding throughput helps product managers see the cost of adding items. Every item added pushes other items further out based on the team's completion rate.

Process investment ROI. When the team invests in process improvements, throughput shows whether they worked. If throughput increases after reducing WIP limits, the change was valuable.

The modern context

Modern product teams increasingly favor flow-based approaches over rigid sprint planning. Throughput fits this paradigm by measuring actual completion rather than estimated work within artificial time boundaries.

Tools like Klero complement throughput thinking by helping teams ensure they're completing the right work, not just more work. When customer feedback informs prioritization, higher throughput translates to more customer value delivered. Throughput without value focus risks optimizing for output rather than outcomes.