For the complete documentation index, see llms.txt. This page is also available as Markdown.

Best Practices

This guide maps common engineering challenges to the Logilica dashboards and metrics that help you understand, diagnose, and improve them.

How to Use This Guide

  1. Identify your challenge from the sections below

  2. Navigate to the recommended dashboards to understand the current state

  3. Look at the suggested indicators to diagnose root causes

  4. Take the recommended actions to drive improvement

  5. Track progress over time using the same dashboards


Delivery Speed & Lead Time

Challenge: Features and tickets take too long from creation to completion.

Dashboards to Use

Dashboard
What to Look For

Stage breakdown — identify whether delays are in Backlog, Pickup, or Resolution

PR stage breakdown — identify whether delays are in Development, Response, Review, or Integration

Long Running WIP tickets — items stuck for extended periods

Key Indicators

  • Long Backlog stage: Tickets are created but sit unassigned — indicates prioritisation or capacity issues

  • Long Pickup stage: Tickets are assigned but work hasn't started — indicates overloaded assignees or unclear requirements

  • Long Resolution stage: Work has started but isn't completing — indicates large scope, blockers, or context switching

  • Long Response stage (PRs): PRs are open but reviews haven't started — indicates insufficient review capacity

  • Break down large tickets into smaller, well-specified tasks (smaller tasks have shorter lead times)

  • Address the specific bottleneck stage rather than applying general pressure

  • If Pickup is long, check Ticket Overload to see if assignees are juggling too many items

  • If Response is long on PRs, allocate more review capacity or reduce PR sizes


Throughput & Velocity

Challenge: The team isn't shipping enough work relative to expectations.

Dashboards to Use

Dashboard
What to Look For

Throughput trends, backlog growth rate, ticket size distribution

PR merge rate, PR backlog accumulation, work focus (New vs. Rework vs. Maintenance)

Sprint overrun %, planned vs. unplanned work ratio

Key Indicators

  • Growing backlog: Tickets or PRs are accumulating faster than they're being closed — capacity shortfall

  • High Maintenance %: A large share of coding effort goes to maintaining old code rather than new features — signals technical debt

  • High Rework Others %: Developers frequently change each other's recent code — may indicate unclear ownership or poor initial quality

  • High unplanned work ratio: Frequent interruptions displace planned sprint work

  • Reduce ticket and PR sizes — smaller items have higher throughput

  • If Maintenance % is high, allocate dedicated time for technical debt reduction

  • If unplanned work is high, implement stronger sprint boundaries or triage processes

  • Monitor the investment breakdown over time to ensure effort aligns with organisational priorities


Sprint Planning Accuracy

Challenge: Sprints consistently overrun or planned work doesn't get completed.

Dashboards to Use

Dashboard
What to Look For

Overrun %, unplanned work ratio, investment breakdown

Whether ticket sizes are consistent and predictable

Whether team members have too many concurrent assignments

Key Indicators

  • Sustained high overrun %: Planning consistently underestimates effort or scope — not a one-off but a pattern

  • High unplanned work: External interruptions are displacing planned items

  • Variable ticket sizes: Inconsistent task sizing makes estimation unreliable

  • Use historical velocity as a planning input rather than aspirational targets

  • Standardise ticket sizing (story points or t-shirt sizes) and track estimation accuracy

  • Ring-fence a portion of sprint capacity for unplanned/interrupt work

  • Focus on trends across sprints rather than single-sprint anomalies


Team Health & Burnout Risk

Challenge: Team members are overloaded, context switching too much, or at risk of burnout.

Dashboards to Use

Dashboard
What to Look For

% of team with too many concurrent tickets — sustained overload is a red flag

% of developers with too many concurrent open PRs

Per-member load status (Available / Active / Overloaded), WIP counts, blockers

What each person is actively working on, how long items have been in progress

Key Indicators

  • Sustained overload across the team: Not a temporary spike but a structural issue — too much work for available capacity

  • Individual outliers: Specific team members consistently overloaded while others are available — work distribution issue

  • High WIP per person: People juggling many items leads to context switching and slower delivery on all of them

  • Redistribute work from overloaded team members to those with available capacity

  • Reduce overall WIP limits — fewer concurrent items means faster completion of each

  • Use Team Pulse for regular check-ins rather than as a performance measurement tool

  • If overload is sustained across the team, escalate as a resourcing issue


Code Quality & Review Process

Challenge: Code quality issues, review process violations, or too many risky changes shipping without proper review.

Dashboards to Use

Dashboard
What to Look For

Merge success rate, % merged without approval, process flow bypass rates

Risky Changes, Complex Reviews, PRs merged outside process

Work focus breakdown — high Rework Others % may indicate quality issues

Key Indicators

  • PRs merged without review/approval: Direct process violation — quality risk

  • High % of Risky Changes: Large PRs that are harder to review thoroughly

  • High Complex Review count: Reviews requiring many comments or cycles — may indicate unclear requirements or large scope

  • High non-productive work: PRs opened but abandoned — wasted effort

  • Enforce review and approval requirements — target 0% merged outside process

  • Break down large PRs to reduce the number of Risky Changes

  • Investigate Complex Reviews to understand whether they stem from PR size, unclear requirements, or domain complexity

  • Track non-productive work to understand why PRs are being abandoned


Build & CI/CD Reliability

Challenge: Builds are slow, unreliable, or blocking team progress.

Dashboards to Use

Dashboard
What to Look For

Build reliability %, average duration, recovery time, per-pipeline breakdown

Key Indicators

  • Low reliability: High failure rate means teams are frequently blocked waiting for fixes

  • Long build duration: Slow builds reduce developer productivity and increase cycle time

  • Long recovery time: When builds break, how long until they're fixed — long recovery means extended team blocking

  • Prioritise improvements where: reliability is low AND many teams depend on the pipeline AND duration is long

  • Reduce build times to remove productivity blockers

  • Improve reliability to reduce developer frustration and stalled releases

  • Track recovery time and establish on-call or ownership for critical pipelines


Epic & Project Delivery

Challenge: Epics or large initiatives are slipping against target dates.

Dashboards to Use

Dashboard
What to Look For

Burndown chart, forecast vs. target variance, code activity correlation

Key Indicators

  • Upward spikes in burndown: New items being added mid-epic — scope creep

  • Forecast date significantly after target: Velocity-based projection shows a miss

  • Low code activity despite many open items: Work isn't progressing even though tickets exist

  • High variance: Widening gap between forecast and target over time

  • Track and limit mid-epic scope additions

  • Use the velocity-based forecast as an early warning signal for stakeholder communication

  • Investigate disconnect between open items and code activity — items may be blocked, poorly specified, or waiting on dependencies

  • Adjust scope or timeline early rather than hoping velocity will increase


Cross-Team Visibility

Challenge: Need to understand how multiple teams are tracking and where to focus management attention.

Dashboards to Use

Dashboard
What to Look For

Per-team metrics: cycle time, WIP, work focus, risk counts

Key Indicators

  • Outlier teams: Teams with significantly different cycle times, risk counts, or work focus compared to peers

  • High risk counts on specific teams: Accumulated Stale PRs, Long Running WIP, or Risky Changes

  • Compare teams with caution — different teams have different work types, cadences, and domain complexity

  • Use outliers as signals for follow-up conversations, not as performance rankings

  • Drill into specific teams that show anomalies to understand root causes before acting


Metrics Anti-Patterns

Engineering metrics are powerful tools for improvement, but they cause harm when misused. Avoid these well-documented anti-patterns:

Don't Use Metrics for Individual Performance Ranking

Metrics like cycle time, PR count, or lines of code should never be used to rank or evaluate individual developers. The most impactful engineers often show lower personal throughput because they spend time unblocking others, mentoring, or doing architectural work that doesn't show up in commit counts. Using metrics for individual ranking encourages gaming (splitting work into trivial PRs, avoiding reviews, cherry-picking easy tickets) and erodes trust.

Use metrics at the team level to identify systemic issues. Use one-on-one conversations to understand individual contributions.

Don't Optimise One Metric in Isolation

Improving one metric at the expense of others is counterproductive:

  • Inflating deployment frequency with trivial changes doesn't improve delivery capability

  • Reducing change failure rate by deploying less defeats the purpose of tracking delivery performance

  • Pushing for shorter cycle times by skipping reviews trades speed for quality

The DORA metrics, work type breakdowns, and risk indicators are designed as a balanced set — interpret them together, not individually.

Don't Treat Thresholds as Targets

When a metric becomes a target, it ceases to be a useful measure (Goodhart's Law). Logilica's thresholds for Risky Changes, overload, and build reliability are signals for investigation, not pass/fail criteria. A team that consistently exceeds a threshold may have a legitimate reason — large PRs in a migration phase, high WIP during a critical launch.

Investigate anomalies before acting on them.

Account for AI-Generated Code

If your team uses AI coding tools (Copilot, Cursor, Claude Code), be aware that AI-assisted development can distort traditional metrics:

  • Deployment frequency may increase without a proportional increase in meaningful output

  • PR sizes tend to grow with AI assistance, which can inflate "Risky Change" counts

  • Review time may increase as reviewers spend more time validating AI-generated code

  • Code quality signals like rework rate may behave differently for AI-generated code

Monitor your AI Coding metrics alongside delivery metrics to understand the full picture.


Setting Up Targets & Thresholds

Logilica includes configurable targets and thresholds with industry defaults. When setting your own:

  • Start with defaults and adjust based on your team's context and baseline

  • Focus on trends rather than absolute values — improvement matters more than hitting a specific number

  • Review thresholds quarterly as team maturity and processes evolve

  • Use thresholds as conversation starters, not enforcement mechanisms

Last updated