Published research
DORA Metrics: What They Show and What They Don't
DORA now tracks five metrics, not four. Here's what they measure, what they can't tell you, and why AI adoption is pulling throughput and stability apart.
4 related · metrics
Engineering measurement · Independent
The median is not the team.
What the frameworks actually measure, what the platforms actually cost, and where both stop being true. Every article states its evidential basis.
The tail is what breaks the roadmap — and it does not appear in the average.
Published research
The DX Core 4, Explained: What It Fixes and What It Doesn't
Core 4 folds DORA, SPACE and DevEx into four dimensions. Here's how it works, why 'diffs per engineer' is controversial, and where it still falls short.
4 related · metrics
Published research
The SPACE Framework in Practice: Five Dimensions, One Common Mistake
SPACE says productivity is multidimensional and no single metric captures it. Here's how to apply it, and why its Activity dimension is now the weak one.
4 related · metrics
Published research
DevEx: Feedback Loops, Cognitive Load, Flow State
The framework that measures the lived experience of developers rather than the artefacts they produce — three dimensions, how to measure them, and its limits.
4 related · metrics
Published research
Free and Open-Source DORA Tooling: What Actually Works in 2026
You can run DORA metrics without paying a vendor. Here's what's available, what each option costs you in engineering time, and when free stops being cheaper.
4 related · tools
Published research
Metadata or Code-Level Analytics: Which Architecture Do You Need?
Engineering analytics platforms split into two architectures. The difference decides what questions you can answer — and how long your security review takes.
4 related · tools
Documentation & public sources
Flow Is Retiring: The Dates, and What to Move To
Appfire Flow (ex Pluralsight Flow, ex GitPrime) reaches end of life 31 December 2027 and renewals closed 30 June 2026. The verified dates and how to pick a replacement.
4 related · tools
Documentation & public sources
LinearB vs Swarmia: Two Different Answers to the Same Problem
LinearB automates the workflow. Swarmia measures the team and asks it to change. Which one fits depends on a question about your organisation, not your stack.
4 related · tools
Published research
What Engineering Analytics Platforms Actually Cost
Most vendors in this category hide pricing. Here's what buyers report paying, where the figures disagree, and how to read a quote before you sign one.
4 related · tools
Documentation & public sources
Jellyfish vs LinearB: Boardroom Tool or Workflow Tool
These two platforms answer different people's questions. Jellyfish is built for the CTO reporting to a board; LinearB for the manager fixing a review bottleneck.
4 related · tools
Published research
Why DORA Metrics Can't Tell You If AI Is Working
Your DORA dashboard went green after rolling out Copilot, and your engineers say they feel slower. Here's why both can be true.
4 related · ai-measurement
Published research
Technical Debt After AI Adoption: What the Data Actually Shows
Two large 2026 datasets on AI-assisted code and maintenance burden — what they found, what they don't prove, and how to read them before quoting them.
3 related · ai-measurement
Published research
AI Code Attribution: The Approaches and Where Each One Breaks
Four ways to work out how much of your codebase came from AI assistants — IDE telemetry, heuristics, commit provenance and surveys — and the failure mode of each.
3 related · ai-measurement
Published research
What to Tell the Board About AI ROI
You've been asked for a return figure on AI coding tools. Here's a defensible answer, why the confident version is worse than the honest one, and what to bring instead.
4 related · ai-measurement
Published research
The Metrics That Stopped Working When AI Started Writing Code
Six measures that were reasonable proxies until assistants changed what they proxy for — and what still holds up.
4 related · ai-measurement
Documentation & public sources
Tools That Claim to Measure AI Impact: What Each Can Actually See
A new product category formed around one question boards are asking. Here's what the different architectures can and cannot establish.
4 related · ai-measurement
Published research
Individual Productivity Metrics Break Teams. The Evidence Is Not Ambiguous.
The argument against per-developer measurement, the 2023 debate that made it public, and the peer-reviewed research on what monitoring does to teams.
4 related · opinion
Published research
Why Your DORA Metrics Are Lying to You
Seven ways a DORA dashboard produces confident, wrong numbers — and how to tell whether yours is measuring delivery or measuring your configuration.
4 related · opinion
Published research
The Dashboard Nobody Opens
Most engineering metrics programmes fail quietly: the data is fine, the dashboard exists, and nobody looks at it. Why that happens and how to fix it.
4 related · opinion
Published research
Metrics for Managing and Metrics for Reporting Are Not the Same Metrics
One set helps a team decide what to change this week. The other explains engineering to people outside it. Most programmes fail by trying to use one set for both.
4 related · opinion
Published research
"Engineering Efficiency" Is the Wrong Frame
Efficiency is output over input. Software development has neither a stable output nor a comparable input, which is why the frame keeps producing bad decisions.
4 related · opinion
Published research
How We Test and What We Disclose
Our testing process, data sources, funding and correction policy — so you can judge how much weight to give what you read here.
0 related · practice
Published research
Your First 90 Days With Engineering Metrics
A sequenced plan for introducing delivery metrics without triggering the resistance that kills most measurement programmes in month two.
4 related · practice
Published research
Measuring a Platform Team Without Measuring the Wrong Thing
Platform teams ship to other engineers, so delivery metrics measure their customers rather than them. Here's what to measure instead.
4 related · practice
Published research
How to Introduce Metrics Without Losing the Room
The conversation that determines whether a measurement programme works. What to say, what not to say, and what the research shows about framing.
3 related · practice
Get new analysis by email
Independent work on engineering measurement. No vendor sponsorship, no affiliate placement, no weekly cadence padded with links.