# LinearB vs Swarmia vs Jellyfish vs DX for engineering bottlenecks

Four engineering analytics platforms plus the free options in GitHub, GitLab and Apache DevLake, compared on what they measure, what they cost in 2026 and when each is the wrong buy.

Updated 25 September 2026 . 11 min read . Alexey Shurov

## Which software actually finds engineering bottlenecks

The software that finds engineering bottlenecks splits the time from first commit to production into stages and shows which stage holds the work. In the delivery data I have looked at, the answer is usually a wait rather than a person, most often a pull request nobody has picked up for review or a merged change queuing for release. LinearB, Swarmia, Jellyfish and Atlassian (DX) are the four platforms most shortlists end up with, and GitHub, GitLab and the open-source Apache DevLake cover much of the same ground for free.

The short version. LinearB and Swarmia are the most direct at showing where a pull request waits, Jellyfish fits when delivery data has to line up with Jira planning and finance, and DX is the pick when the friction is something developers feel but no timestamp records. Export your own pull request timestamps before any trial.

I build and run production AI agents for operational teams, so my bias is towards measuring the waiting yourself before paying a platform to do it. No vendor has paid to be here. Every price was checked on the vendor's own page on 25 September 2026, and where none is published I say so rather than guess.

### 1. LinearB

LinearB splits cycle time into four stages. Coding runs from first commit to the pull request opening, pickup from opening to first review, review from first review to merge, and deploy from merge to production. Pickup is usually the revealing number, because it is pure waiting. Both plans include DORA metrics and developer satisfaction surveys.

It wins for organisations with several teams and a recurring argument about where delivery time goes, and the 45-day free trial is long enough to settle it. It is the wrong buy for a small team. Essentials is 29 dollars per contributor per month billed annually, built, in LinearB's words, for teams of 50 or more developers, and supports GitHub Cloud only. Enterprise is 59 dollars per contributor per month billed annually, aimed at 100 or more developers, and adds GitLab, Bitbucket Cloud and Azure DevOps.

### 2. Swarmia

Swarmia measures the same flow from the team's side. Pull request cycle time is time in progress, time in review and time to merge, with a separate time to first review for the wait before anyone looks. Around that sit DORA metrics, CI visibility, developer surveys, investment balance and working agreements, the targets a team sets itself, such as how fast reviews should start.

It wins for teams that want to change review habits, not only report on them. It is free for companies with fewer than ten developers, with every Standard feature except HR system integrations and Swarmia AI. Standard is 45 dollars and Enterprise 55 dollars per developer per month billed annually, and its Productivity and AI impact plan, with code, issue and DORA metrics, is 23 dollars on its own. It is the wrong buy if nobody will act on what it shows, and its code integrations are GitHub and GitLab, so check first if you are on Bitbucket or Azure Repos.

### 3. Jellyfish

Jellyfish reads delivery through Jira as well as Git. Its Life Cycle Explorer maps work across four phases called refinement, work, review and deployment, so it also covers the gap between a ticket being created and work starting, which Git-only tools never see. Its Developer Productivity tier adds DORA metrics, DevEx surveys, benchmarking and investment allocation, and a DevFinOps tier adds software capitalisation for finance.

It wins for large organisations where delivery data has to line up with planning and finance. Pricing depends on seats and modules, with no figures published, so expect a sales process. It is the wrong buy for one team's review queue, and it needs reasonably clean Jira statuses, because the refinement and work phases are only as good as the ticket data.

### 4. Atlassian (DX)

DX, which now calls itself Atlassian (DX), starts from the other end. Its core is a developer survey, the Developer Experience Index, combined with system data in a framework called DX Core 4 that measures speed, effectiveness, quality and impact. DX says it captures all four DORA metrics inside that framework, and it also measures AI tool adoption and impact.

It wins when you suspect friction that developers feel but no timestamp records, such as flaky test environments or unclear requirements. Atlassian completed the purchase on 10 November 2025 and is adding DX to its Software Collection alongside Bitbucket and Compass. No price is published, so it is a demo and a quote. It is the wrong buy for a single team's review queue, and if you are not an Atlassian shop, ask how the roadmap treats GitHub and GitLab data.

### 5. Waydev

Waydev covers the same delivery ground. Its Pro plan includes DORA metrics, cycle time from first commit to production release, pull request insights and CI/CD reports, and unlike LinearB's plan at the same price, it connects to GitHub, GitLab, Bitbucket, Azure DevOps and Gerrit. A developer experience module sends surveys through Slack.

It wins for mid-size teams on mixed Git hosts who want delivery analytics without an enterprise contract. Pro is 29 dollars and Premium 49 dollars per active contributor per month billed annually, with Premium adding AI impact tracking and 24 months of data. The free Start Up plan is the trap. It covers AI usage and cost on up to 10 repositories with three months of data, so it is the wrong buy if a free view of your review queue is what you came for.

### 6. GitHub Actions performance metrics

GitHub has shown Actions performance metrics on every GitHub Cloud plan since March 2025, with average run time, queue time and failure rate for workflows and jobs. For the human side, the open-source issue-metrics action, built by GitHub's open source office and now maintained under github-community-projects, reports time to first review, time in draft and time to close for pull requests.

This wins when the complaint is slow or flaky builds, and as a free first look at review waits. It is the wrong place to stop if you need every stage side by side, or DORA metrics, because GitHub does not assemble those, and the action sits outside GitHub's support contracts.

### 7. GitLab Value Stream Analytics

GitLab Value Stream Analytics is on every tier, Free included, and shows the median time work spends in six default stages called issue, plan, code, test, review and staging. Review runs from the merge request being created to merge, and staging from merge to the first production deployment. Custom value streams need Premium and DORA metrics need Ultimate.

It wins for any team already on GitLab, because looking costs nothing. The catch is that waiting for a reviewer and reviewing are one number, so you still need the free first pass or a paid tool to split them. It is the wrong tool on its own if your issues live in Jira or you deploy outside GitLab, because those stages depend on GitLab's own data.

### 8. Apache DevLake

Apache DevLake is the open-source option, and a top-level Apache project since October 2025. It pulls data from GitHub, GitLab, Bitbucket, Azure DevOps, Jira and Jenkins into its own database and ships Grafana dashboards, including DORA metrics and pull request coding, pickup, review and deploy times.

It wins for teams with an engineer who will own it and who want DORA metrics and pull request stages without a per-seat licence. It is the wrong buy if nobody owns it, because you host, upgrade and troubleshoot it yourself. Free software is not free operation.

## What changed in 2025 and 2026

Gartner published its first Magic Quadrant for Developer Productivity Insight Platforms on 5 May 2026, and the vendors' own announcements say it named Atlassian (DX), Jellyfish, LinearB and Opsera as leaders, Opsera for its Unified Insights product. A leader badge speaks to scale and direction, not to your team's question.

Ownership moved too. Atlassian bought DX. Appfire, which took over Pluralsight Flow, is discontinuing it, with no renewals beyond 30 June 2026 and end of life on 31 December 2027. Code Climate replaced Velocity with a platform custom-built for each organisation and now sells a twelve-week engagement with its own engineers. Older lists still recommend both. Neither is a sensible start now.

The other change is on every pricing page. LinearB bundles monthly AI credits, Swarmia sells AI adoption and cost as a module, Jellyfish's entry tier is AI Impact and Waydev's free plan measures AI spend. That is a different question from where delivery waits, so do not let the AI dashboards take over a trial you started to find a bottleneck.

## Run the free first pass yourself

Before any trial, export the last quarter of merged pull requests with three timestamps for each, opened, first reviewed and merged, plus the first commit and production deploy if you can get them. The GitHub command-line tool lists merged pull requests with created, merged and review times as JSON, GitLab's API gives merge request and comment timestamps, and a short script or spreadsheet does the rest.

Pickup is first review minus opened, and review is merged minus first review. Take the median and the 85th percentile per team rather than the average, because a few stale pull requests distort averages badly, then sort by the longest stage. The stage where work waits longest is your constraint, and it is usually visible within an afternoon.

Then read it honestly. If pickup dominates, the fix is a review rota or smaller pull requests, not more developers. If deploy dominates, look at the release process and its approvals. If coding dominates, look at pull request size and unclear requirements. Buy a platform when you need this continuously across many teams, not to answer a question a spreadsheet has already answered.

## Comparison at a glance

| Tool | Cycle time breakdown | DORA metrics | Developer surveys | Pricing model | Published price | Best fit |
|---|---|---|---|---|---|---|
| LinearB | Coding, pickup, review, deploy | Yes, both plans | Yes | Per contributor, billed annually | Essentials 29 dollars, Enterprise 59 dollars per contributor per month | Several teams, 50 or more developers |
| Swarmia | In progress, in review, to merge, plus time to first review | Yes | Yes, or 9 dollars per developer per month alone | Per developer, billed annually | Free under 10 developers, Standard 45 dollars, Enterprise 55 dollars per developer per month | Teams changing review habits |
| Jellyfish | Refinement, work, review, deployment from Jira and Git | Developer Productivity tier | Developer Productivity tier | Seats plus modules | Not published | Large organisations tying delivery to finance |
| Atlassian (DX) | Delivery reports from system data, alongside surveys | All four, inside DX Core 4 | Yes, the core of the product | Quote after a demo | Not published | Large organisations measuring developer experience |
| Waydev | First commit to release, plus pull request insights | Yes, Pro and above | Yes, through Slack | Per active contributor, billed annually | Free Start Up for AI metrics, Pro 29 dollars, Premium 49 dollars per active contributor per month | Mid-size teams on mixed Git hosts |
| GitHub Actions performance metrics | Build queue and run time, time to first review via issue-metrics | No | No | Included in GitHub plans | Free on every GitHub Cloud plan | Slow builds, a first look at review waits |
| GitLab Value Stream Analytics | Issue, plan, code, test, review, staging | Ultimate only | No | Included in GitLab tiers | Free view, custom streams on Premium, DORA on Ultimate | Teams already on GitLab |
| Apache DevLake | Coding, pickup, review, deploy | Yes | No | Open source, self-hosted | Free, plus hosting and engineer time | Teams with an engineer to own it |

## How to pick

Under ten developers, do not buy anything yet. Run the free first pass once a quarter, use the analytics already in GitHub or GitLab, and if you want something continuous, Swarmia's free plan covers you.

From ten to about fifty developers, a per-seat tool earns its keep once someone owns acting on it. At published annual prices, 30 developers cost 16,200 dollars a year on Swarmia Standard, 8,280 dollars on its Productivity and AI impact plan and 10,440 dollars on Waydev Pro. LinearB Essentials, positioned for 50 or more developers, is 17,400 dollars a year at that size.

From fifty to a few hundred, trial LinearB and Swarmia on the same repositories over the same weeks, compare their pickup figures with your own export, and keep the one your team leads open without being asked. Above that, when finance wants software capitalisation and leadership wants AI spend explained, Jellyfish and Atlassian (DX) are built for the conversation. Choose Jellyfish when the question is where work and money go, and DX when you suspect friction that surveys catch and timestamps miss.

At every size, the platform does not remove the bottleneck. Someone still has to change the review rota, split the pull requests or fix the release process, then measure again, because the constraint moves.

## Common questions

### What software is best for finding engineering bottlenecks?

LinearB, Swarmia, Jellyfish and Atlassian (DX) are built for delivery bottlenecks, because they split the time from first commit to production into stages and show where work waits. LinearB and Swarmia expose review waits most directly, Jellyfish adds Jira planning and finance context, and DX adds developer surveys. If the code itself is slow, application performance monitoring is the right category. Before buying, time each stage from a quarter of pull request timestamps, which often settles it for free.

### What are the best tools for identifying bottlenecks in SDLC workflows?

Start with what you have. GitHub shows Actions queue time, run time and failure rate on every Cloud plan, and GitLab Value Stream Analytics shows the median time in each stage, with DORA metrics on Ultimate. Apache DevLake is a free open-source option. For continuous views across teams, LinearB, Swarmia and Waydev publish per-seat prices, while Jellyfish and Atlassian (DX) sell on quotes.

### How much does LinearB cost?

LinearB Essentials is 29 dollars per contributor per month billed annually, built for teams of 50 or more developers on GitHub Cloud. Enterprise is 59 dollars per contributor per month billed annually, aimed at 100 or more developers, and adds GitLab, Bitbucket and Azure DevOps. There is a 45-day free trial with no payment method required. Swarmia Standard, for comparison, is 45 dollars per developer per month billed annually.

### What is the best engineering analytics tool for a small team?

Under ten developers, Swarmia is the easiest start, because its free plan includes the Standard features, pull request and DORA metrics among them. The analytics in GitHub and GitLab cost nothing, and a spreadsheet of pull request timestamps answers most one-off questions in an afternoon. LinearB's entry plan is built for 50 or more developers, and Jellyfish and DX aim at large organisations.

### Is DX owned by Atlassian?

Yes. Atlassian announced the deal on 18 September 2025 and completed it on 10 November 2025, and the product now appears as Atlassian (DX). It still combines developer surveys with system data through its DX Core 4 framework, and it publishes no pricing.

### What happened to Pluralsight Flow?

Flow now belongs to Appfire, which is discontinuing it. Appfire stopped offering renewals beyond 30 June 2026 and will retire Flow on 31 December 2027, after which it cannot be bought or used. Existing customers stay supported until then, so plan the move and export your data well before the end date.

## Field notes on this[BottlenecksYour Developers Are Not the Bottleneck in Software Delivery](https://shurco.ai/insights/your-developers-are-not-the-bottleneck-in-software-delivery/)[MeasurementMeasure Queue Time Not Touch Time to Find Where Work Actually Stalls](https://shurco.ai/insights/measure-queue-time-not-touch-time-to-find-where-work-actually-stalls/)[BottlenecksPull Three Timestamps Before You Buy Any AI Tool](https://shurco.ai/insights/pull-three-timestamps-before-you-buy-any-ai-tool/)[BottlenecksYou Automated the Bottleneck and the Constraint Moved](https://shurco.ai/insights/you-automated-the-bottleneck-and-the-constraint-moved/)

## More software guides[GuideBest AI Agent and LLM Evaluation Tools in 2026Every eval platform promises to tell you whether your agent got better or worse. Here is what each one actually does, what it costs this month, and when it is the wrong buy.](https://shurco.ai/guides/best-ai-agent-evaluation-tools/)[GuideBest AI agent frameworks in 2026, compared for productionNine frameworks judged on what they do after the demo, with checked prices and a plain note on when each one is the wrong choice.](https://shurco.ai/guides/best-ai-agent-frameworks/)[GuideBest intelligent document processing software in 2026Nine options for invoices, purchase orders, delivery notes and timesheets, what each costs and when it is the wrong buy. Reading the page is now cheap, and checking it is where the work sits.](https://shurco.ai/guides/best-intelligent-document-processing-software/)[GuideBest LLM observability tools in 2026 for production AI agentsTracing, cost, prompt versions, alerts and online evaluation for AI agents in production. Nine tools alive and selling in 2026, what they cost, and the acquisitions older lists miss.](https://shurco.ai/guides/best-llm-observability-tools/)[GuideBest process mining tools in 2026, honestly comparedWhich tool will show exactly where your orders get stuck, what each one really costs, and why many mid-size companies should fix their event log before they buy any of them.](https://shurco.ai/guides/best-process-mining-tools/)[GuideBest sales order automation software in 2026Distributors, wholesalers and manufacturers still key emailed purchase orders into their ERP by hand. Here is what each tool actually automates, what it costs, and when it is the wrong buy.](https://shurco.ai/guides/best-sales-order-automation-software/)

## How this guide was checked

Every tool, plan and price on this page was checked against the vendor's own pages or reputable news on 25 September 2026. Prices change, so confirm with the vendor before you budget. No vendor paid to be here.Sources, 30 pages checked

- [LinearB pricing](https://linearb.io/pricing)
- [LinearB cycle time definition](https://linearb.helpdocs.io/article/v9pckvmkbj-cycle-time)
- [Swarmia pricing](https://www.swarmia.com/pricing/)
- [Swarmia pull request cycle time](https://help.swarmia.com/features/metrics/pull-request-cycle-time)
- [Swarmia time to first review](https://help.swarmia.com/features/metrics/pull-request-cycle-time/time-to-review)
- [Jellyfish pricing](https://jellyfish.co/pricing/)
- [Jellyfish Life Cycle Explorer](https://jellyfish.co/blog/life-cycle-explorer-remediate-bottlenecks-unlock-process-obstacles-improve-trends-throughout-the-sdlc/)
- [DX Core 4 guide](https://docs.getdx.com/dx-core-4/)
- [DX on DORA metrics](https://getdx.com/blog/dora-metrics-tools/)
- [DX SDLC analytics](https://getdx.com/sdlc-analytics/)
- [Atlassian agrees to acquire DX](https://www.atlassian.com/blog/announcements/atlassian-acquires-dx)
- [Atlassian completes DX acquisition](https://www.businesswire.com/news/home/20251110683591/en/Atlassian-Completes-Acquisition-of-DX-Advancing-Engineering-Intelligence-for-Enterprises)
- [Atlassian (DX) named a Leader](https://getdx.com/news/dx-named-a-leader-gartner-magic-quadrant-developer-productivity/)
- [Jellyfish named a Leader](https://jellyfish.co/blog/jellyfish-named-a-leader-in-gartner-magic-quadrant-for-developer-productivity-insight-platforms/)
- [LinearB named a Leader](https://www.businesswire.com/news/home/20260511848721/en/LinearB-Named-a-Leader-in-the-2026-Gartner-Magic-Quadrant-for-Developer-Productivity-Insight-Platforms)
- [Opsera named a Leader](https://www.prnewswire.com/news-releases/opsera-named-a-leader-in-the-2026-gartner-magic-quadrant-for-developer-productivity-insight-platforms-302770434.html)
- [Waydev pricing](https://waydev.co/pricing/)
- [Waydev surveys](https://docs.waydev.co/docs/developer-experience)
- [Appfire Flow end-of-life notice](https://appfire.com/flow/info)
- [Code Climate replaces Velocity](https://codeclimate.com/legacy/code-climate-quality-is-now-qlty-software)
- [Code Climate today](https://codeclimate.com/)
- [GitHub Actions metrics changelog](https://github.blog/changelog/2025-03-14-actions-performance-metrics-are-generally-available-and-enterprise-level-metrics-are-in-public-preview/)
- [GitHub Actions metrics docs](https://docs.github.com/en/enterprise-cloud@latest/organizations/collaborating-with-groups-in-organizations/viewing-github-actions-metrics-for-your-organization)
- [issue-metrics action](https://github.com/github-community-projects/issue-metrics)
- [GitLab Value Stream Analytics](https://docs.gitlab.com/user/group/value_stream_analytics/)
- [GitLab DORA metrics](https://docs.gitlab.com/user/analytics/dora_metrics/)
- [DevLake introduction](https://devlake.apache.org/docs/Overview/Introduction)
- [DevLake cycle time metric](https://devlake.apache.org/docs/Metrics/PRCycleTime)
- [DevLake pickup time metric](https://devlake.apache.org/docs/Metrics/PRPickupTime)
- [DevLake top-level status](https://news.apache.org/foundation/entry/the-apache-software-foundation-announces-new-top-level-projects-2)

## Want this in your operation

I build and run production AI agents that take repetitive work off operational teams. [Tell me what your team spends too long on.](https://shurco.ai/#contact)

Web version https://shurco.ai/guides/linearb-vs-swarmia-vs-jellyfish-vs-dx/