# How do I see which repositories and workflows use the most GitHub Actions minutes?

Metered usage gives the total by SKU; the detailed usage report gives repository and workflow\_path. Where each is, what it holds, and how to rank workflows.

By Grant Unwin, published 2026-11-10. Checked against GitHub's documentation on 2026-11-17. The web page: https://www.runners.io/blog/which-github-actions-workflows-use-the-most-minutes

Open Billing & Licensing, then Metered usage, for the total by SKU. For the repository and the workflow, request the detailed usage report, which covers up to 31 days and has repository and workflow\_path columns, and add up quantity for each pair. GitHub’s Actions usage metrics show minutes by workflow, job, repository, operating system and runner type in the browser, with an optional CSV download, but they apply no minute multipliers. Our [GitHub Actions cost](https://www.runners.io/blog/github-actions-pricing-2026) guide has the rates.

**In short**

- Billing & Licensing, then Metered usage, gives the total by SKU; the example on GitHub's page is Actions grouped by SKU.
- The detailed usage report covers up to 31 days and has repository and workflow\_path columns; the summarised report has no workflow\_path.
- workflow\_name was removed from the report and replaced by workflow\_path.
- Actions usage metrics show minutes by workflow, job, repository, OS and runner type, but apply no minute multipliers.
- Rank workflows by adding up quantity for each repository and workflow\_path, and compare the total with Metered usage.

## Where can I see minutes by repository and workflow?

Three places, each answering a different part of the question. Read from GitHub’s own pages on the date this post was last checked.

| Where | What it shows | What to know |
| --- | --- | --- |
| Billing & Licensing, Metered usage | Usage of every metered product, including Actions. The page's own example is Actions grouped by SKU. | Who can see it depends on your role and plan; owners and billing managers of an organisation or enterprise can. The page names SKU as its example and says the grouping options vary with the filters you set. |
| Insights, Actions Usage Metrics | Minutes by workflow, job, repository, runtime OS and runner type, for a period you choose. Can be downloaded as CSV. | It applies no minute multipliers, so it will not match the bill. Days are in UTC. A custom range spans up to 100 days. |
| The detailed usage report | Usage by date, SKU, organisation, repository, cost centre, user and workflow path, for up to 31 days. | Emailed to you, the link expires after 24 hours, and only one can be requested at a time. It is available only through the web interface, not the REST usage endpoint. |

## What is in the detailed usage report?

Fourteen columns: `date`, `product`, `sku`, `quantity`, `unit_type`, `applied_cost_per_quantity`, `gross_amount`, `discount_amount`, `net_amount`, `username`, `organization`, `repository`, `workflow_path` and `cost_center_name`. The one that names the workflow is `workflow_path`, which GitHub describes as the path of the Actions workflow that generated the usage, if applicable. The summarised report leaves out `username` and `workflow_path`, so it cannot answer this question. GitHub’s page also says `workflow_name` has been removed and replaced by `workflow_path`, so a spreadsheet or a script written for the old name needs the new one.

## How do I rank the workflows from the report?

Sum `quantity` for each repository and workflow path, for the rows of one product and one unit. Read the distinct values of `product` and `unit_type` in your own file first, because the page does not list them and the sample below, which is made up, uses the ones we guessed. Then compare the total with the Metered usage page: GitHub does not say here whether `quantity` carries a minute multiplier, and the two should agree before you trust a ranking.

```text
import csv, sys
from collections import defaultdict

path, product, unit = sys.argv[1:4]
totals = defaultdict(float)
with open(path, newline="") as f:
    for row in csv.DictReader(f):
        if row["product"] == product and row["unit_type"] == unit:
            totals[(row["repository"], row["workflow_path"])] += float(row["quantity"])
for (repo, workflow), amount in sorted(totals.items(), key=lambda kv: -kv[1]):
    print(f"{amount:6.0f}  {repo}  {workflow}")
```

```bash
python3 -I top_minutes.py usage.csv actions minutes
#    500  acme/web  .github/workflows/e2e.yml
#    120  acme/api  .github/workflows/ci.yml
#     50  acme/api  .github/workflows/win.yml
```

## Why do the usage metrics and my bill not match?

Four differences, all on GitHub’s pages. The usage metrics apply no minute multipliers, and GitHub says jobs on its Windows and macOS runners cost more to run than jobs on Linux, which is our reading of why a Windows minute and a Linux minute count the same there and not on the bill. Skipped and zero-minute runs are left out. Days are cut in UTC. And the billing dashboard may show Actions as a dollar amount instead of minutes. The [GitHub Actions pricing post](https://www.runners.io/blog/github-actions-pricing-2026) has the rates.

## How do I see it for jobs that run on runners.io?

**Usage** in the dashboard groups by repository by default and, for the month in progress, by machine size; a closed month shows its invoice’s share per repository. **Download CSV** on that page exports the month line by line for owners, admins and anyone with the Billing role. Our [billing page](https://www.runners.io/docs/billing/how-you-are-billed) has the mechanism, and the [calculator](https://www.runners.io/calculator) estimates a month before it happens. Our per-minute rate for every size, beside GitHub’s, is on the [GitHub Actions runner pricing](https://www.runners.io/pricing) page.

Through an agent, `list_repositories` gives each shared repository with how many jobs it has run and when the last one was, and `list_jobs` narrows recent jobs by repository. `get_spend` is per runner size, not per repository, and says what its sum leaves out; it needs a role that can see billing. See [the MCP servers for GitHub Actions](https://www.runners.io/blog/mcp-server-for-github-actions) to connect one, and [tracking CI metrics](https://www.runners.io/blog/track-ci-metrics-github-actions) for the wait and run times behind the minutes.

## Sources

- [Viewing your usage of metered products and licenses (GitHub Docs)](https://docs.github.com/en/billing/how-tos/products/view-productlicense-use)
- [Billing reports reference (GitHub Docs)](https://docs.github.com/en/billing/reference/billing-reports)
- [Viewing GitHub Actions metrics for your organization (GitHub Docs)](https://docs.github.com/en/organizations/collaborating-with-groups-in-organizations/viewing-github-actions-metrics-for-your-organization)
- [About GitHub Actions metrics (GitHub Docs)](https://docs.github.com/en/actions/concepts/metrics)
- [GitHub Actions billing (GitHub Docs)](https://docs.github.com/en/billing/concepts/product-billing/github-actions)
- [How you are billed (runners.io documentation)](https://www.runners.io/docs/billing/how-you-are-billed)

## Frequently asked questions

### How do I see which repository uses the most Actions minutes?

Request the detailed usage report from the Metered usage page, then add up quantity by repository. Actions usage metrics under Insights also show minutes by repository, workflow, job, operating system and runner type, and can be downloaded as CSV.

### Which column in the usage report names the workflow?

workflow\_path, which GitHub describes as the path of the Actions workflow that generated the usage, if applicable. The summarised report leaves it out, and the old workflow\_name column was removed and replaced by it.

### How long a period can the detailed usage report cover?

Up to 31 days. It is emailed to you, the link expires after 24 hours, and only one report can be requested at a time. The summarised report can cover up to a year.

### Why do Actions usage metrics not match my bill?

The metrics apply no minute multipliers, and GitHub says jobs on its Windows and macOS runners cost more than Linux. They also leave out skipped and zero-minute runs and cut days in UTC.
