← FinOpsAid blog
What is GitHub storage even charging us for? Reading the Storage page
Net cost, month-over-month change, category splits and the SKU table — every figure on the FinOpsAid Storage page, and why this is the most trustworthy page in the product.
Storage is the page with the least explaining to do, because almost everything on it is billed. No modeling, no allocation rule, no attribution caveat — GitHub reports storage cost by category and SKU, and the page shows you that. If you want a number to take to finance without a footnote, it's here.
What the page can't do is tell you why the number is what it is. Storage has no owner in most organisations, and its cost is the accumulated residue of decisions nobody remembers making.
The KPI row
| Tile |
What it counts |
Direction |
| Net Storage Cost |
Total net storage spend in scope, after discount. Billed. |
Worse. |
| This Month |
Current-month net, with a month-over-month delta chip. |
Worse. |
| Previous Month |
Last full month's net — the comparison baseline for the tile beside it. |
— |
| Categories |
How many distinct storage categories carry spend. |
Neutral, but see below. |
The Categories count is more useful than it looks. Storage on GitHub isn't one thing — it's Packages, Actions artefacts, Actions cache, LFS, Codespaces disks and more, each with its own retention behaviour and its own owner. A high category count means the cost is diffuse, and diffuse cost needs a different approach from concentrated cost.
This Month against Previous Month is a genuine like-for-like comparison, which is rarer than it sounds. Storage bills for existence rather than activity, so month-over-month movement is a real signal: it means something was created and not cleaned up, or something was finally deleted.
Cost by category
A donut with the net total in the centre, alongside a legend listing every category with its amount. This is where you learn what your storage bill actually is.
Read the donut as an ownership map, not just a cost split. Every category is created by a different process and fixed by a different team, and none of them will notice on their own — storage bills silently forever.
Storage spend over time
A daily area chart of net storage cost over the last 30 days.
Storage series behave differently from compute series and should be read differently. Compute is spiky because it tracks activity. Storage is a staircase: flat stretches punctuated by step changes when something large is created or deleted. A smooth upward drift means accumulation with no cleanup. A sudden step down means someone actually deleted something — worth finding out who and what, because that's your repeatable saving.
By SKU
The detail table: Product, SKU, Category (as a chip), Net — top 20 by net cost, sorted descending.
SKU is finer-grained than category, and it's where the answer usually is. "Packages" as a category tells you the family; the SKU tells you which package registry and which storage tier. When you need to hand a finding to a specific team, the SKU row is the evidence.
What this page doesn't tell you
There's no per-repository split. GitHub reports storage at the org level by category and SKU, not by repository — the same limitation as Copilot, for the same reason. The one exception is Actions cache storage, which does have a per-repository breakdown, and which lives on the GitHub Actions page because a workflow configuration is what created it.
There's no age dimension. You can see what storage costs, not how old it is. A 90-day-old artefact and a 900-day-old one bill identically and look identical here.
There's no "is this needed" signal. That judgement lives with whoever created the thing, which is exactly why storage accumulates.
What people get wrong on this page
- Treating storage as too small to matter. It's small per month and permanent, which makes it large per year. It also never stops on its own.
- Reading the month-over-month delta as usage. Storage bills for existence. A rise means things were created and kept.
- Trying to allocate storage per repo. Only Actions cache splits that way.
- Ignoring a flat line. Flat means nothing is being cleaned up either.
- Stopping at category. The SKU table is where the actionable detail is.
The 60-second read
- Net Storage Cost and the month-over-month chip — is it drifting up?
- Donut: largest category, and therefore which team owns the conversation.
- Time chart shape: staircase up with no steps down means no cleanup process exists.
- SKU table top rows — that's the evidence to attach to the ask.
- Cross-check Actions cache on the Actions page; it's the one storage line with a per-repo owner.
Frequently asked questions
Is GitHub storage cost billed or estimated?
Billed. GitHub reports storage cost directly by category and SKU, with no modeling, allocation rule or attribution caveat involved. That makes the Storage page one of the safest sources in FinOpsAid to quote to finance without a footnote.
Can I see storage cost per repository?
Not in general. GitHub reports storage at organization level by category and SKU, not by repository. The single exception is Actions cache storage, which does have a per-repository breakdown and appears on the GitHub Actions page because a workflow configuration created it.
Why does our storage bill keep growing?
Storage bills for existence rather than activity, so it only falls when someone deletes something — and nothing in GitHub prompts anyone to. A smooth upward drift on the daily chart means accumulation with no cleanup process; a step down means somebody actually removed data.
Want to see what your storage is actually made of? Connect your GitHub org — read-only and free during beta — or explore the demo first. There's a longer piece on the causes in why is GitHub storage on the bill. Next: how to read the Copilot page.