Even if a coding agent builds your test dashboard for free, its infrastructure may cost more than using Flakiness.io.
These days, agents and LLMs can generate (and do generate) enormous test suites for small teams. At some point, debugging test failures via CI logs stops being pleasant, so it is tempting to ask those same agents to build an internal test analytics dashboard.
The strongest argument for building rather than buying has always been cost: “Building is cheaper than buying.” And coding agents only strengthen this argument further.
But even if an agent writes the code for free, the dashboard’s infrastructure alone might cost more than a subscription to a purpose-built service.
That’s the case with Flakiness.io.
Let’s estimate the storage cost of running an in-house dashboard.
A conventional test analytics system keeps test results in a database. Allure TestOps, a mature and widely used test management platform, uses PostgreSQL as its primary data store. Its self-hosting documentation provides the following estimate:
Allure TestOps relies on PostgreSQL as the main data storage.
The required disk space depends on the amount of test results you process. A project with
100,000tests can take about300 GBin PostgreSQL after one year of daily runs.
That is about 36 million test results: 100,000 tests, run daily, for a year. Using Supabase’s pricing (as of August 2026), the database alone would cost us ~$60/month.
By comparison, our measurements suggest that roughly 40 million test results, excluding artifacts, fit within Flakiness.io’s free 1 GB plan.
Thirty-six million test results may sound like a lot, but many projects far exceed this testing volume. For example, WordPress’s Gutenberg generates more than that in a single week - roughly two billion a year: WordPress/Gutenberg Dashboard.
Storing these results in Supabase’s PostgreSQL would cost roughly $1,900/month versus $49/month on Flakiness.io.
How did we calculate this estimate?
Flakiness.io began collecting Gutenberg’s results on June 22. By August 17, 56 days later, it had received 368,546,634 results, or roughly 46 million per week.
To keep the comparison conservative, we use 36 million results per week:
- Annual volume: 36 million × 52 = 1.872 billion results.
- PostgreSQL storage: Allure TestOps estimates 300 GB for 36.5 million annual results. At the same density, 1.872 billion results require roughly 15.4 TB.
- Supabase: Its Pro plan includes 8 GB, with additional storage priced at $0.125 per GB. That works out to roughly $1,947 per month.
- Flakiness.io: At our observed density of roughly 40 million results per GB, 1.872 billion results require about 47 GB, fitting within the $49 plan.
The Flakiness.io estimate excludes test artifacts.
Flakiness.io can do this because its purpose-built analytics engine stores and processes testing data unusually efficiently. Lower infrastructure costs allow us to pass those savings directly to customers.
So don’t build the same dashboard again. Try ours instead.
Flakiness.io Test analytics for GitHub and GitLabFlakiness.io Team