Learning paths
The hub holds the strategy. The Academy teaches you why each part of it is there.
Seven paths through quality assurance, in the order the argument runs: understand why quality is an economic property, move the accountability rather than just the tests, choose the typology that answers your actual question, design the topology it runs in, write tests worth keeping — and only then take the option that requires the most prerequisites.
In preparation.The curriculum below is settled; the lessons are still being written, and there is nothing to enrol in yet. Every path links to the documentation that covers its ground today.
Told as stories
Testing is taught badly as a taxonomy and well as a sequence of decisions someone had to live with. So the lessons carrying astorymark come with a short video in narrative form: a team with a problem, the option they took, what it bought them and what it cost. 16 of the 37lessons are planned that way — the rest are short enough to read.
The curriculum
- FoundationComing soon
Quality-driven development
Why quality is an economic property rather than a moral one, and what changes when a team treats it as the thing that decides the cost of the next change instead of the thing that delays this release.
- The cost curve: what a defect costs at each stage it survives Includes a storytelling video
- Quality attributes are requirements, and unowned ones are wishes
- Definition of done, and why "tested" is not on the list
- The quality gate that made things worse Includes a storytelling video
- Reading a defect as feedback about the process, not the person
Read today: Quality-driven development
- FoundationComing soon
Shift-left accountability
Moving the tests earlier is the easy half and the useless half. This path is about moving the answerability: who is asked, when, and what they are expected to have already done.
- Shift-left testing versus shift-left accountability Includes a storytelling video
- The three amigos, from the tester’s seat
- Acceptance criteria that a machine can disagree with
- What a test manager does when the teams own the tests Includes a storytelling video
- The handover that never happened and the bug that survived it Includes a storytelling video
Read today: Shift-left accountability
- CoreComing soon
Choosing a typology
Nine kinds of test, each answering a question the others cannot. The skill is not knowing the definitions — it is knowing which question you actually have, and refusing to buy the answer twice.
- The nine typologies and the question each one answers Includes a storytelling video
- Unit and component: the design feedback, not the safety net
- Integration versus contract: what each one can and cannot catch Includes a storytelling video
- End-to-end: the most expensive test you should own the fewest of
- Non-functional typologies: performance, resilience, security, accessibility
- Exploratory testing as a discipline rather than a lack of one Includes a storytelling video
Read today: Typologies
- CoreComing soon
Designing a topology
Where the tests run and against what. Suite shapes, environment strategy, test data, and the honest trade between a fast signal and a trustworthy one.
- Pyramid, trophy, honeycomb — and what they are all arguing about Includes a storytelling video
- Ephemeral environments, and what they cost when they are not
- Test data: synthesised, seeded, subsetted, or masked
- Testcontainers and the end of the shared integration environment Includes a storytelling video
- When the environment is the thing under test
Read today: Topologies
- AppliedComing soon
Writing tests worth keeping
A test is code that outlives the code it tests. This path is about the properties that decide whether it earns its maintenance: intent, determinism, isolation, and a failure message that names the cause.
- A test name is a specification, so write it as one Includes a storytelling video
- Arrange–act–assert, and the assertion that tests three things
- Determinism: time, randomness, ordering, concurrency and the network
- Flakiness has a cause, and quarantine is not a fix Includes a storytelling video
- Coverage measures what ran, mutation measures what mattered
- The failure message as the first line of the investigation
Read today: Guidelines
- AppliedComing soon
Performance as a first-class test
Load, stress, soak and spike, driven by service-level objectives rather than by round numbers — and written as code that a reviewer can read, which is what the DSL research is chasing.
- From SLO to workload model: what to simulate and why Includes a storytelling video
- The four shapes: load, stress, soak, spike
- Reading a percentile without lying to yourself Includes a storytelling video
- Why performance scripts rot faster than any other test
- A DSL for workloads: the case, the survey, the sketch
Read today: Research
- AdvancedComing soon
Testing in production
The option, with its entry conditions. Synthetic monitoring, canaries, feature flags, dark launches and controlled experiments — and the observability and rollback you have to have first.
- What only production can tell you Includes a storytelling video
- The prerequisites: observability, rollback, blast radius, consent
- Synthetic journeys and the alert that means something
- Canary analysis: deciding automatically, with a rule agreed in advance
- The dark launch that found the bug staging never could Includes a storytelling video
Read today: Testing in production
You do not have to wait
Everything the first lessons will teach is already written down, and the fastest way to see how it fits together is to take one campaign this week and write down its scope, its owner and its exit condition.