3 pages in this topic
SaaS Growth Experimentation Course
A three lesson course on running experiments that survive scrutiny: find the constraint, design a readable test, and decide what to scale, with worked examples.
On this page 7 sections
The short answer
This three lesson course teaches B2B SaaS growth teams to run experiments at low traffic, where classic A/B testing advice breaks down. Lesson one locates the actual constraint in the funnel. Lesson two designs a test that can be read at your sample size, or rules it out. Lesson three sets the decision rule before results arrive. Total time is about 90 minutes, and the materials include an experiment brief, a sample size calculator and a decision rule card.
Key points before you start
Fair warning before you enrol: this course will tell you not to run most of the tests currently on your backlog. Not because the ideas are bad, but because at your traffic level the result would be unreadable and you would ship it anyway based on a 9% lift that was noise. That is the problem the three lessons are built around.
Who should take this
Growth marketers and product managers at B2B SaaS companies with limited traffic. Call it 2,000 to 50,000 sessions a month, a few hundred signups, and a boss who has started asking about experiment velocity.
If you run a consumer product with millions of sessions, skip it. Standard experimentation practice works fine when you can detect a 2% lift in four days. The whole design of this course comes from the opposite situation, where a homepage test needs eleven weeks to resolve and the homepage will have changed twice by then. The wider context on where experiments fit sits in SaaS Growth Marketing.
The uncomfortable premise
Most B2B SaaS experimentation programs are theatre. They report tests run, not decisions changed. If your last quarterly review counted experiments rather than naming what you stopped doing because of one, this course is aimed at you specifically.
What you will be able to do after each lesson
| Lesson | Time | Capability you gain | Material |
|---|---|---|---|
| 1. Find the constraint | 30 min | Locate the one funnel step where a 10% improvement changes revenue, and ignore the rest | Constraint worksheet |
| 2. Design a readable test | 35 min | Calculate minimum detectable effect at your traffic and kill tests that cannot resolve | Sample size calculator |
| 3. Decide and scale | 25 min | Write the decision rule before launch and run the post test meeting in ten minutes | Decision rule card |
Lesson one is funnel arithmetic, not brainstorming. You will compute the revenue impact of a 10% improvement at each step, and in almost every case one step dominates so heavily that everything else becomes a distraction. Teams routinely discover their constraint is trial to paid, then realise every test they ran last quarter was on the signup form.
- Lesson two is where the backlog gets cut. You enter your weekly traffic and baseline conversion rate, and the calculator returns the smallest effect you could actually detect in four weeks. If that number is 35%, no button copy change qualifies, and you need a structurally different test: a new offer, a removed step, a changed pricing display. Building a growth experimentation program goes deeper on how to build a backlog that survives this filter.
Lesson three is the part that saves relationships. The decision rule gets written into the brief before launch: what result ships it, what result kills it, what result means run it longer. Without that, the post test meeting becomes a negotiation between whoever wants the feature and whoever wants to be right.
Self-paced learning
Build your SaaS marketing study plan
Choose a free course and work through its published lessons at your own pace. Save the course index for later.
The materials
Three files, all editable, all delivered by email as you work through the lessons.
- Experiment brief template. Hypothesis, primary metric, guardrail metric, minimum detectable effect, decision rule, owner, end date. The growth experiment brief template is the same file, available separately if you only want that.
- Sample size calculator. Input weekly traffic and baseline rate, output minimum detectable effect and required run length.
- Decision rule card. One page, printed or pinned in Slack, that the post test meeting follows.
40 to 60%
Share of a typical B2B SaaS experiment backlog that fails the power check in lesson two
Aggregated practitioner reports, saas-marketing.net estimate
How this differs from generic CRO training
The core assumption. Standard CRO material teaches you to test constantly and let volume find the winners, which is correct advice at consumer scale and actively harmful at 8,000 sessions a month.
Here the advice inverts. Run fewer, bigger tests. Accept that some decisions will be made on directional evidence and label them as such in writing. Use qualitative research, session replay and sales call review to generate hypotheses, because at your traffic the hypothesis quality matters more than the test count. Tooling choices follow from that, and A/B testing tools for SaaS compares what PostHog, Amplitude and the rest actually give a low traffic team.
Velocity as a vanity metric
Experiment count is the easiest growth number to report and the least informative. A team running 14 tests a quarter, none powered to detect less than a 30% lift, has generated 14 coin flips and a lot of engineering tickets. Published velocity figures are worth reading as context, not as targets, which is how experiment velocity benchmarks frames them.
What the course does not cover
Statistical theory beyond what you need to act. There is no derivation of the t test, no Bayesian versus frequentist debate, and no discussion of multi armed bandits, which are genuinely useful and genuinely irrelevant until you have enough traffic to feed them.
It also does not cover SEO testing, which behaves differently because the unit is the page and the feedback loop runs in months. SaaS SEO Sprint handles that. And it assumes you already know your funnel definitions; if you do not, take the SaaS marketing foundations course first.
Editable working copy
Download this template
Save an editable working copy of the framework on this page. Add your own owners, evidence and decisions.
Time commitment and format
About 90 minutes of reading and worksheet time, delivered one lesson per day over three days. Every lesson page is fully readable on the site without an email address. The sign up exists to pace the lessons and send the templates, not to gate content.
Realistically, add two to three hours in week one to actually compute your funnel constraint with real data, because pulling clean numbers out of Amplitude or your warehouse is always slower than the arithmetic itself.
How to get value from it
- Pull four weeks of funnel data first
Sessions, signups, activated accounts, paid conversions. You know you have enough when you can compute each step rate without guessing.
- Do lesson one with that data open
The constraint should be obvious once the revenue impact column is filled in.
- Run your backlog through lesson two
Expect to cancel roughly half of it. Write down what you cancelled and why.
- Rewrite two surviving tests as bigger swings
If a test cannot detect less than a 25% lift, the change itself has to be big enough to plausibly produce one.
- Write the decision rule before launch
Get the person who will argue about the result to sign it in advance.
- Hold the post test meeting to ten minutes
If it runs longer, the decision rule was vague.
Start here
Begin with lesson one, or if experimentation is one part of a broader growth remit, read B2B SaaS growth for where it sits alongside channel work. For the tactics that get mislabelled as experimentation, growth hacking for SaaS is the honest version.
Pull your funnel numbers before you start lesson one. The course is close to useless without them.
Editable CSV worksheet
SaaS Growth Marketing planning worksheet
A practical growth planning worksheet: decisions, owners, evidence and next actions.
Frequently asked questions
Who is this growth experimentation course for?
Growth marketers, product managers and founders at B2B SaaS companies with real but limited traffic, roughly 2,000 to 50,000 monthly sessions. If you have 500,000 sessions a month, standard CRO training applies and you do not need this. If you have 300, the course will tell you to run qualitative research instead and that is a legitimate answer.
Do I need statistics knowledge to take it?
No. The sample size calculator does the arithmetic and lesson two explains what the output means in plain terms. You need to be able to read a conversion rate and hold the idea that a 12% lift on 40 conversions is not evidence of anything. That is the entire prerequisite.
How is this different from generic CRO training?
Generic CRO assumes you can reach significance in two weeks and spends its time on button colours and heuristics. This course assumes you cannot, so it prioritises picking bigger swings, running fewer tests, using sequential decision rules, and accepting directional evidence with a documented confidence level rather than pretending to certainty.
What materials come with the course?
Three files: an experiment brief template with the hypothesis, metric, guardrail and decision rule fields; a sample size calculator that tells you the minimum detectable effect at your traffic; and a one page decision rule card for the post test meeting. They arrive by email alongside the lessons.
Is the course free?
Yes. All three lesson pages are open and readable without a form. The email sign up delivers one lesson a day plus the three templates. Nothing is gated, and there is no upsell to a paid cohort at the end.
How many experiments should a B2B SaaS team run per month?
Fewer than you think. Teams under 20,000 monthly sessions usually get more value from two well powered tests a quarter than from eight underpowered ones a month. Velocity benchmarks are popular because they are easy to report, but a high count of unreadable tests is negative value once you count the engineering hours.
The saas-marketing.net editorial team Research and editorial
We research, write and maintain every page on this site. The library explains marketing decisions through practical frameworks, explicit assumptions and references. Corrections can be requested through the contact page.
Published September 11, 2026. Last updated .
Lessons
Work through them in order. Each one ends with an exercise.
- LessonFind the constraint in the growth modelChoose the customer transition that limits retained value before building an experiment backlog. Complete the exercise, produce a working artifact and review it against a practical rubric.
- LessonDesign a test that can answer the questionDefine assignment, outcome, sample requirements and guardrails before changing the customer experience. Complete the exercise, produce a working artifact and review it against a practical rubric.
- LessonDecide from the evidence and preserve the learningInterpret the result against the original question, then choose a bounded next action. Complete the exercise, produce a working artifact and review it against a practical rubric.
Related topics
Growth experiment velocity: measure learning capacity
Experiment velocity is the rate at which a team completes interpretable tests, not simply the number of variants it launches. Review definitions, sampling choices and common comparison errors.
Growth experiment brief template
Specify the hypothesis, assignment, outcome and decision rule before a growth experiment starts. Use the editable worksheet, example and review steps to turn it into a working document.
Building a Growth Experimentation Program
Set up an experimentation program that ships weekly: idea intake, prioritisation, the review meeting, decision rules and the documentation that stops repeats.
Evaluate SaaS experimentation tools
An experimentation tool must support the unit of assignment, outcome timing and operating controls your product needs. Compare practical options, review the tradeoffs and choose a relevant next step.
A/B test sample size calculator
Plan an approximate sample for a two-sided conversion-rate test at 5% significance and 80% power with equal allocation. See the formula, change the inputs and save your results.
Design a test that can answer the question
Define assignment, outcome, sample requirements and guardrails before changing the customer experience. Complete the exercise, produce a working artifact and review it against a practical rubric.