Quick Start

Step-by-step guide for your first launch of AB-Labz Workbench.

Step 1: Registration and Login

After registration, you land on the main page — Dashboard. Here you can see:

• Active experiments — tests that are running right now
• Upcoming experiments — scheduled tests
• Ready for analysis — completed experiments awaiting data analysis

If you're just starting, the dashboard will be empty — that's normal, now let's create your first hypothesis.

Step 2: Create a Hypothesis

Navigate to the Hypotheses section via the sidebar menu. Click the "Add Hypothesis" button and fill out the form:

• Name: for example, "New CTA button on homepage"
• Hypothesis: "If we add a bright CTA button, registration conversion will increase"
• Key metric: select from the dropdown (metrics are created in the metrics catalog)
• Expected change: for example, 10
• External ID (optional): experiment identifier from your split system (used in API)
• Tags (optional): for grouping hypotheses

Click "Save" — the hypothesis will appear in the list.

Step 3: Calculate Sample Size

Navigate to Tools → Sample Size Calculator. Enter metric parameters and desired statistical power.

The tool will show how many users are needed in each group. This will help plan the experiment duration.

Step 4: Schedule the Experiment

Open the hypothesis and click the "Schedule" button. Specify the start and end dates of the experiment (based on sample size calculation). After saving, the experiment will appear in the Calendar, where you can track all scheduled tests.

Step 5: Run the Experiment

Implement the change in your product, divide users into groups (control/treatment), and collect data. AB-Labz Workbench does not manage splitting — this is done in your system.

Step 6: Upload Data and Analyze

When the experiment is complete, prepare a dataset for analysis:

user_id,group,registration_completed
user_1,control,0
user_2,control,1
user_3,treatment,1

Navigate to the Data Preparation section and upload data:

• CSV file — upload a file directly
• API — get data from storage

After upload, run the analysis. The system will calculate and display statistics: metric change, p-value, confidence intervals.

Step 7: Save Results

After analysis, click the "Save Results" button on the analysis page. In the modal window:

• Select the hypothesis this experiment relates to
• Or create a new hypothesis directly from results
• Results will be automatically saved to the selected hypothesis

Important: If you analyzed data with segmentation (multiple calculations for different segments), save each result separately to one hypothesis. After saving, the hypothesis will show toggles to switch between segment results.

The hypothesis with saved results will automatically move to the Experiments Archive — this way you accumulate knowledge about what works in your product.