Multi-Page User Journey (Tutorial)

In this tutorial, you'll create a multi-step automation that navigates across several pages — simulating a real user journey like browsing a store, adding items to a cart, and proceeding to checkout. You'll learn how to monitor progress and review results for complex flows.

Prerequisites

• Donobu Studio installed and running (download)
• At least one LLM provider key configured in Studio settings
• Completed the Your First Automation tutorial (recommended)

Step 1: Plan the user journey

For this tutorial, we'll automate an e-commerce browsing flow. The AI will:

1. Navigate to an online store
2. Search for a product
3. Select a product from the results
4. Add it to the cart
5. Proceed to the checkout page

This involves multiple page navigations and interactions across different parts of the site.

Step 2: Create the flow

1. Open the New Flow screen in Studio.
2. Enter the URL of an e-commerce site (use a demo store, such as https://www.saucedemo.com/, or a site you have permission to test).
3. Write a detailed objective that describes the full journey:

   Search for a fleece jacket, add it to the cart, and check out. Assert that the order confirmation screen appears with no errors.

4. Under Options, increase the Max Actions to 75 (the default may be too low for a multi-step journey).

Step 3: Monitor the flow in progress

Click Start Automation and watch the flow execute:

• The Timeline updates in real time as each action is completed.
• You can see the browser navigating between pages.
• Each step shows what the AI decided to do and why.

For multi-page journeys, the AI may take exploratory steps — trying different approaches to find the right button or link. This is normal, especially on the first run.

Step 4: Review the step-by-step results

Once the flow completes, open the detail view:

1. Scroll through the timeline — each action is a separate step, with a description like "Adding the Sauce Labs Fleece Jacket to the cart as per the objective." or "Clicking 'Continue' to proceed to the order overview."
2. Compare screenshots — click through the screenshots to see the visual state at each step. This is especially useful for verifying that the AI navigated to the correct pages.
3. Check for unexpected steps — if the AI took a wrong turn (e.g. clicked the wrong product), you'll see it in the timeline. This helps you refine the objective for future runs.

Step 5: Rerun for regression testing

Once you have a successful multi-page flow:

1. Click Execute to replay the exact same action sequence.
2. The rerun executes in Deterministic mode — no AI calls, just replaying the recorded steps.
3. If the site has changed (e.g. the product was removed or a button was renamed), the deterministic rerun will fail at the point where the page no longer matches.

When a rerun fails, you can create a new flow with the same objective to let the AI adapt to the changes.

Tips for multi-page flows

• Be specific in the objective — "Select the first product" is better than "select a product" because it gives the AI a clear target.
• But don't be too prescriptive - A specific, step-by-step objective might be more predictable, but might be more difficult to self-heal if the website changes. For example, "Add the first product to the cart and check out" is better than "1. Click the cart button in the top-right of the screen ..." since the locations and names of buttons may change overe time, or there may be intermediate steps that need to be navigated.
• Set a reasonable max tool calls — Multi-page journeys typically need 15-30 actions, though some long and complex ones can take more than 50. Start with the default (50) and adjust based on results.
• Break very long journeys into parts — If a journey has 10+ distinct steps, consider splitting it into 2-3 shorter flows (e.g. one for browsing, one for checkout). This makes debugging easier.
• Use the video — For complex flows, the video recording gives the best end-to-end view of what happened.

What you learned

• Multi-page flows can navigate across different parts of a website by describing the full journey in the objective.
• Max tool calls controls how many actions the AI can take — increase it for longer journeys.
• Deterministic reruns replay the exact steps from a successful flow for fast regression testing.
• Screenshots and video provide visual verification of the entire journey.

What to try next

• Try the Donobu Playwright Extension tutorials to write the same kind of journey as a Donobu-enhanced Playwright test.
• Learn about Logged-In Browser State to automate flows that require authentication.