Skip to main content

Mobilize AI: Building Complex Branching Logic Workflows

This guide explains how to create workflows with multiple levels of branching logic where users make sequential selections that determine which questions they see.

Lauren Baird avatar
Written by Lauren Baird
Updated today

Step 1: Format Your Questions (Excel, Word, or Text Editor)

What You'll Do: Create a properly structured document with your workflow data using your preferred tool.

Actions:

  1. Open Excel/Word Text Editor and create a new document

  2. Set up your levels based on your branching needs:

    • Level A, B, C, etc.: [Your branching category headers]

    • Final Level: [Your questions]

  3. Enter your data following this format example:
    Recommended Indented Format:

    LOCATION: Building 1
    DEPARTMENT: Safety
    SHIFT: Morning
    - Are fire exits clear and unobstructed?
    - Is emergency equipment easily accessible?
    - Are safety signs visible and legible?
    SHIFT: Evening
    - Are fire exits clear and unobstructed?
    - Is emergency lighting functional?
    DEPARTMENT: Maintenance
    SHIFT: Morning
    - Are tools properly stored and secured?
    - Is equipment clean and operational?
    LOCATION: Building 2
    DEPARTMENT: Safety
    SHIFT: Morning
    - Are fire exits clear and unobstructed?
    - Is emergency equipment easily accessible?

    Alternative Bracket Format:

    [Building 1] [Safety] [Morning]
    - Are fire exits clear and unobstructed?
    - Is emergency equipment easily accessible?

    [Building 1] [Safety] [Evening]
    - Are fire exits clear and unobstructed?
    - Is emergency lighting functional?

    Formatting Guidelines:

Next Steps:

  1. Ensure consistency - use exact same spelling for your categories throughout

  2. One question per row/line - repeat category combinations as needed

  3. Review for completeness - ensure all category combinations have their associated questions

  4. Save your file with a clear filename for documentation purposes

Quality Checklist:

  • Consistent spelling across all entries

  • Complete questions in final column

  • Clear, descriptive headers

Step 2: Access Mobilize AI

What You'll Do: Navigate to the workflow creation area in Mobilize AI.

Actions:

  1. Log into ANVL Insights (https://insights.anvl.com/)

  2. Open Mobilize AI

Step 3: Provide the Detailed Prompt

What You'll Do: Give Mobilize AI specific instructions using the standardized prompt.

Actions:


Copy this prompt into Mobilize AI (edit as needed)


Create a branching logic workflow using the questions below. Follow these requirements exactly:

  1. Use Exact Language: Use the questions exactly as written below - do not modify, paraphrase, or reword any questions.

  2. Branching Structure:

    • Create branching levels in the order of my columns (left to right)

    • Each header (except the final questions) becomes a branching level

    • Show questions only after ALL selections are made

  3. Required Tags: Add a "reportName" tag to every single question for reporting purposes.

  4. Navigation Logic: Users must select in the order of my headers:

    • First Question: Select from Heder A options

    • Second Question(s) (one per response option from first question): Select from Header B options

    • Continue: Through all remaining columns in sequence

    • Finally: Display questions that match ALL selections in sequential order as provided

  5. Intervention Logic:

    • For questions where "No" or other undesirable responses indicate issues, automatically add a follow-up question asking, "What action will you take to address this?" or similar intervention prompt

    • Ensure follow-up questions only appear when the triggering response is selected

Additional Requirements:

  • Preserve exact question wording from the questions column

  • Ensure proper filtering based on all branching column selections

  • Include reportName tags for all questions

  • Test that users see only relevant questions for their complete selection path

  • Handle the correct number of branching levels based on my column structure

[ADD YOUR QUESTIONS HERE]


  1. Review the prompt to ensure it is complete

  2. Send the prompt to Mobilize AI along with your uploaded file

  3. Review the responses from Mobilize and add detail or clarity if needed
    IMPORTANT: Ensure you pasted in your questions, if not do that now and hit Send

  4. IMPORTANT: Ask Mobilize "Please provide all of the questions you will create"

  5. Review the questions and ensure they are correct

  6. Click Mobilize to create the Draft

  7. Enter a Name for the Draft and click Create

  8. Wait for the Draft to be created and click the "Take me there link" to open it

If Draft Creation Fails

This typically occurs when your dataset contains too many questions for a single workflow. Mobilize AI may timeout or fail to process the entire structure.

Solution: Split Into Multiple Workflows

Actions:

  • Identify your top-level categories (Header 1 items)

  • Create separate workflows - one per top-level category

  • Example: If you have "Building 1", "Building 2", "Building 3" as top-level categories, create three separate workflows

Step 5: Review the Generated Draft

What You'll Do: Examine the AI-generated workflow for accuracy and completeness.

Actions:

  1. Wait for AI processing (progress bar will display)

  2. Review the workflow structure that Mobilize AI created

  3. Check these key elements:

    • Branching levels match your header structure

    • All unique values from each column appear as options

    • Questions appear only after all selections are made

    • Question text matches your question text exactly

    • All questions have "reportName" tags

    • Filtering works correctly for all branching paths

  4. Verify completeness - all your questions should be represented

  5. Make edits as needed and Save the Draft

  6. Publish the workflow to your testing site

Step 6: Test and Refine

What You'll Do: Validate the workflow functions as intended.

Actions:

  1. Run a test of the workflow from start to finish

    1. Log out of the mobile app or workflows.anvl.com

    2. Log back in to the mobile app or workflows.anvl.com

    3. Click Start Workflow and select the workflow to start it

  2. Try different combinations of selections to ensure proper filtering

  3. Check question accuracy against your original Excel file

  4. Verify reporting tags are present on all questions

  5. Test all branching paths to ensure complete functionality

  6. If issues exist:

    • Note specific problems

    • Provide corrective feedback to Mobilize AI

    • Request specific modifications

  7. Once satisfied, save and deploy your workflow

Step 7: Document and Save

What You'll Do: Preserve your work for future updates.

Actions:

  1. Save your Excel file with a clear filename and date

  2. Document the workflow name and purpose in Mobilize AI

  3. Save any custom prompts you used for future reference

  4. Test the final workflow one more time

  5. Share with others about what worked will and what to avoid in the process

Troubleshooting Common Issues

If questions don't appear correctly:

  • Verify consistent spelling in all category columns

  • Ensure complete data across all rows

If branching logic is wrong:

  • Review your header order

  • Check for duplicate or inconsistent values

  • Verify the prompt was submitted correctly

  • Provide more detailed branching instructions in your prompt.

If questions are modified:

  • Emphasize "exact language" in your feedback

  • Provide the specific original wording

  • Request regeneration with stricter adherence to source text

Success Criteria

Your workflow is complete when:

  • Users can navigate through all branching levels in sequence

  • Only relevant questions appear based on complete selection path

  • All question text matches your source exactly

  • ReportName tags are present on all questions

  • The workflow flows logically through all branching levels

  • All your original data is represented in the workflow

  • Filtering works correctly for every possible selection combination

This process typically takes 15-30 minutes for a standard dataset and ensures you get exactly the branching logic workflow you need, regardless of how many branching levels your data requires.

Did this answer your question?