Test Scripts, Test Data and User Training With AI

Testing and training are where SAP projects often run short of time. Test scripts get written in a hurry, negative cases are skipped, and user guides arrive a week before go-live. AI can take a large part of this writing load, so more of your time goes into actually testing and supporting users.

Test scripts from a process description

Describe the process (anonymised) and the format your project uses:

Write test scripts for this process in S/4HANA: [describe, e.g. create a
purchase requisition, convert to purchase order, goods receipt, invoice
verification, with these specific rules: ...].
Use a table with columns: Step, Role, Action, Input data, Expected result.
Include:
- the main (happy path) scenario
- 5 negative scenarios (e.g. missing approval, price variance above
  tolerance, blocked vendor)
- 2 edge cases
Write [TO CONFIRM] where an expected result depends on our configuration.
Do not invent transaction codes or app names; write [APP/TCODE] instead.

Replace [APP/TCODE] and [TO CONFIRM] with the real values from your system. Expected results are the part that most needs your knowledge of the configuration.

Thinking of the cases you would miss

AI is good at listing scenarios people forget:

For the process above, list scenarios that are often missed in SAP
testing: period-end timing, cancellations and reversals, partial
deliveries, currency and unit conversions, authorisation differences,
integration with FI/CO postings, and output (print/email) issues.

Choose the ones that matter for this client and add them to your scripts.

Planning test data

Do not ask a public AI tool for real master data, and never copy production data into prompts. Instead, ask for a test data plan:

For these test scripts, list the test data needed: master data (material
types, vendors, customers, with the attributes that matter for each
scenario) and transactional data. Group by scenario and mark data that
must be created in the test system before testing starts.

Then create the data in the test system following the project's data rules. If your project uses anonymised copies of production data, that process belongs to the basis and data teams, not to an AI chat.

From test scripts to user guides

Once a process is tested, the test script is a great source for training material:

Turn this tested script into a user guide for [role, e.g. purchasing
clerk] who is new to S/4HANA. Numbered steps, one action per step, with
notes on what to check on the screen. Add a "Common mistakes" section and
a "Who to contact" section with placeholders.

Add your own screenshots from the test system (with data anonymised if needed). Real screenshots make guides far easier to follow.

Quick-reference cards and quizzes

Create a one-page quick-reference card for this process: the 6 main
steps, 3 key checks and 3 common errors with what to do.

Write 8 multiple-choice questions to check that users understood this
process, with the correct answer and a one-line explanation for each.

Quizzes in training sessions show quickly which users need extra help before go-live.

Translating training material

If users prefer Hindi or another language, AI can draft translations. Keep SAP terms, field names and button labels in the language the system shows, and have a bilingual key user check the result.

Practice task

Pick one process you are testing or supporting. Generate test scripts with negative cases in your project's format, fill in the real transactions and expected results, then turn the script into a user guide and a five-question quiz.

Key takeaways

• Generate test scripts in your format, with negative and edge cases, and placeholders for system-specific details.
• Ask AI for missed scenarios, then choose the ones relevant to your client.
• Plan test data with AI, but create it in the system; never use real production data in prompts.
• Reuse tested scripts for user guides, quick cards and quizzes, with real screenshots.