Now accepting guest editorial from SAP, Oracle, Workday and ServiceNow practitioners.Pitch a guide
The Sign-Off Table

Lesson 1 of 6

Why regression dominates SAP run

7 min

Listen to this lesson

0:00 / 9:03

Go-live is not the end of testing. It is the start of a different kind. Contoso Motor Co. is now live on SAP S/4HANA 2025 in a private cloud, and the project team has handed over to a run organization. From this point on, most of the testing effort Contoso spends will not go into new features. It will go into one question asked again and again: does everything that worked yesterday still work today? That question is regression testing, and in the run phase of an SAP system it is the dominant test activity. The reason is simple: an SAP system never stops changing. SAP delivers new software, the business asks for new configuration, developers fix and extend custom code, and the platform underneath is patched. Every one of those changes arrives as a transport or a technical update, and every one of them can break a process that nobody meant to touch. The diagram below shows the six layers that change underneath a live system.

SAP software deliveriesRelease upgrades, feature package stacks and support package stacks: thousands of changed objects at once.
SAP NotesIndividual corrections, legal changes and monthly security notes, implemented one by one.
ConfigurationCustomizing transports: pricing, output, account determination, new plants or sales areas.
Custom code and extensionsEnhancements, reports, interfaces and Fiori extensions built by Contoso or its partners.
Data and integrationMaster data mass changes, new interfaces, changed partner systems and certificates.
PlatformDatabase, kernel, operating system and browser updates that change behavior under the application.
Six sources of change in a live SAP landscape. Each one is a regression risk.
The Sign-Off Table · thesignofftable.com/academy/sap-tst-205

Each layer changes on its own schedule. SAP Notes can arrive any week; security notes follow a monthly rhythm; support package stacks and feature package stacks follow SAP's release plan; configuration and custom code follow the business's change requests. The result is a steady stream of change that never really pauses. Every system has regression risk, but SAP concentrates it, because a change in one area often shows up somewhere else, days later, in somebody else's process. The example below is typical of a first quarter after go-live.

  1. Small changeA pricing transport adjusts freight conditions for one scenario.
  2. Retest passesThe requester checks the scenario the change was meant for; it works.
  3. Import to productionNobody tests the other processes that use the same pricing procedure.
  4. Quiet side effectIntercompany billing between plants 1710 and 1720 now calculates freight differently.
  5. Found days laterFinance spots the wrong margin at month-end, after many documents have already posted.
How one routine transport became a production incident in a process nobody meant to touch.
The Sign-Off Table · thesignofftable.com/academy/sap-tst-205

Five reasons SAP is especially exposed

  • Integration. A sales order drives availability, delivery, goods issue, billing, revenue and margin. A change in pricing can surface as a wrong journal entry in finance.
  • Shared master data. The product CMC-ALT-100 is used by sales, purchasing, production, warehouse and costing. A change to one view can affect all of them.
  • Configuration-driven behavior. Much of SAP's behavior is set by customizing tables, not code. A single table entry can change determination logic for every document of a type.
  • Custom code hooks. Enhancements and user exits run inside standard processes. When SAP changes the standard code around them, they can behave differently.
  • Periodic processes. Month-end close, payment runs and year-end activities run rarely, so a break may not be noticed until weeks after the change that caused it.

Retesting is not regression testing. Two kinds of test follow every change, and they answer different questions. A confirmation test (often called a retest) checks that the change itself works: the defect is fixed, or the new field appears. A regression test checks that the change has not damaged anything else. Teams often do the first and call it the second. The confusion is understandable: both happen after a change, both are often run by the same key user, and both end with a pass or fail. But a retest is designed by looking at the change, while a regression test is designed by looking at everything around it. If the only person who tests a change is the person who asked for it, you almost certainly have a retest and no regression test.

TestQuestion it answersContoso example
Confirmation test (retest)Does the change do what it was meant to do?After a correction to freight pricing, the intercompany invoice shows the right freight.
Regression testDid the change break anything that used to work?Standard sales from plant 1710, returns and credit memos still price and post correctly.
Smoke testIs the system basically usable after an import or update?Users can log on, create a sales order and post a goods receipt; key jobs and interfaces run.

Three tests that follow a change, and the different questions they answer.

The economics of run

In a project, regression is a phase near the end. In run, it is a repeating activity that happens dozens of times a year. That changes the economics. A regression approach that needs three weeks and twenty key users is tolerable once before go-live; it is impossible every month. Without a strategy, run teams drift to one of two extremes. Some test everything every time, which is slow, exhausts the key users and delays fixes, so the business starts to bypass the process. Others test only the change itself, which is fast until a quiet break reaches production and costs more to repair than all the testing that was skipped. Production defects in SAP are expensive because they are found late, after wrong documents have already been posted, shipped or paid, and each wrong document needs its own correction. The rest of this course gives you the middle path: a regression suite in tiers, ranked by risk, scheduled around SAP's delivery cadence and automated where automation actually pays.

ApproachTest everything every timeTest only the changeTiered, risk-based regression
Effort per changeWeeks of key-user timeMinutes to hoursMatched to the size of the change
Speed of fixesSlow; fixes queue behind testingFastFast for small changes, planned for large ones
Production riskLow, until the business starts bypassing the processHigh; quiet breaks reach productionLow where it matters most
Sustainable dozens of times a year?NoYes, until the first big incidentYes
The two extremes run teams drift into, and the middle path this course builds.
The Sign-Off Table · thesignofftable.com/academy/sap-tst-205

What would you do?

Maria, Contoso's order-to-cash key user, raises a change to add a field to the sales order header in the Fiori app. The developer says: "It is only a screen change, so no regression testing is needed. Maria can retest the field and we transport it Friday." What do you recommend?

Key points
  • An SAP system in run changes constantly: SAP deliveries, notes, configuration, custom code, data and platform.
  • Every transport is a regression risk because SAP is deeply integrated and configuration-driven.
  • A confirmation test proves the change works; a regression test proves nothing else broke.
  • Run-phase regression repeats many times a year, so it must be lean, prioritized and partly automated.

Check your understanding · not graded

Lesson check

  1. 1. What question does a regression test answer?

  2. 2. Why can a pricing configuration change break a finance result?

  3. 3. Why are month-end processes a particular regression risk?

  4. 4. Which approach to run-phase regression is sustainable?