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

Lesson 3 of 5

Baselines

9 min

A KPI without a baseline is a number without meaning. Was DSO of 48 days good? Only if you know it was 52 before. Every significant change, whether a go-live, an upgrade or a process redesign, should have a baseline captured before and a comparison after.

Capturing a baseline

  1. Fix the definitionsWrite down how each KPI is calculated and which data it includes
  2. Choose the periodUsually several months, covering the same season as the comparison period
  3. Calculate in the old systemUse the same definition the new system will use
  4. Record contextVolumes, seasonality, known one-offs such as a large customer bankruptcy
  5. Store itWith the sign-off pack, so the comparison can be made later
Capturing a KPI baseline before a change.
The Sign-Off Table · thesignofftable.com/academy/sap-pow-302

The hardest part for an ERP migration is calculating the baseline in ECC with the S/4HANA definition. It is worth the effort: comparing an ECC "home-grown" KPI with an S/4HANA standard KPI tells you nothing about whether the process changed.

Comparing after the change

  1. Months -6 to -1BaselineKPIs calculated in ECC with the agreed definitions
  2. Go-liveCutoverExpect disruption; KPIs not yet meaningful
  3. Weeks 1–6HypercareTrack weekly; look for direction, not precision
  4. Months 2–3StabilizationCompare with the same months in the baseline
  5. Month 6Benefit reviewCompare against baseline and the business case targets
Baseline and comparison around a go-live.
The Sign-Off Table · thesignofftable.com/academy/sap-pow-302

Expect a dip after go-live. Users are learning, master data issues surface, and blocks catch things the old system let through. A temporary worsening of DSO or the touchless invoice rate is normal; what matters is that it recovers, and how fast. Agree in advance what "recovered" means and by when, and put that in the hypercare exit criteria.

Plan the baseline as part of the cutover and test plan, not as an afterthought. Someone has to run the legacy calculations before the old system is switched to read-only, and someone has to check that the S/4HANA app produces comparable values from migrated data. At Contoso, Grace added a task to the cutover plan for Dan's team to calculate the final ECC month of each KPI using the agreed definitions, and to file the results with the go-live sign-off pack.

Baselines are not only for go-lives. Capture one before each release upgrade, each major feature pack and each process redesign. For a release upgrade, the four weeks before the upgrade weekend are usually enough, as long as they do not include a period-end that the weeks after will not.

Making the comparison fair

  • Compare like with like: the same months, so seasonality does not masquerade as change.
  • Adjust for volume: a rate (touchless invoices as a share) is fairer than a count.
  • Note one-offs: a large customer paying late can move DSO by days on its own.
  • Look at the distribution, not just the average: a stable average can hide a growing tail of very late payments.
Investigate nowWorse and unexplained. Find the cause; may be a defect or a broken control.
Validate and celebrateBetter but unexplained. Check the definition before claiming a benefit.
Track to recoveryWorse but explained, such as learning curve. Set a recovery date.
Report the benefitBetter and explained by the change. Feed into the benefit review.
Reading a KPI change against its baseline.
The Sign-Off Table · thesignofftable.com/academy/sap-pow-302

What would you do?

One month after an S/4HANA feature pack is applied, Contoso's days to close rises from 5 to 7. There is no other known change. What should Dan do?

Key points
  • Capture baselines before every significant change, using the same definitions as afterwards.
  • Expect a post-go-live dip; agree what recovery means and by when.
  • Compare like with like: same season, rates not counts, one-offs noted, distribution not just average.
  • Worse and unexplained means investigate now; better and unexplained means check before celebrating.

Check your understanding · not graded

Lesson check

  1. 1. Why capture a KPI baseline before an upgrade?

  2. 2. What makes a baseline comparison fair across an ERP migration?

  3. 3. What is normal right after go-live?

  4. 4. A KPI improved but nobody can explain why. What should you do?