Automation Team Playbook: The Telenor Waiver Process

A practical, prioritised guide for the automation team working on the supplier waiver process at Telenor Group Sustainability. It answers three questions in order: what to correct first, what to redesign, and what to automate. The reasoning is grounded in two discovery sessions with Annabelle (process owner) and Zainab, interviewed by Mike Riess (initial interview 5 May 2026; validation call 22 May 2026), and scored against the Devil's Quadrangle of Time, Cost, Quality and Flexibility.

The one idea that should drive everything

The single most important finding from the validation call is that the biggest wins are upstream, not downstream. The instinct is to automate the tedious word-by-word document comparison that the process owner dislikes. That is the wrong target. The comparison is partly redundant because procurement already does a first-pass comparison, and the reviewer's real value is judgment, not re-doing that comparison. The leverage is therefore in stopping the wrong work from reaching the reviewer at all: triage low-risk cases out, force a complete intake so there is no email ping-pong, and let procurement own the comparison artefact.

High leverage upstream → low leverage downstream Triage filter low-risk Complete intake no email loop Review human judgment Compare low payoff Effort spent here removes work the reviewer never needs to see
Spend the automation budget on the left of the chain, not the right.

How the team should proceed

Work in three tiers. Do not start by writing automation code. Correct the structural misunderstandings first, redesign the flow second, and only then automate the steps that survive the redesign, because automating a flawed flow just makes the flaws faster.

1 Correct first (cheap, mostly non-technical)

These are clarifications and definitions. They cost almost nothing and they prevent you from automating the wrong thing.

Correct Make roles explicit: recommend ≠ approve

The coordinator repeatedly treats the sustainability recommendation as the approval, when accountability sits with the local CEO who signs. The process owner partly attributes this to the team not having explained the role clearly enough. Build the role distinction into the form and the BPMN as an explicit, visible step. A1, C1

Correct Define the right document type at the source

Vendors recurrently submit their internal employee code of conduct as a "match" for Telenor's supplier-facing principles. They are not equivalent. A definition and an example, surfaced at intake, removes a whole class of wasted comparison cycles. A3

Correct Close the post-transition paper-trail gap

Since the ABC to SCP transition only the vendor signs, and with no common repository the process owner cannot verify completion. She judged this a bigger risk than before. Decide where the signed waiver is stored and who confirms completion, before automating anything that depends on that record. A5, A6

2 Redesign second (structural, highest payoff)

These change the shape of the flow. They are where the Time, Cost and Flexibility gains concentrate.

Redesign Add a triage gateway at procurement intake

Small local contracts with no reputational impact do not need group involvement. A rule-based or AI-assisted classifier keyed on contract size, criticality, geography and reputational signals routes those cases to a local resolution and keeps them off the reviewer's queue. This is the highest-leverage change in the whole process. Already modelled as Gateway_Triage from BPMN v10 onward. B9, C11

Redesign Replace the email loop with a complete structured intake

Much of the friction is informal email back-and-forth before the form arrives. The process owner herself acknowledged this makes the process more complicated than it needs to be, and pointed to the earlier "submit the form first, then we talk" practice as better. Enforce a self-service template that procurement completes before any interaction. A11, A17, B4, B8, C4

Redesign Re-position procurement as the comparison owner

Procurement already performs the first-pass comparison. Formalise that they own the comparison artefact and share their conclusion when raising a case; the reviewer then operates as a review/escalation layer instead of re-doing the comparison. Removes the duplicated effort and clarifies responsibility. Modelled as a dedicated procurement comparison task from v10. A18, C12

Redesign Split waiver and modification into separate paths

Waiver and modification differ fundamentally: modifications change legal text and carry a higher risk of unintended commitment, so legal must be looped in closely. Model modification as an iterative wording loop between reviewer, legal, coordinator and vendor, planned for N rounds rather than one pass. Represented as a loop activity from v10. A13, C3, C13

3 Automate last (only what survives the redesign)

After triage and intake are fixed, these are the safe, valuable automations.

Automate "Homework done" pre-check at intake

The process owner's own top wish: a formalised checklist confirming basic screening and risk assessment are complete before she is involved, so the discussion concentrates on substance. This is a form-and-validation automation, not an AI model. B4, C4

Automate AI-assisted high-level comparison as a first impression

AI already gives a useful first impression of document differences, but it is not always accurate and must be reviewed. Keep it strictly as a starting point that the reviewer confirms. B1, B3

Automate Modification pattern library, with guardrails

Modification requests often resemble each other, so a library of "this kind of text → typical response" is feasible. Legal warned that no two cases are identical and risk varies widely by vendor type, so the library is a starting point that must be reviewed each time, never a copy-paste. B2

Automate Document aggregation via the incoming tooling

The new vendor-assessment tool (target 2026) consolidates documents and scans public data, which can attach all vendor documents in one place and chip away at the IDD backlog. It does not replace the review; it removes the email-attachment shuffle. B5, B6, B7

Non-negotiable guardrail. The process owner set an explicit constraint: any AI may produce recommendations, never decisions. The human stays in the loop, and accountability remains with the local CEO. Every automation above must preserve this. Treat it as an acceptance criterion, not a nice-to-have.

Reading impact with the quadrangle

Score every candidate change against the four competing dimensions before committing. The triage gateway and the structured intake touch all four positively, which is why they rank first. The pattern library mainly buys Time and Cost but risks Quality if the guardrail slips, which is why it is tier three.

ChangeTimeCostQualityFlex.
Triage gateway++·+
Structured intake+++·
Procurement owns comparison++··
Waiver/modification split··++
Homework pre-check+·+·
Pattern library++at risk·

Key concepts

Further learning

Sources

  1. Dumas, M., La Rosa, M., Mendling, J., & Reijers, H. A. (2018). Fundamentals of Business Process Management (2nd ed.). Springer. link.springer.com/book/10.1007/978-3-662-56509-4
  2. Reijers, H. A., & Limam Mansar, S. (2005). Best practices in business process redesign: an overview and qualitative evaluation of successful redesign heuristics. Omega, 33(4), 283–306. doi.org/10.1016/j.omega.2004.04.012
  3. High-Level Expert Group on AI, European Commission (2019). Ethics Guidelines for Trustworthy AI (human agency and oversight). digital-strategy.ec.europa.eu/en/library/ethics-guidelines-trustworthy-ai
  4. Object Management Group. Business Process Model and Notation (BPMN) Version 2.0. omg.org/spec/BPMN/2.0
  5. Primary material: waiver-process discovery interviews with Annabelle and Zainab, conducted by Mike Riess. Initial interview 5 May 2026; validation call 22 May 2026 (findings A1–A18, B1–B9, C1–C13). Internal Telenor Group Sustainability source material.