By Christopher Sanchez

Organisations are racing to deploy AI, automation and workflow technologies. But many are discovering the same uncomfortable reality: speeding up broken processes just creates faster dysfunction.

The organisations seeing measurable outcomes are taking a different approach.

They are fixing the process first.

Ahead of the Process to Practice Summit 2026, I spoke with Sneha Gadkari, Business Transformation Leader at Transport for NSW, Alan Skinner, Director of the Lean Six Sigma Program at UTS Business School, and Stefan Norrvall, Founder and Principal Consultant at Synexia.

Together, they explored why process discipline, governance and operational clarity are becoming the critical foundations for successful AI and automation initiatives.

The common thread across the discussions? Technology is rarely the starting point.

Why does AI need structure before speed?

“Automation does not fix unclear ownership or weak decision rights; it usually makes those issues move faster.”

Many organisations still approach AI implementation as primarily a technology exercise. But sustainable transformation depends on first understanding how work actually happens across the organisation.

As Stefan Norvall explains, “Automation does not fix unclear ownership or weak decision rights; it usually makes those issues move faster. A good BPM foundation helps expose where the process is structurally ready for automation, and where it is simply carrying ambiguity that needs to be resolved first.”

Conversely, Sneha Gadkari notes that BPM gives organisations the governance and visibility needed to make AI usable and defensible in complex environments.

“BPM gives AI a governed canvas to make faster decisions to replace complex approval loops that exist in most hierarchical structures,” she explains.

Importantly, that structure also creates explainability. “You get explainability by design. When regulators or leadership ask why a decision was made, the answer lives in the model, not in a black box.”

This challenge becomes even more critical as organisations attempt to balance speed with oversight.

“High speed automation may not always produce desired results or outcomes, hence having controlled speed with decision making is key,” says Gadkari. “A real-time faster process doesn’t mean everything happens instantly. It means your process is always aware of where the customer is, what they need, and what your next best action is.”

Why does process design matter more than the technology?

For Alan Skinner, one of the biggest risks organisations face is implementing automation into workflows that were never properly analysed or designed in the first place.

“Effective AI and automation that don’t result in high frequencies of errors or escalation and rework depend on process analysis and design where improvement has been brought to bear for the implementation of automation or AI components.”

Skinner emphasises that staff involvement is essential to avoiding unintended consequences. “If the staff who interact with the system are involved in contributing to the flow and design, it helps avoid unintended consequences that create waste and rework.”

He also points to prototyping and testing as critical safeguards before workflows go live. “The process that is speeded up and is intended to remove errors needs to be well designed based on observation, analysis and improvement.” Further, “the people who interact with the system need to be involved to prevent unforeseen, practical errors e.g., Failure Modes Analysis. Use of prototyping with staff can help reduce catastrophic events when systems go into production / practice.”

Why is process mining only the beginning of effective process improvement?

“Seeing deviation, rework, or delay is only the first step; the harder question is why it is happening.

Another recurring theme across the process improvement landscape is the growing role of process mining and operational data in transformation initiatives. But experts warn that organisations often misunderstand what process mining can, and cannot, solve on its own.

“Process mining is not a one-time activity to fix broken processes,” says Gadkari. “Its value is long term and continuous.”

She argues that organisations only realise real ROI when process mining is continuously connected to live execution data and embedded into an ongoing improvement cycle.

Norrvall agrees, while also warning against treating process maps as complete diagnoses.

“Process mining is very powerful for showing what is actually happening rather than what the documented process says should happen. The risk is treating the map as the diagnosis. Seeing deviation, rework, or delay is only the first step; the harder question is why it is happening. Often the cause sits in unclear authority, unresolved handoffs, or accountability boundaries that do not match the work. Process mining becomes much more valuable when paired with structural diagnosis, not just performance analysis.”

Skinner adds that many organisations still struggle with one of the most fundamental disciplines required for effective process improvement: measurement architecture.

“The set up of measurement architecture is known to be one of the weakest aspects of management practice,” he says. Without clear strategy maps, process performance measures, and operational feedback loops, organisations struggle to identify capability gaps or prioritise improvement opportunities effectively.

How do you measure ROI beyond efficiency metrics?

“Anchor ROI to customer outcome metrics, not just internal efficiency gains. Measure the cost of deviation, not just the cost of the process.”

One of the biggest challenges facing process leaders is proving the value of BPM and operational improvement initiatives beyond isolated efficiency gains. It’s often the case that organisations focus too narrowly on cycle time reduction or throughput improvements while overlooking broader operational impacts.

“Anchor ROI to customer outcome metrics, not just internal efficiency gains,” says Gadkari. “Measure the cost of deviation, not just the cost of the process. Set a baseline before you optimise. Organisations that skip this can’t prove what changed; a pre-transformation process model is your baseline.”

Norrvall similarly cautions against ignoring hidden coordination costs. “A BPM system may improve the visible process while leaving senior people still spending large amounts of time resolving cross-functional friction.”

Instead, he argues that “A better ROI view includes whether the system has reduced decision latency, exception handling effort, duplicated work, and the amount of informal coordination needed to keep the work moving.”

Skinner also argues that ROI must be viewed through a broader operational lens. “Organisations should develop an expectation and consistent practice of measuring both performance changes (stability and capability) and then the return in monetary terms.” At the end of the day, “ROI needs to be seen in terms of the Balanced Scorecard, not just the dollars.” This includes tracking customer perception shifts, collaboration outcomes, leadership capability, and long-term process stability alongside financial improvements.

How do you get operational excellence right by thinking differently?

The conversations being had between process improvement leaders point to a broader shift underway in operational excellence and transformation.

Leading organisations are moving away from treating AI, BPM, automation and continuous improvement as separate initiatives. Instead, they are building integrated operating models where process visibility, governance, data and human decision-making work together.

The result is not simply faster processes, but more adaptive, resilient and scalable organisations.

As organisations face increasing pressure to modernise operations, reduce friction and deliver measurable transformation outcomes, the question is becoming less about whether to adopt AI and automation, and more about whether the underlying processes are capable of supporting them.

Or, as Stefan, Alan and Sneha would argue: before you automate the work, make sure the work itself actually works.

Join the conversation at the Process to Practice Summit 2026, Hear more from Sneha Gadkari, Alan Skinner, Stefan Norrvall and other thought-leaders from 4 - 6 August 2026 at Parkroyal Darling Harbour, Sydney. Learn more.