Outcome Over Process: Automate, Innovate, Eliminate

We love our processes. They make us feel competent, consistent, and in control. But people don’t actually want the process, they want the outcome it delivers. The process is only a means to an end. Patients don’t want “healthcare”, they want to be healthy. We don’t want “legal proceedings”, we want fairness and respect in society. In the age of AI, we have the most powerful tool ever created, yet much of the focus remains on applying it to existing processes rather than rethinking them entirely. Too often, the aim is to replicate the old way with AI instead of asking: is there a better route to the same outcome, or better still, can we imagine outcomes that far exceed today’s standards?

The strategic question shifts from “How do we add AI to this process?” to “What outcome are we here to deliver, and what’s the fastest, safest, most scalable way to get there?” This reframe reveals three distinct plays: Automate, Innovate, Eliminate.

This post is a practical playbook for those who want to stop worshipping the “how” and start optimising for the “why.”

The Process Trap (and Why It’s So Comfortable)

AI’s extraordinary capabilities have triggered a defensive reflex in many people. It forces us to re-evaluate our value and relevance, especially in work that is heavily process driven. When a tool emerges that can potentially do almost anything, it’s natural to wonder: if AI can do my tasks as well, or even better, then where does that leave me?

This often leads to process-first thinking, defending the way we work as a way of defending our place in the world. It plays out at both individual and organisational levels. Layered on top is a common bias: most of us quietly believe our own jobs are safe, yet we’re far more certain that others’ roles could be replaced. That bias keeps us wedded to familiar routines. We become attached to the steps, tools, and artefacts we’ve mastered, mistaking them for the actual value we deliver. Ironically, clinging too tightly to a process can make us even more vulnerable to being displaced.

Outcome-first organisations flip this logic:

• Define what must be true for customers to experience value.

• Enable multiple pathways to that destination (human led, AI augmented, AI led).

• Measure outcomes, not activity.

Litmus test: If your KPIs focus mainly on internal throughput and process compliance, you’re optimising the factory. If they focus on customer results and access, you’re optimising the outcome.

Machines Don’t Need to Think Like Us

A common form of process-first thinking is the belief that machines must mimic our reasoning steps to match our performance. They don’t. Pattern driven systems can achieve the same results by entirely different means. That can feel unsettling, especially if your sense of value comes from how the task is done, because the task itself might soon disappear altogether.

The focus should always return to the outcome:

• Process fidelity is optional. If the outcome is what matters, don’t discount alternative routes just because they look unfamiliar.

• Human judgement shifts up a level. We spend more time defining the goal, deciding when to intervene, and handling ambiguity, and less time grinding through each step ourselves.

Start with the destination. Then evaluate each route based on reliability, safety, explainability, and cost.

The Tech Trinity: Automate, Innovate, Eliminate

Technology can change performance in three distinct ways:

A) Automate: Deliver the same outcome faster, cheaper, and with fewer errors by enhancing existing workflows such as drafting, summarising, classifying, retrieving, and checking. It’s the path of least resistance because it leaves the operating model intact.

B) Innovate: Redesign how you achieve the outcome. This could mean self service experiences, agentic orchestration across systems, minimally invasive diagnostics, or real time data fabrics that remove handoffs. The destination is the same, but the route is entirely new.

C) Eliminate: Remove the problem altogether by addressing it upstream so the outcome is no longer required. Prevention beats response: build a fence at the top of the cliff instead of parking an ambulance at the bottom. For example, rather than building robots to iron clothes, use materials science to create crease proof fabrics that make ironing obsolete.

Choosing between A / I / E:

• If customers are waiting and errors are frequent → Automate.

• If the process itself is the bottleneck (handoffs, specialist scarcity, travel, scheduling) → Innovate.

• If the demand is avoidable (preventable defects, predictable issues, structural causes) → Eliminate.

Many organisations over rotate on Automate because they never stop to ask the deeper question: What problem are we actually trying to solve by working this way?

Vision Led Restructuring: Ask the Right Question

The questions you ask determine the possibilities you see.

• Ask “What’s the future of ironing?” and you’ll picture a robot doing the same task.

• Ask “How do we remove creases?” and you’ll explore textiles, storage, washing, and design choices that eliminate the task entirely.

• Ask “What’s the future of lawyers/doctors/teachers?” and you’ll default to preserving those roles.

• Ask “What are today’s best answers trying to solve, and what better solutions could exist?” and you’ll open up space for innovation by avoiding attachment to current solutions.

First, articulate the job in plain language ("Get me healthy", "Keep me safe", "Teach me to apply this skill", "Resolve my issue quickly without bouncing between teams"). Then brainstorm multiple routes, human led, AI augmented, or AI led, and choose based on evidence, not habit. AI isn’t always the answer, but it has greatly expanded the range of viable options. The biggest constraint is how we think about the problem.

A vision led organisation defines outcome statements that are short, measurable, and human:

• Healthy without repeat visits

• Resolved in a single interaction

• Learn the skill in half the time

• Onboard in minutes, not days

You Can’t Change the Wheel While the Car Is Moving

Incumbents often promise “transformation”, but for anything beyond automation, the physics are unkind. You can’t truly re-engineer the car while driving at motorway speed.

The pragmatic answer is ambidexterity:

• Exploit (Core): keep the current engine reliable, automate to lower cost and raise quality.

• Explore (New Vehicle): build a separate vehicle that delivers the same outcomes via new methods (innovate and eliminate). Govern and fund it differently, then migrate passengers when it’s roadworthy.

Why separation matters:

• Different cadence: the core values stability, the new vehicle values learning velocity.

• Different KPIs: the core chases margin and reliability, the new vehicle chases adoption, outcome learning, and time to value.

• Different talent mix: operating the known ≠ discovering the new.

• Reduced antibodies: the core’s governance won’t accidentally strangle the new.

Designing for Resilience

Many organisations stay locked into familiar ways of working because they fear the risks of changing their core. Innovating or eliminating can mean taking big, uncomfortable decisions, but if you stay anchored to the outcomes you’re pursuing, the disruption is usually worth it.

Resilience is not just about enduring disruption, it’s about ensuring outcomes are maintained when conditions shift, systems fail, or demand surges. It means building redundancy, fallback modes, and manual overrides into your solutions from day one. It requires constant monitoring, clear escalation paths, and well practised recovery plans so the organisation can return to a safe, functional state quickly. In an AI driven world, where transparency and explainability can be limited, these safeguards are more critical than ever.

Equally, resilience is about anticipating potential failure points before they happen, whether technical, operational, or human. It’s about designing processes and choosing technologies so that performance and reliability reinforce each other, not compete.

Measure What Customers Value

Up to this point, we’ve explored how to rethink processes, build new vehicles, and design for resilience. But none of that matters if you don’t know whether those changes are genuinely improving life for the people you serve. The only way to know you’re on the right track is to measure what customers actually value.

That means shifting from process led efficiency metrics to indicators that reflect real world impact. Move beyond counting activities, like tickets closed or hours logged, and focus on how well you deliver the promised outcome. Track measures such as time to outcome, first interaction resolution, and the quality and consistency of the result from the customer’s perspective. You might also assess access (who benefits and how easily) and cost per successful outcome. By aligning metrics with what customers care about most, you ensure that every optimisation effort drives meaningful, visible improvement.

Your Checklist

Here’s how to put all of these ideas into action, distilled into clear, practical steps you can follow immediately.

Step 1: Clarify the outcome in plain language. Write the outcome as the customer or end-user would. Keep it to a sentence.

Step 2: Map three routes to the destination. Human led (baseline), AI augmented, AI led.

Step 3: Classify opportunities by the Tech Trinity (Automate, Innovate, Eliminate).

Step 4: Build the new vehicle (don’t retrofit the old one). Ring fenced explore unit with clear governance and budget.

Step 5: Design for resilience as well as performance. Fallback modes, human oversight, monitoring.

Step 6: Update metrics - measure what customers value. Time to outcome, first interaction resolution, defect rate, access widened, cost per outcome.

Step 7: Move passengers when roadworthy. Planned migration with clear triggers.

Choose the Destination, Then Build the Right Vehicle

In stable times, process is a comfort. In compounding times, process can be a trap. The organisations that win from here will be ruthless about outcomes and flexible about methods. They’ll bank the quick wins from automation, push into innovation to change constraints, and invest in elimination so whole classes of problems quietly disappear.

Don’t bolt AI onto yesterday’s way of working and call it transformation. Build the new car. Keep the passengers safe. Move them when it’s ready.