The Placebo Button

In 2025, AI adoption stalled because organisations prioritised User Experience over Reliability Experience. To master the art of adoption, we must understand the Placebo Button effect, where tools promise action but are disconnected from the underlying data. This article explores why employees are reverting to manual workarounds and how to go about reconnecting the wiring of the digital enterprise in 2026.

The Illusion of Control: The Psychology Behind the Placebo Button

Next time you step into an office elevator, I want you to watch what people do.

It is a fascinating study in human psychology. They walk in, turn around, and almost immediately press the "Close Door" button. Then they wait a second. Then they press it again. Then, usually, a third time with a bit of aggressive rhythm.

Here is the open secret of the elevator industry: In the vast majority of elevators built in the United States after 1990, that button isn’t connected to anything.

It is a placebo button.

Following the Americans with Disabilities Act of 1990, elevator doors were required to stay open for a minimum duration to ensure accessibility. To comply, manufacturers simply disconnected the manual override from the main controller for standard passengers. The doors are on a timer. They will close when the system decides they should close, and not a moment sooner.

So why do manufacturers leave the button there? Because if they took it away, complaints would skyrocket.

It turns out, humans don't mind waiting. But we hate feeling helpless. The button gives us an illusion of agency. It tells our anxiety: I am doing something. I am in control.

I’ve been thinking about that button a lot as we analyse the AI adoption gap of 2025. Looking back at the last twelve months, I see a corporate landscape littered with stalled pilots, abandoned tools, and frustrated leaders. We spent millions deploying Next-Gen AI agents and fancy dashboards. We gave our teams brand new, shiny interfaces that promised instant results.

But we forgot to check if the button was actually connected to the machine.

The Disconnected Enterprise: Why Shiny Interfaces Fail Without Solid Infrastructure

The primary reason adoption has stalled in 2025 is simply that the wiring behind the interface was broken.

We gave our teams a button called "Resolve Incident." They pressed it. But because the underlying data in the CMDB (Configuration Management Database) was fragmented and outdated, the AI agent couldn't actually resolve the incident. It just hallucinated a solution or returned an error.

We gave them a button called "Onboard Employee." They pressed it. But because the service catalogue wasn't standardised, and because HR and IT have different definitions of, say, "Manager", the workflow hit a dead end.

When a user presses a button and nothing happens, they don't blame the wiring. They blame the button. They say, This AI is useless, and they stop using it.

This is the Placebo Problem of our industry. We are obsessing over the user experience, including the chat window, the voice assistant, and the Magic Button, while ignoring the fact that the cables in the elevator shaft are a tangled mess.

Taking the Stairs: Why Employees Revert to What They Know

When people realise the elevator is broken, or when they realise the buttons don't work, they do something predictable.

They take the stairs.

In a corporate context, taking the stairs is manual work. It is spreadsheets. It is email chains. It is the manual workaround.

For years, IT leaders have labelled this behaviour as Shadow IT—a rebellion against the system. We assume employees use Excel because they are stubborn, or because they refuse to modernise. But the elevator analogy teaches us something different.

Taking the stairs is inefficient. It is tiring. It is slow. But it has one massive advantage over your fancy new AI elevator: It is reliable.

When I walk down the stairs, I know exactly where I am going. I have full control over my speed. I know I won't get stuck between floors.

This explains the AI adoption gap we have seen throughout 2025. Leaders are baffled. They ask, Why are my teams still using Excel when we bought this multimillion-dollar AI-enabled Service Management platform?

They are using Excel because Excel offers certainty. When they type a number in a cell, it stays there. They trust the physics of the stairs more than they trust the magic of your disconnected elevator.

The Industry Waking Up: Software Alone Is No Longer Enough

So, how do we get people back in the elevator in 2026?

We have to stop trying to impress them with how nice the doors look and start fixing the wiring in the shaft.

This is exactly what we are seeing across the market. A recent analysis by Andreessen Horowitz (a16z) confirms that the buy-and-fly era of software is failing in the age of AI. They note that successful vendors are shifting focus towards Forward Deployed Engineering, teams dedicated to the messy, complex work of integration and data context.

The industry is waking up to the hard truth that outcomes must be built, not bought.

For complex enterprise workflows, the software itself is not enough. The a16z research highlights that agents and automation tools break without deep context. They require the heavy lifting of connecting to internal databases, mapping historical records, and defining business logic.

In our previous blog, AI Readiness for Service Management, we argued that getting this foundation right is the only way to scale. The implementation is the product.

You cannot buy a working elevator; you have to build the shaft. You have to wire the sensors. You have to program the logic. The true value lies in the heavy lifting of integration.

The Three Wires You Must Connect: Establishing Discipline Before Automation

At Fusion GBS, we have long argued that adoption is fundamentally a discipline problem.

If you want the "Close Door" button to work, and if you want your AI agents to actually execute tasks rather than just talk about them, you need to connect three specific wires. These have nothing to do with LLMs or GPUs, and everything to do with discipline:

1. Data Lineage: The Cables - The system needs to know exactly where the data comes from and where it goes. In a placebo system, the AI guesses. In a connected system, the AI follows a traceable path. If your Asset Management data is a swamp of duplicates and unknown owners, the elevator is stuck between floors. You need to do the hard work of mapping the lineage, so the AI has a physical cable to pull.
2. Service Standardisation: The Logic Board - You need to define what "Close Door" actually means. Does it mean immediately? Does it mean after the safety check? Does it mean only if the user has a badge? In many organisations, a simple request like "Order a Laptop" triggers five different processes depending on who you ask. If the process isn't standardised, the button is just a suggestion. You cannot automate a mess. You must standardise the logic first.
3. Governance: The Safety Inspection - Users need to know that the elevator won't plummet. This is the biggest blocker to adoption in 2026:

FEAR

Employees are terrified that if they let the AI decide, it will break something, and they will be blamed. Governance is the safety certificate on the elevator wall. It tells the user: This system has been tested. The risks are managed. You are safe to press the button. This assurance is explicitly engineered. In our AI Readiness for Service Management guide, we detail exactly how to engineer these guardrails iteratively, delivering quantifiable value while building the safety net that finally convinces your users to step inside.

The Outlook for 2026: Prioritising Reliability Experience Over User Experience

We are standing on the edge of the Uncanny Valley. We have technology that can act autonomously. We have AI agents that can run the elevator for us, predict which floor we want, and take us there instantly.

But autonomy without connection is just chaos.

If you are looking at your roadmap for 2026, I challenge you to look past the shiny buttons. Stop focusing on the User Experience of the interface and start focusing on the Reliability Experience of the infrastructure.

Your employees are smart. They know the difference between a tool that works and a placebo. They can feel when a system is disconnected.

If you want them to stop taking the stairs, and if you want true, high-velocity adoption, you don't need to promise them a faster ride. You just need to prove that when they press the button, the machine actually listens.

Let’s reconnect the wiring.

About the author:

Keyvan Shirnia - Chief Revenue Officer at Fusion GBS, guiding strategy, growth, and customer value realisation with a focus on service management and digital transformation.