You need to enable JavaScript in order to use the AI chatbot tool powered by ChatBot
Marks Sattin logo

Why Most AI Initiatives Are Failing

Technology
Rhys Scott-Samuel our consultant managing the role
Author: Rhys Scott-Samuel

Over the past 12–18 months, AI has moved from experimentation to expectation. Across every industry, leadership teams are investing heavily in tools, platforms, and talent with the aim of unlocking value from AI. The assumption is simple: If we implement the right tools, we will unlock better decisions, efficiency, and growth.

Yet when speaking to Heads of Data, AI Directors, and Chief Data Officers on a weekly basis, a consistent pattern continues to emerge:

Very few organisations are struggling with the technology itself.

OpenAI, Microsoft Copilot, Databricks, Snowflake, AWS, and Google Cloud have dramatically lowered the barrier to entry. Capabilities that once required years of investment can now be deployed in weeks.
And yet, despite this AI adoption is still falling short of expectations.The reasons are far less technical, and far more uncomfortable.

1. The trust gap: Why people aren’t using AI properly

The first and most significant barrier is not capability—it’s trust.

Across organisations, there is a quiet uncertainty around how AI should be used:

  • "Can I input sensitive client data into this tool?”
  • “Where does this data go?”
  • “Is this compliant?”
  • “Am I making my own role redundant?”

At the same time, experienced professionals often don’t feel a strong incentive to change:

My current way of working delivers results - why should I adopt something new?


From a leadership perspective, AI represents a clear productivity and capability upgrade.

From an employee perspective, it introduces risk—personal, professional, and regulatory. This disconnect creates a trust gap, and that gap is where adoption stalls.

Until people feel confident and safe using AI tools, usage will remain inconsistent, cautious, and ultimately ineffective.

2. Fragmented Data: The problem AI is exposing, not solving

AI is often positioned as the solution to data challenges. In reality, it is amplifying them.

Most organisations still operate with fragmented data landscapes:

  • Marketing operates off its own dashboards
  • Sales relies on CRM extracts
  • Finance maintains independent models
  • Operations produces separate reporting
  • Data teams are reacting to ad hoc requests
Each function works with different definitions, metrics, and pipelines.

When AI is layered on top of this environment, the outcome is predictable:

  • Inconsistent outputs
  • Low confidence in insights
  • Limited scalability
AI requires connected, reliable, and well-governed data to generate meaningful value. Without that foundation, it simply highlights how disconnected the organisation really is. You cannot build intelligent systems on fragmented data.

3. The ROI trap: starting with AI instead of the problem

A common pressure point for leadership teams is demonstrating return on investment.

This often leads to a familiar pattern:

  • Roll out copilots
  • Launch chatbot initiatives
  • Run multiple AI pilots
  • Invest in licences and tooling
However, very few organisations start with the most important question: What business problem are we trying to solve?

Instead, they begin with: How can we use AI?

This leads to disconnected use cases, unclear outcomes, and difficulty measuring success.AI, in this context, becomes an expensive experiment rather than a strategic enabler.

The reality is simple: AI is not a strategy. It is a tool to solve clearly defined problems.

Without clarity on the problem, even the most advanced AI solutions will fail to deliver meaningful impact.

What successful organisations are doing differently

Organisations that are seeing tangible results from AI adoption are not necessarily the ones with the most advanced technology. They are the ones with the strongest foundations.

In practice, this means:

1. Creating Clear AI Guardrails They define how AI can and should be used across the business:

  • Data usage policies
  • Compliance boundaries
  • Approved tools and use cases
This removes uncertainty and gives employees the confidence to adopt AI safely.

2. Prioritising Data Foundations Rather than aiming for perfection, they focus on:

  • Establishing a usable single source of truth
  • Aligning key metrics across functions
  • Improving data quality and accessibility
  • This enables AI to produce consistent and trusted outputs.

3. Starting With Business Problems, Not Technology They identify high-impact operational challenges first:

  • Inefficient processes
  • Manual workflows
  • Decision bottlenecks
Only then do they apply AI where it can deliver clear, measurable improvements.

4. Internal Training and upskilling Huge initiatives on internal training and ensuring the business knows what AI is, how data is stored and creating a collaborative and inclusive culture heading towards the same business goals. 

The reality: AI adoption is an organisational challenge

AI transformation is often framed as a technology journey. In reality, it is a combination of three things:

  • Cultural change (trust, behaviour, adoption)
  • Data maturity (quality, consistency, accessibility)
  • Clarity of purpose (problem definition and outcomes)

Until these elements are addressed, organisations will continue to:

  • Invest in tools
  • Run pilots
  • Talk about AI
…but struggle to translate that activity into meaningful business outcomes.

Final thought

AI has reached a point where the technology is no longer the limiting factor. The organisations that succeed will not be those that move fastest to implement tools, but those that build the right foundations to use them effectively.

A question for leaders 

If AI adoption within your organisation isn’t delivering the expected impact:

Is the issue really the technology?

Or is it a question of trust, data, and direction?

26/05/26

Tags:

Popular Searches

Quick links