AI-generated image
Contents
Most AI projects fail to deliver business value. Not because the technology does not work. Because organisations spray technology at problems instead of aiming it at specific outcomes. The fix is not a better model - it is clearer thinking and cleaner data before any model gets built.
The spray-and-pray approach
"We need AI." "We need machine learning." "We need advanced analytics."
These statements sound strategic. They are actually symptoms of unclear thinking.
AI is a capability, not a strategy. Machine learning is a tool, not an outcome. Asking for "advanced analytics" without specifying which decisions need to improve is like asking for "more power" without knowing what you are trying to move.
The spray approach: buy AI platforms, hire data scientists, build models, hope something valuable emerges.
The result: expensive infrastructure producing insights nobody uses.
Why precision is the starting point
The organisations that extract value from AI start differently.
They do not start with "we need AI." They start with "we need to solve X problem, and here is how we will measure success."
Precise: "Our reps spend too much time on customers who will not grow. We need to identify which customers have genuine potential before each call."
Imprecise: "We need AI to optimise our sales force."
The precise version defines success criteria. The imprecise version hopes for magic.
The governance prerequisite
Here is what most AI vendors will not tell you: AI quality depends entirely on data quality.
Models trained on bad data produce confident bad answers. Models trained on inconsistent data produce inconsistent results. Models trained on incomplete data miss the patterns that matter.
Data governance is not a bureaucratic checkbox. It is the foundation that determines whether AI investments produce value or waste.
Before AI: Can you trust your customer data? Are your sales records complete? Do different systems agree on basic facts?
If the answer is "sort of" or "we're working on it," you are not ready for AI. You are ready for data governance.
The failure pattern
Failed AI projects share common characteristics:
Unclear outcomes. The project was funded because AI seemed important, not because a specific business problem demanded a solution.
Data denial. Known data quality issues were minimised. "We'll clean it up as we go." Nobody ever does.
Model obsession. Resources focused on model sophistication rather than deployment practicality. A brilliant model that nobody uses is worthless.
Integration avoidance. The AI system was built alongside existing workflows rather than within them. Adoption required users to change behaviour, and they did not.
Success metrics missing. Nobody defined what success looked like before starting, so nobody could tell whether the project worked.
What precision execution looks like
Step 1: Problem definition. What specific decision are we trying to improve? Who makes that decision? What information would change how they decide?
Step 2: Data assessment. Do we have the data to answer this question? Is it accurate? Is it accessible? Is it current?
Step 3: Simple first. Can we solve this with basic analysis before building complex models? Often, simple logic rules outperform sophisticated AI.
Step 4: Integration design. How will insights reach decision-makers? In their existing workflow? At the moment of decision? Without requiring extra steps?
Step 5: Success measurement. How will we know this worked? What metrics will change? Over what timeframe?
The uncomfortable truth
Most organisations do not need AI. They need cleaner data and clearer thinking.
AI amplifies what exists. If your data is a mess, AI will amplify the mess. If your decision-making is vague, AI will produce vague recommendations.
The organisations that succeed with AI do not succeed because of AI. They succeed because they did the hard work of governance and precision that makes AI valuable.
Stop spraying technology at problems. Start aiming at outcomes.
The projects that succeed share one trait: they knew exactly what they were aiming for before they fired.
Written by
Dieter Herbst
CEO at Herbst Group. Working with pharmaceutical commercial leaders across South Africa, Kenya, and Brazil to transform sales force effectiveness through evidence-based approaches.
Connect on LinkedIn
