It is not pessimism. This is what KPMG, one of the largest consulting firms in the world, documented in a study carried out together with HFS Research with almost six hundred executives from thirteen countries. The study is several years old — and that is precisely why it is worth reading today: everything it warned about is still happening. Companies are still making the exact same mistakes, only now with more powerful tools and bigger budgets.
In this article I want to summarize what that study found, and above all, translate it into what it means for an entrepreneur in any market. Because the conclusion, I anticipate, is not what many expect: SMEs have an advantage here that multinationals would envy.
The diagnosis: a lot of investment in AI, little impact
The study numbers speak for themselves. More than half of the companies surveyed confirmed investments of more than $10 million in intelligent automation — the term that encompasses artificial intelligence, advanced analytics, and robotic process automation.
And the result? Only seventeen percent of companies had managed to scale these technologies to the entire organization. The rest were stuck in what the study calls "pilot mode": lab projects, internal demonstrations, isolated initiatives that never touched the actual operation of the business.
Cliff Justice, the head of Intelligent Automation at KPMG in the United States, summed it up bluntly: “Without a comprehensive digital transformation strategy that supports investments across the organization, these projects remain stuck in pilot mode and do not achieve the expected results.”
Ten million dollars. Stuck in pilot mode. Let that sink in for a moment.
Why artificial intelligence projects fail: five causes (none are technological)
Here is the most revealing part of the study. When the researchers asked about obstacles, none of the main ones had to do with the technology itself. The executives pointed out five fronts:
Expectations without preparation. Companies have high hopes for automation, but recognize that they are not ready — especially in change management and governance. That is to say: they want the result, but they have not prepared the way.
Lack of internal talent. Many respondents cited lack of in-house automation knowledge as one of the biggest barriers. They bought the tool, but no one in the house knows how to take advantage of it.
Investment uncertainty. Although they plan to increase spending in the coming years, many already fear that even that will not be enough. They invest without knowing how much it is needed or why exactly.
Team culture and fear. The promise of automation is to make staff more productive and creative — not to cut jobs. But if the company does not manage the expectations of its people during the transition, fear kills the project from within.
Lack of focus. Automation offers opportunities in all areas of the business, and that is precisely why many companies disperse: each department sets up its own isolated experiment, without clear priorities and without connection to the general strategy.
Look at the pattern: money, organization, people, priorities. The technology almost doesn't appear.
Companies that do it well multiply their investment by five to ten
The study is not just a list of failures. He also found growing evidence of something notable: Companies that approach automation strategically from the start — understanding that it can create new business models, not just cost savings — reap dividends of five to ten times their investment.
The difference between burning millions and multiplying them is not in the budget or the brand of the software.
Five to ten times. The difference between burning millions and multiplying them is not in the budget or the brand of the software. It lies in two decisions that the study clearly identifies:
First: automation is a management decision, not a systems project. Investments have to be a strategic imperative of the highest level. In an SME that means something very simple: the owner or general manager personally leads the matter. It is not delegated to "the computer guy."
Second: it is a transformation of the operating model, not a software installation. The study describes it as moving from a “people supported by technology” model to a “technology supported by people” model. It sounds abstract, but in practice it is concrete: processes change, roles change, and you have to design that change — not expect it to happen just because you paid for a license.
The three stages of intelligent automation: where is your business?
One of the most useful parts of the study is its map of intelligent automation into three stages of maturity. It is worth knowing it, because it tells you exactly where your business is and where it is best to move.
Basic
Fixed rules for repetitive tasks.
Improved
The system learns, interprets and converses.
Cognitive
Predictive analysis and continuous learning.
Stage one — basic automation. Systems that follow fixed rules for repetitive tasks: scanning documents and converting them to text, moving data from one system to another, managing expense reports. Useful, but limited: the machine only does exactly what it is programmed to do.
Stage two — enhanced automation. Here the system begins to "learn." Recognizes patterns, understands natural language, processes structured and unstructured information. It no longer just executes rules: it interprets. It is the stage where a artificial intelligence agent You can have a real conversation with a customer on WhatsApp, understand what they need even if they write it in their own words, ask them the right questions, and qualify them before passing them on to a human salesperson.
Stage three — cognitive automation. The full expression of artificial intelligence: systems that continuously learn from multiple sources, process large volumes of data and do predictive analysis. The territory of large corporations with data science teams.
In the words of the study: automation goes from being based on rules, to being based on learning, to being based on reasoning. From acting like a human to thinking like a human.
And where are the majority of SMEs in international markets? Being honest: before stage one. Many businesses still write down orders in notebooks and respond to WhatsApp when there is time.
That sounds bad, but it's exactly the opportunity.
The advantage of SMEs from international markets over multinationals
Let's go back to the question at the beginning: if multinationals with ten million dollars fail, what hope does an SME have?
Much more than it seems. And for three specific reasons.
First: stage two technology is already mature and accessible. What a few years ago required corporate budgets is today available at a fraction of the cost. An SME does not need to invent anything or finance experiments: it can jump straight to proven solutions, right at the stage where the real business value is.
Second: the SME does not carry the burden of the corporate. The big ones fail because of endless committees, systems inherited from twenty years ago, and departments that don't talk to each other. In an SME, the owner decides today and implementation starts tomorrow. That speed, which the corporate cannot buy with money, the SME has for free.
Third: the focus is natural. The study warns against projects dispersed in silos. A well-advised SME does not have that problem: choose a specific use case — for example, stop losing sales due to unanswered messages — you implement it, measure the result, and only then move on to the next one. Exactly what the study recommends and what the big guys fail to do.
But be careful: the advantage only exists if you avoid the mistakes of the greats. An SME that buys “a chatbot” because it is fashionable, without a diagnosis, without a measurable objective and without preparing its team, is going to fail just like the corporate one — only without the financial cushion to absorb the blow.
I saw it before, when the internet first arrived
Allow me a personal note, because this pattern is not new to me.
More than twenty years ago I opened one of the first Internet cafes in different cities and markets. In those years I saw exactly the same movie that the KPMG study describes, only with different technology: some businesses bought computers “because they had to have them” — and the machines ended up collecting dust in a corner. Others first asked the right question: How does this help me sell more or serve my customers better? Those were the ones that grew.
Technology changed. The principle did not change anything: the tool without a strategy is an expense; The strategy with the right tool is an investment.
The three questions before investing in artificial intelligence
If from the entire KPMG study there was only one practical lesson to be learned for the international businessman, it would be this: the order of the steps is everything. Those who fail buy first and ask questions later. Those who multiply their investment first answer three questions:
What specific problem am I going to solve?
I don't "want artificial intelligence." But a problem with first and last name.
Who is responsible for the result?
Someone answers with a figure: more appointments, more quotes or fewer lost clients.
How am I going to prepare my team?
The tool eliminates repetitiveness so that your people sell and serve better.
One: what specific problem am I going to solve? I don't "want artificial intelligence." Otherwise: I lose sales because no one answers the messages at night. My saleswoman wastes half a day with curious people who were never going to buy. Nobody follows up on the quotes sent. A problem, with first and last name.
Two: who is responsible for the result? Not about the project — about the result. Someone in the company, ideally the owner himself at the beginning, who answers for a figure: more appointments scheduled, more quotes answered, fewer clients lost.
Three: how am I going to prepare my team? The study is clear: managing staff expectations is critical. The correct message is not "a machine is coming", but "this tool takes away the repetitiveness so that you can dedicate yourself to selling and serving." If your people are afraid of the project, the project dies — no matter how good the technology is.
Frequently asked questions about AI projects in SMEs
Why do most artificial intelligence projects fail?
According to the KPMG and HFS Research study, the main causes are not technological: lack of a clear strategy, absence of defined managers, poor team preparation and isolated projects without connection to the business. Only seventeen percent of companies surveyed managed to scale their automation technologies.
Is artificial intelligence worth it for an SME across different markets?
Yes, as long as you start with a diagnosis and a specific use case — for example, an AI agent that responds to WhatsApp and qualifies customers. SMEs decide faster than multinationals and can implement in weeks what takes a corporation years.
What is the first step to automate my business with AI?
It's not buying a tool. It means answering three questions: what specific problem are you going to solve, who is responsible for the result, and how are you going to prepare your team. At Ezelero this analysis is done in a free forty-five minute diagnosis.
The right question (closing)
After reading a study like this, the honest conclusion is not "you should invest in artificial intelligence." It is more precise: you have to invest in clarity first, and in technology later.
So, if you're evaluating this topic for your business, don't start by asking "what tool should I buy?" Start by asking “where am I losing money that I could get back?” That question costs zero and is worth more than any software.
And if you want us to answer it together, at Ezelero we always start the same way: with a free forty-five minute diagnosis. Without commitment and without technicalities. We analyze your operation, identify where your money is leaking, and honestly tell you if artificial intelligence is right for you — and if it's not right for you yet, we'll tell you that too.
Because after seeing so many large companies burn millions without a strategy, our conviction is simple: clarity first, technology second.
Your same team, selling much more. That's the goal. Everything else is tools.
Schedule your free forty-five minute diagnosis
We review your operation, identify where time or money is wasting and define whether an AI pilot makes sense for your company.