PART OF FORWARD GROUP
In the past year I saw three businesses that together invested over NIS 2 million in AI implementation and came out with nothing. Zero ROI, frustrated employees, systems that are not in use.
The technology was not the problem. The tools they bought were reasonable. They made the same 7 mistakes that I see again and again, in every industry.
Since 2008, 18 years of guiding established businesses through digitization projects, I have seen these 7 mistakes repeat in restaurants, importing from China, medical devices, real estate, international development and education. This is not a coincidence. It is a pattern.
In this article I break down each mistake: the story, the actual cost, and the practical lesson that prevents it.
Mistake 1: starting from the tool instead of the problem
The story: The CEO of a medical device company, 15 years in the market, read an exciting article about ChatGPT Enterprise and bought 50 licenses. Three months later he asked the team: "So? How are you using it?" The answer was silence.
The cost: About NIS 80,000 a year on licenses, with no defined use case, no KPI, no measurement. A whole year of partial, random use. Most of the licenses became a subscription nobody opens.
The practical lesson: Before you buy a tool, define three specific use cases that the tool is supposed to solve, and a measurable KPI for each. If you cannot phrase them in three lines, you are not ready to buy.
Mistake 2: not involving the team that will actually use the tool
The story: A restaurant chain, 5 branches. The owner brought in an AI based inventory management system. The CTO implemented it. The managers in the field, who had managed inventory in Excel for 10 years, were not involved. After two months, 4 of the 5 branches "forgot" to enter the recommendations into the system.
The cost: NIS 140,000 on a system that in practice serves one branch. The owner insisted on trying again with "intensive training", another NIS 30,000. Same result.
The practical lesson: Bring at least one user from the field to the POC meeting with the vendor, before you sign. Ask them: "Does this solve a real pain for you, or add a step?" If they hesitate, you have a usage problem.
Mistake 3: believing the vendor's ROI promises (without measuring yourself)
The story: A veteran importer from China. The vendor promised: "The system will cut 30% of your inventory costs." A year later the importer could not tell whether that had happened. He had no baseline recorded before the implementation.
The cost: NIS 95,000, and no tool to prove ROI or lack of ROI. The vendor claimed yes, the customer suspected not, but without baseline data the discussion could not be settled.
The practical lesson: Before an implementation starts, write down three numbers: Time spent on the task today, error rate, and direct cost. Without a baseline, every discussion about ROI is an argument about feelings.
Mistake 4: neglecting data hygiene before the implementation
The story: An importer from China, 20 years in the field, wanted an AI agent to manage the supply chain. His data was scattered across 4 different Excel files, with supplier names written differently in each ("Shanghai Textile", "Shanghai Textiles", "Shanghai T.", "ST Co."). The agent that was implemented worked, and made scary mistakes because it thought these were 4 different suppliers.
The cost: NIS 60,000 for the project plus about NIS 40,000 more afterwards on a data cleanup that had to be done in advance. Garbage in: garbage out.
The practical lesson: Before implementation, budget two weeks to a month for a data audit. It is not a sexy stage, but it is decisive. If your data is fragmented, contradictory or incomplete, fix it first.
Do you recognize yourself in one of the mistakes? It is not too late. A free Readiness Assessment call: 30 minutes. Together we will check where you are exposed, before anyone sends you a quote.
Or directly: phone 052-331-3334 · WhatsApp
Mistake 5: letting the CTO or an outside vendor decide without the business owner
The story: A manufacturer of equipment for the real estate sector, with an active owner for years. In this case the owner "let the CTO handle" the AI implementation. The CTO chose a technically excellent tool, but one that did not fit how the company markets and prices to large customers.
The cost: NIS 160,000 on a system the company used for only half a year before abandoning it. The owner discovered only in hindsight that the system did not support the sensitive workflow of large sales.
The practical lesson: The decision to implement AI is a strategic decision, not a technical one. Technology supports it, but the owner must be in the room. If you do not understand the use case, keep asking until you do.
Mistake 6: Big Bang instead of a pilot
The story: A private education network with 8 branches chose to implement an AI system for managing learning tracks inall 8 branches at once, in order to "save time". It was about 7 months of chaos: confused coordinators, parents who received contradictory information, and staff turnover.
The cost: Beyond the licensing cost, by the CEO's estimate the network lost about 12% of its students for the next semester, which translates to several hundred thousand shekels of lost revenue.
The practical lesson: Always a pilot. One branch, one team, 30 days. The extra effort of running a small pilot is worth 10 times the risk of a company wide rollout that fails.
Mistake 7: forgetting change management
The story: A business development company with a presence in Europe and the Emirates implemented an AI tool for analyzing leads. Technically, it worked excellently. Business wise, six months later the sales managers were still working in Excel like in 2019. They did not trust the tool's recommendations, because nobody bothered to explain to them how it reaches them.
The cost: NIS 110,000 on a system that was not in use, even after a year.
The practical lesson: Change management is 15% of the budget, not 2%. That includes face to face training (not video), a written SOP, and an internal Champion who answers the team's questions for at least two months. Without that, the technology is wasted.
5 early warning signs: how to know you are on the way to a mistake
Before you sign, stop and check. If two or more of the following signs are present, stop the process until you resolve them:
- ⚠️ You cannot phrase the use case in two sentences a sign you do not yet understand what you are buying.
- ⚠️ The vendor avoids committing to a measurable KPI a sign there is no real promise, only hope.
- ⚠️ You do not know any vendor customer who resembles your business a sign you are the vendor's pilot, not a customer.
- ⚠️ No employee from the field was present at the demo meeting a sign the purchase is disconnected from the actual users.
- ⚠️ The offer states only year 1 costs a sign that in years 2 to 3 you will be surprised (for the worse).
Summary: 18 years in the field taught me one thing
The technology is almost never the problem. The problem is always the decision of how, when, with whom, and at what pace.
Established businesses in 6 different industries (restaurants, importing from China, medical devices, real estate, international development and education) repeat the same 7 mistakes. Not because they are less smart. Because they did not work with someone who had seen this movie before.
At Forward, since 2008, this is exactly what we do: we give established businesses the experienced eye that separates a successful implementation from a NIS 100,000 mistake.
A free Readiness Assessment call: 30 minutes. Before you receive an offer, let us neutralize 3 or 4 of these mistakes. Cheaper than fixing afterwards.
To schedule a Readiness Assessment
Or directly: phone 052-331-3334 · WhatsApp
