What happened
At twenty-nine, I delivered 241 per cent sales growth at Electrolux. I was dismissed within the year. The result had broken the team that produced it.
I tell this story because I was responsible for what happened. The growth was real. So was the damage to the people who produced it, and that damage is why I lost the job.
What it taught me
Results are not the same as progress. A result is a number at the end of a year. Progress is whether the people who made that number can make the next one, and still want to.
That year I had the result. I did not have the progress, and the first does not last without the second.
I would like to say the lesson changed me completely. It changed me less than that. I still move too fast, and I still default to outcomes. A strong number still holds my attention, and I have to work to look past it to the people behind it.
Why it matters more now
AI now drafts plans, and it drafts them from the numbers it is given. It does this quickly, and it writes well. It is capable enough to be trusted too early. Sounding right is a style, not a signal. Finding an answer is rarely the hard part any more. The hard part is knowing which answer to act on.
A plan can rest on a strong number and say nothing about what the number cost. Those costs may be missing from the data. Overtime may be recorded separately. A seller may not tell the team about a conversation with a recruiter.
AI can speed up existing work. People still decide what needs to change. People still have to check what a strong number cost the team. A draft based on that number may leave those costs out. The person who sends the plan still owns what it asks of people.
A sales target for next year
An example I made up.
The situation
A country sales head at a consumer goods distributor had to send next year's sales target to her regional director by Friday. Her team had grown sales strongly this year, and the region wanted the same growth again. The target would decide how hard the same people worked for another twelve months. If it was set too high, she would not see the cost in the sales report. She would see it in the people who left.
What the AI produced
She asked SwiftletAI to draft the target from this year's sales data. It answered:
"Raise next year's target by 40 percent. This year the team grew sales by 38 percent with no extra headcount. I am 87 percent sure."
The draft was clear, and the sales figures in it were correct.
The check
She asked her team leads for the hours their people had worked this year. Then she opened the list of who had left or asked to move to another team.
The hours showed overtime in most months. The list showed that the growth had come with resignations, each one replaced, so headcount looked level. One team lead added that two of the senior sellers had been talking to recruiters.
None of this was in the data SwiftletAI had been given. It had sales and headcount, so it read this year's growth as something the same team could repeat and then exceed.
The decision, and who was told
She sent a lower target, with one new hire to share the work. She wrote to the regional director with her figure and SwiftletAI's side by side. In the same message she gave the reason they differed. This year's growth had come from overtime and had cost the team people, and the AI's data held neither fact.
She did not set the AI's figure aside because it came from an AI. She set it aside because of what its data did not hold. She didn't hide the AI's draft. The director could see both figures and the reason for the gap, and decide from there.
What happened after
Three months later, the two senior sellers who had been talking to recruiters were still on the team. No sales figure shows that they stayed. That is the progress a number leaves out.
What I do now
The work, for me, is noticing earlier and listening longer.
I give keynotes, briefings and workshops on this.