In the summer of 1999, Procter & Gamble told its shareholders that it was changing almost everything at once. Organization 2005, in the words of its chief executive, Durk Jager, meant “far-reaching changes in structure, work processes and culture.” Regions would give way to global business units built around products. People were asked to go for stretch targets, and new reward systems put more of senior management’s pay at risk. About 15,000 positions worldwide would be affected.
The chairman, John Pepper, called it “the boldest change effort in Procter & Gamble’s history.”
On March 7, 2000, P&G warned that its third-quarter earnings would fall 10 to 11 percent from a year earlier. Its shares lost 31 percent that day, the Associated Press reported, dragging on a Dow that fell about 374 points that day.
The illusion of speed
By August, A.G. Lafley had replaced Jager, and he and Pepper gave shareholders their diagnosis. In March the company had blamed raw-material and manufacturing costs. Now they wrote: “We tried to do too much too fast.” The new structure itself, they added, “remains fundamentally right.”
In 1995, John Kotter of Harvard Business School had described that trap. “Over the past decade, I have watched more than 100 companies try to remake themselves into significantly better competitors,” he wrote in the Harvard Business Review. A few succeeded and a few were utter failures. Most fell “somewhere in between, with a distinct tilt toward the lower end of the scale.” His main lesson: “Skipping steps creates only the illusion of speed and never produces a satisfying result.”
In his 1996 book Leading Change, Kotter set out eight steps, which he later listed as “establishing a sense of urgency, creating a guiding coalition, developing a change vision, communicating the vision for buy-in, empowering broad-based action, generating short-term wins, never letting up, and incorporating changes into the culture.” Urgency and short-term wins have their own pages.
In Kotter’s observation, change that skips stages tends to stall. The speed isn’t real.
What each stage hands the next
One way to read his list: each stage builds what the next one spends. Without a powerful coalition, Kotter wrote, an effort “can make apparent progress for a while. But, sooner or later, the opposition gathers itself together and stops the change.” A vision announced at one meeting uses “about 0.0001% of the yearly intracompany communication,” and the leaders are “startled when few people seem to understand.”
Of a dozen reengineering efforts he watched, in all but two “victory was declared” when the first major project was finished, after two to three years. “Within two more years, the useful changes that had been introduced slowly disappeared.”
Read P&G’s own later accounts against that list and the fit is tempting. Lafley recalled that by the summer of 2000 leaders “had retreated to their bunkers,” business units blamed headquarters and headquarters blamed the units. That is hindsight, not a test. Nobody ran the same overhaul at a slower speed to compare.
A map drawn from experience
Kotter’s evidence is his own observation, mostly of unnamed companies. A 2012 review by Steven Appelbaum and colleagues noted that his work “was based on his personal business and research experience.” Reviewing 15 years of research, they found support for most of the steps on their own, but “no formal studies” covering the model as a whole. They found no evidence against it either. Its popularity, they wrote, appears to come “more from its direct and usable format than from any scientific consensus.”
A number often quoted alongside it is shakier still. “The brutal fact is that about 70% of all change initiatives fail,” Michael Beer and Nitin Nohria wrote in HBR in 2000. Mark Hughes reviewed five published instances of the 70 percent figure and, in each, found no “valid and reliable empirical evidence” behind it.
Kotter himself later loosened the sequence. In 2012 he wrote that his steps “are often used in rigid, finite, and sequential ways,” and proposed versions that run “concurrent and always at work.”
The paper book comes back
Imagine a clinic manager moving her front desk from a paper appointment book to online booking. She sends one email and runs one training. The most senior receptionist keeps the paper book “just in case.” In week two online bookings rise, and the manager tells her boss the switch is done. By spring, the paper book is back on the counter.
Run the same switch through Kotter’s list and different moments stand out. The senior receptionist was the coalition she never built. The email was a vision communicated once. Week two was a short-term win announced as the end.
Seven hundred agents in a month
Klarna’s version of week two came in February 2024. A month after switching on an AI assistant, the company announced it had handled two-thirds of its customer service chats and was doing “the equivalent work of 700 full-time agents.”“The numbers speak for themselves,” the release said.
In May 2025, chief executive Sebastian Siemiatkowski told Bloomberg, as Fortune reported, that “as cost unfortunately seems to have been a too predominant evaluation factor when organizing this, what you end up having is lower quality,” and Klarna began recruiting human agents. A spokesperson told Fortune the company was “very much still AI-first.”
The record is mixed. In its March 2025 IPO filing, Klarna said internal surveys showed no drop in consumer satisfaction, and the reports here don’t say how “lower quality” was judged. The release didn’t say the rollout was finished; reading it as a victory lap is Kotter’s lens, not Klarna’s words.
Kotter’s own yardstick ran in years. In one of the most successful transformations he saw, the amount of change was scored every year for seven years. The peak came in year five, “fully 36 months after the first set of visible wins.”