Jim Collins published Good to Great in 2001 and named eleven companies that had outperformed the market by a factor of seven over fifteen years. The book became the dominant management text of its decade. Two and a half decades later, three of those eleven — Circuit City, Fannie Mae, and Wells Fargo — are case studies in how disciplined operators miss the future they are sitting inside.
The most common reading of these failures is governance. Greedy management, weak boards, fraudulent culture. Those readings are not wrong, but they are downstream. The deeper failure is disciplinary. These companies did not lose because they were undisciplined. They lost because their discipline was the wrong shape for the world they were operating in.
Circuit City — the discipline of the prior decade
Circuit City's collapse is usually told as an Amazon story. It is not. By the time Amazon was a serious threat in consumer electronics, Circuit City had already broken its own retail discipline. The 2003 decision to fire its highest-performing commissioned salespeople and replace them with hourly staff made sense on a spreadsheet — labor costs were the largest line item, and commissions were the largest variable in that line. It made no sense as a strategic read of where the consumer-electronics category was heading.
The category was bifurcating. At one end, increasingly commoditized digital goods sold on price. At the other, increasingly complex installed home systems sold on expertise. Both ends had clear signal. The middle — undifferentiated big-box retail — was getting squeezed from both sides. A predictive read of the operating environment would have shown that, and shown it years before the eventual collapse.
The failure was not that Circuit City lacked information. The failure was that its discipline had been built for the prior decade — when scale and selection won — and had not been re-pointed at the operating environment it was actually living in. The Hedgehog Concept is dangerous when the world your hedgehog evolved for stops existing.
Fannie Mae — the discipline that priced out judgment
Fannie Mae was the longest-tenured Good-to-Great company in 2001. By 2008 it was effectively nationalized. The standard story is political pressure to expand subprime exposure. That story is true but partial.
The deeper story is that Fannie Mae's risk discipline had been increasingly automated, and the automation had been calibrated on a regime — long-running housing-price appreciation — that was about to end. The signals that the regime was ending were visible to anyone willing to look at human data instead of model output. Realtors had started talking about buyers who could not actually afford the homes they were closing on. Mortgage brokers were moving paperwork that made no operating sense. The default patterns inside subprime tranches were starting to shift in 2006 in ways the models did not capture because the models had no analog regime to draw on.
A predictive discipline would have treated those signals as a cleavage to write a scenario against. A scenario in which the decade-long regime broke and the model assumptions decoupled from reality. The discipline that was actually in place treated the models as ground truth and the human signal as noise. The model held until the moment the regime changed, and then it didn't, and the company that had been Good to Great for thirty years had to be rescued in a weekend.
Wells Fargo — the discipline that ate its own context
Wells Fargo's account-fabrication scandal broke in 2016. The most common reading is incentive design. Aggressive cross-sell quotas produced fraudulent accounts. That reading is true at the surface and misses the operating discipline that allowed those quotas to run unchecked for nearly a decade.
Wells Fargo's executive layer had built a measurement discipline so precise — accounts per household, products per customer, minutes per branch visit — that the metrics themselves became the operating environment. Internal signal that the metrics were producing perverse behavior was visible from at least 2009 onward. Branch managers were raising flags. Internal audit was raising flags. Customers were filing complaints in patterns that, read in aggregate, were unmistakable.
The discipline at the top did not have a posture for treating internal signal as a forecast input. Internal signal was treated as an operating issue to be managed, not as a leading indicator that the strategy was producing exactly the wrong outcomes. By the time the signal became external — first as journalism, then as regulatory action — the optionality to correct quietly was gone.
The common thread
Read the three failures together and the pattern is not greed, not weak boards, not bad luck. It is a missing disciplinary layer — the layer that converts a stream of weak signals into scenarios and re-prices commitments against them. Each company had a strong operating discipline. Each lacked a discipline for the questions that operating discipline could not answer.
Circuit City had inventory discipline, store-format discipline, sales-floor discipline. It did not have a discipline for "what if the category bifurcates and the middle disappears." Fannie Mae had model discipline, capital-allocation discipline, secondary- market discipline. It did not have a discipline for "what if the regime our models were trained on stops applying." Wells Fargo had measurement discipline, compensation discipline, branch discipline. It did not have a discipline for "what if our metrics are producing exactly the customer behavior we don't want."
The disciplines they had were the disciplines that had made them Good to Great. The disciplines they lacked were the disciplines that would have kept them there.
The lesson
The point of this is not that Collins was wrong. The point is that operating discipline and predictive discipline are not the same thing. Operating discipline runs the company you have. Predictive discipline detects the company you will need to become. A company with strong operating discipline and weak predictive discipline will execute its way into obsolescence with great precision.
The Good-to-Great companies that are still great today — and there are several — share something the failed three did not. A working habit of treating their operating environment as something they had to read continuously, not something they had configured once and could now optimize against. That habit is the practice that the Predictive Planning Loop names and codifies.
It is also, almost without exception, the discipline that organizations imagine they have and don't.