The Right Way to Deliver Bad Earnings News

On July 14, Arvind Krishna did something CEOs spend their entire tenure hoping to avoid. Eight days before IBM was scheduled to report second-quarter earnings, he published an open letter to investors warning that the results would fall well short of expectations. By the closing bell, IBM’s stock had fallen roughly 25%—the worst single day in the company’s 115-year history. Within days, Wall Street was debating whether activist investors might try to break up Big Blue.

It would be easy to file this away as a cautionary tale about candor: the CEO who volunteered bad news and got a historic selloff for his trouble. That’s exactly the wrong lesson. Having spent more than two decades covering companies in crisis as a journalist and nearly two more advising executives on how to communicate through them, I can tell you the instinct inside many C-suites at a moment like this is to wait—close the quarter, perfect the script, and deliver the bad news on your own stage, flanked by your CFO and reciting a rehearsed Q&A.

The research on corporate disclosure says that instinct to wait is an expensive one. Krishna was right to go early, and no letter, however worded, was going to spare IBM from a drop in the stock price. But how a warning is written shapes everything after: whether the stock keeps sliding, whether investors believe the future results management gives them, and how much time they allow for the fix to work.

IBM’s warning left gaps—and markets fill gaps with fear. The letter never sized the miss, so investors sized it themselves—and rounded up, harshly. It referred to big deals that hadn’t closed but never answered the question the selloff turned on: Had those deals merely slipped, or had demand dried up? And it spent a quarter of its words promoting new products, so a profit warning closed like a sales brochure.

Over the next eight days, the stock bled another 5% while the market wrote its own account. When the answers arrived on IBM’s scheduled call, they were smaller than the fear: a modest trim to full-year guidance, cash flow intact, a third of the slipped deals already closed. That distance—between what the market imagined and what IBM eventually said—is what a flawed warning costs. What Krishna got right and what he got wrong are equally instructive. (IBM did not respond to multiple request to comment on this article.)

Why Going Early Was Right

Krishna was not the first chief executive to face this choice and take it. On January 2, 2019, Tim Cook published an open letter to Apple’s investors announcing the company’s first revenue warning since 2002. Apple cut its quarterly revenue guidance from a range of $89 billion to $93 billion to approximately $84 billion, attributing most of the shortfall to weaker-than-expected iPhone sales in Greater China and a slowing Chinese economy. Apple’s stock fell about 10% the next day.

Cook, like Krishna, chose to break the news himself rather than let the market break it for him, and the two men made the same correct call to go early. Where their letters diverged was in the execution, and that gap is what separates a bad day from a worse one of the company’s own making.

Contrast those cases with a company that saves the surprise for earnings day. Netflix told investors in January 2022 to expect 2.5 million new subscribers that quarter. But in mid-April, the streaming giant instead reported a loss of about 200,000, its first decline in more than a decade. Netflix’s stock fell roughly 35% in a single day, not because it concealed anything. The stock fell because an optimistic forecast ran into a business that had turned, and investors heard it only at the earnings report—not before.

Bad news does not age well inside a company. Research on company-specific crashes finds they are rarely caused by bad results alone; they are caused when bad results are withheld, in whole or in part. Managers sit on negative information until it exceeds what can be hidden. When released all at once, the price collapses. Firms whose communications conceal problems rather than reveal them carry measurably higher crash risk into the following year. The choice IBM faced was never between a bad day and a good one. It was between a controlled detonation now and an uncontrolled one later.

Markets are also harsher on silence than most executives believe. A study of 545 companies that missed their scheduled earnings dates found the average announcement of a delay cost 6.3% in a single day. But delays with no explanation at all cost 10.4%, more than any stated reason, including accounting problems. Those companies had missed their reporting dates—a different issue than IBM’s—but the mechanism is similar. Given nothing, investors write the narrative themselves, and they write a darker one than the truth.

Going early also ends a quieter risk: the longer that materially bad news sits undisclosed, the longer insiders hold an informational advantage over the shareholders they serve, with all the legal and reputational exposure that implies.

IBM’s board had this debate before the letter went out: warn early, or wait for the scheduled call, The Wall Street Journal later reported. Any general counsel can make the case for waiting: the books aren’t closed; a preliminary number that later moves is fodder for a securities suit; a partial disclosure can invite a duty to keep updating. The board warned early, rightly. When they take their lawyers’ advice, companies are not deciding whether to go early but how: state the direction and rough size of the damage now, clearly labeled as preliminary, and save precision for the closed-books call. Go early enough to correct the record, and no earlier than the facts can support.

Where the Letter Went Wrong

The letter’s first flaw was hiding in plain sight, in the numbers. IBM reported selected preliminary figures that, eight days later, proved exact: revenue up 1%, operating earnings per share up 5%, margins and cash flow all laid out. Read cold, those numbers look almost routine—and that was the problem.

Nothing in the letter said what the shortfall meant for the full-year revenue guidance IBM had issued in January and reaffirmed in April, in dollars or percentage points. The letter showed the results and left the disappointment for the reader to size. A 25% decline against a letter whose own arithmetic looks benign suggests the market was pricing more than the stated arithmetic, including what went unsaid.

Second, the remedy got one sentence: “we are undertaking new initiatives and accelerating others.” That is reassurance, not a plan. Research on the language of profit warnings finds that troubled companies often write them to sound as routine as possible, acknowledging the problem while saying little about what management will do about it.

Veteran investors have a rule for exactly this moment: Sell on the first profit warning, because there is usually a second. The way a company breaks that expectation is specificity—named actions, owners, dates. Studies of management disclosure find the market rewards forward-looking statements it can verify and discounts the ones it cannot. At IBM, the fuller plan arrived on July 22. But investors weren’t pricing the company’s roadmap that week. They were pricing whether management had command of the problem, and that answer took eight days.

Third—and this is what I believe cost IBM the most—the letter never answered the question the market actually repriced: Is this a timing problem or a demand problem? Krishna wrote that large deals “failed to close on the timelines we expected.” The obvious follow-up: Have they closed since? How much of the slipped value remains committed?

If the answer was favorable, omitting it was extraordinarily expensive, because it allowed a deal-slippage story to be read as a story about AI structurally displacing IBM’s business. One former IBM executive, in a LinkedIn post quoted by The Wall Street Journal, said that the selloff wasn’t about a quarter. It was, he wrote, “the market repricing a question: can a 115-year-old enterprise company lead the agentic era, or merely survive it?”

By the July 22 earnings call, about a third of those deals had closed—“an indication, not yet full evidence, but a good indication,” Krishna said on the call, “that this was deferral and not destruction.” Whether that evidence existed on July 14, the record doesn’t say. And the letter did tell investors when they would learn more (the July 22 call). What the letter didn’t give them was a yardstick for the eight days in between—how many deals had slipped, what share normally close, and how quickly. Krishna had those numbers, which he recited on the call.

Finally, a quarter of the pre-announcement warning was devoted to promotion: a product announced seven weeks earlier, naming the same 11 early adopters; a quantum computing foundry, a $10 billion investment roadmap. That impulse is understandable; no leader wants a document this painful to end without hope. But when hope is stapled onto a profit warning, the market sees through it.

The research here is unforgiving. Markets price abnormal tone—optimism beyond what the reported numbers support—so a promotional message inside a profit warning reads as spin, not news. Investors penalize communication whose emotional register is at odds with the news it delivers. And longer releases tend to dilute the market’s response to the substantive message. In the research, complexity itself is a red flag because cluttered disclosure correlates with bad news. The launch material belonged on July 22. And when it got there, it worked. The same roster and roadmap, delivered on the scheduled call, drew engagement, not objection. The final analyst question was about the new product, not the miss.

The Playbook

For CEOs, CFOs, and chief IR officers, the IBM episode offers five lessons:

Go early and go once. The first warning must also be the complete one. Companies that dribble out bad news in installments teach the market to expect another shoe to drop, and it prices accordingly. Nvidia is the model: on August 8, 2022, more than two weeks before its scheduled report, it pre-announced that quarterly revenue would come in at roughly $6.7 billion against the $8.1 billion it had guided. The company identified gaming as the principal cause and took the hit at once, so the earnings call weeks later could be about recovery rather than shock.

Intel shows what the drip costs. A soft first-quarter outlook came in January 2024; the heavier blow came in August with a dividend suspension, lowered forecast, and a 15% workforce reduction. The two disclosures covered different quarters and a changing restructuring—but the market prices the pattern, not the intent. By the second announcement the market had learned to brace, and the stock fell about a quarter in a single day.

Size the miss yourself. State the shortfall against your own prior guidance, in numbers, in the first hundred words. Whatever figure you withhold, the market will replace with a worse one.

FedEx broke the rule outright. On September 15, 2022, it withdrew its full-year earnings forecast entirely rather than resetting it—arguably because it could no longer forecast reliably. FedEx’s CEO told investors only that “global volumes declined as macroeconomic trends significantly worsened later in the quarter.” Pulling the number instead of restating it, whatever the reason, handed the market a hole where the floor should have been, and the stock fell 21%, its worst day in more than four decades.

Never leave a vacuum. Set an upper limit on the full-year impact, even roughly, and don’t let the written warning be management’s only voice. Investors under-process one-way disclosure, so answer the obvious questions inside the document and put a senior leader in front of investors quickly. Cook gave a TV interview the day his letter published. IBM Vice Chairman Gary Cohn appeared on CNBC the next morning, relaying Krishna’s view that the pullback was “relatively temporary”—a characterization, not evidence. The first numbers behind it didn’t arrive for eight days.

When Best Buy warned in October 2012 that its quarterly results would disappoint, the same release announced an investor day where its new CEO would lay out its turnaround plan (“Renew Blue”) and take questions. The formal earnings release followed a week later, and the stock fell about 13% on the earnings news. As the turnaround took hold, Best Buy went on to be one of the best-performing stocks in the S&P 500 the next year.

Own the cause, then name the cure. IBM got the first half right: it blamed external forces only where they were verifiable and industry-wide—the conditions research shows separate believable external explanations from self-serving ones—and paired them with an admission (“this quarter we faltered”). What IBM didn’t provide was the cure.

Ford shows what finishing it looks like. Amid mounting losses that would reach $12.7 billion in 2006, it published a named turnaround plan, “Way Forward,” that accelerated when Alan Mulally arrived eight months later. Ford moved to raise roughly $18 billion, and in time nearly $23.5 billion, by mortgaging its assets before the credit markets froze. Under Armour got only the first half: in January 2017 it owned the cause (“challenges and disruptions in North American retail”) but offered little measurable detail on how it would reverse a growth rate that had just fallen from above 20% to a guided 11-to-12%, and a CFO departure in the same release deepened the doubt. The stock fell about 23%.

Don’t sell while you’re apologizing. Keep the proof points that bear on recovery but cut everything that reads like a press release. Markets price abnormal tone and a promotional surge inside a warning is graded as spin. Rolls-Royce is the counter-example: in February 2016 it delivered a weak outlook and halved a dividend it had paid for a generation. Management framed it plainly as clearing the decks rather than dressing it up, and the shares rose about 10% over the following week. A warning must sound like a warning.

. . .

For IBM, the scoreboard is already running. The final results matched the letter’s preliminary figures exactly and the full-year reset was modest. The shares rose 4% in the following days, but if investors fully believed “deferral and not destruction,” the stock should have reversed more of the 25% plunge. It didn’t, and the lack of a reversal after reassuring answers is what a credibility discount looks like. The markets score management’s story at zero until further results confirm it.

Credibility is a balance-sheet asset, accumulated in deposits and destroyed in withdrawals. Krishna’s early warning may yet prove to be a deposit, but the letter’s omissions spent some of the principal. The deeper lesson for every leadership team: The market will forgive you for a bad quarter. It is far slower to forgive discovering that you knew and said little or nothing. That is the experiment IBM, to its credit, chose not to run.

Copyright 2026 Harvard Business School Publishing Corporation. Distributed by The New York Times Syndicate.

Explore AAPL Membership benefits.