Independent newsroom The Wyre News Network OpEd desk

Analysis 5 min read

The Profit Word Companies Keep Redefining

Adjusted earnings began as a reasonable idea, strip out the genuinely one-off item so investors could see the underlying business, and have become something closer to a house style: a recurring exclusion of anything unflattering, repeated so often that the adjustment is no longer the exception but the norm. This piece argues that the practice has quietly shifted the object investors are pricing. When restructuring costs, share-based pay, litigation charges and impairments are routinely waved away as "non-recurring" despite recurring every year, the adjusted figure stops describing the business and starts describing the story management wants told. The argument here is not that every adjustment is dishonest, some are defensible, but that the cumulative effect of the practice has been to train markets to discount the very number that used to anchor valuation: statutory profit.

Listen to this piece 7 min

There is a specific moment worth noticing in most quarterly earnings calls now: the pause after "adjusted" and before the number. That pause is doing work. It is where a company tells you, without quite saying so, that the figure you are about to hear is not the figure its auditors signed off on. It is the figure management would prefer you use instead.

None of this is new in the sense of being a recent invention. Adjusted or "non-GAAP" earnings have existed for decades, and the original justification was not unreasonable. A company that sells a factory takes a one-off gain that has nothing to do with how its actual business performed that quarter. A company hit by a genuine natural disaster incurs costs that will not recur. Strip those out, the argument goes, and investors see the underlying engine of the business more clearly than statutory profit allows.

The trouble is that "one-off" has stopped meaning one-off. Restructuring charges appear most years. Share-based compensation, a real cost that dilutes real shareholders, gets added back as though it were a rounding error rather than a recurring transfer of value from investors to employees. Litigation settlements, impairments, acquisition costs: all routinely excluded, all recurring with enough regularity that describing them as exceptional has become a kind of institutional fiction that everyone involved is choosing not to examine too closely.

The adjustment becomes the business model

What happens when a company adjusts its earnings every single quarter, for years running, always in the same direction, is that the adjustment stops being a correction and starts being the actual reporting practice. The statutory number, the one produced under standard accounting rules and subject to audit, becomes the footnote. The adjusted number, produced by management with no external constraint on what gets excluded, becomes the headline.

This is not a technical quibble. It changes what investors are actually pricing. A share price built on statutory earnings reflects, however imperfectly, the business as its accountants are willing to certify it. A share price built on adjusted earnings reflects the business as management would like it to be understood, net of whatever items management has decided do not count. Those are not the same exercise, and treating them as interchangeable is where the damage happens.

The word "adjusted" implies a correction toward accuracy. In practice it has come to function as a licence: once a company establishes that certain categories of cost are routinely excludable, it can produce a version of its performance that is more flattering in every period, not just the unusual ones. The word has been redefined by use, quietly, without anyone voting on it.

Why investors keep buying the story anyway

It would be easy to blame this entirely on management and walk away satisfied. But adjusted metrics persist because investors, analysts and financial media have found them convenient. They are easier to compare across companies than statutory figures riddled with one-off items that genuinely do differ by business. They often show growth where statutory numbers show stagnation, which makes for a better story on the earnings call and a more defensible price target in the analyst note.

Analysts build models around adjusted figures because that is what companies guide toward, and guidance sets the terms of the conversation. A company that reports adjusted earnings and beats its own adjusted guidance gets rewarded with a headline about a beat, even if statutory profit tells a flatter or worse story. The market has, in effect, agreed to grade companies against a target the company itself designed.

This is the part of the arrangement that deserves more scepticism than it gets. Nobody outside the company decides what counts as an adjustment. There is no independent body approving the exclusions, no consistent rule across companies or even across quarters within the same company for what qualifies as non-recurring. The definition can shift when it is convenient, and it usually shifts in one direction.

Pricing a story instead of a business

When every loss becomes an adjustment, the thing being valued in the market is no longer the business as it actually operates: its costs, its liabilities, its real dilution of shareholders through compensation schemes. It is the narrative the business tells about itself, refined quarter after quarter to exclude anything that complicates the growth story. Investors who price shares off adjusted earnings are, whether they realise it or not, buying the story rather than underwriting the balance sheet.

This matters most acutely at the point where the story and the business diverge, which tends to be exactly the point at which investors most need accurate information. A company under real financial stress has the strongest incentive to lean on adjusted figures, because the statutory numbers are the ones showing the strain. The metric that was supposed to reveal underlying performance becomes, in the moments that matter most, the metric best designed to conceal it.

None of this requires assuming bad faith on the part of every finance department. Some adjustments really are defensible, and treating all non-GAAP reporting as fraud would be its own kind of distortion. But the direction of travel over time, toward more frequent adjustment, broader categories of exclusion and greater reliance on adjusted figures in official guidance, has not been neutral. It has consistently favoured the more flattering number over the more accurate one, and it has done so with the market's quiet cooperation.

What a sceptical read looks like

The corrective is not complicated, even if it is unfashionable. Read the statutory figure first. Ask what specifically has been excluded to arrive at the adjusted number, and ask whether the excluded item has, in fact, recurred in prior periods despite being labelled non-recurring. A share-based compensation add-back that appears every single quarter is not describing an exception; it is describing a cost of doing business that the adjusted figure has simply chosen not to count.

None of this means ignoring adjusted earnings altogether. Management commentary about underlying performance can be genuinely informative when the exclusions are narrow, consistent and clearly explained. The problem is not that adjustment exists as a concept. The problem is that the word has been stretched to cover so much recurring cost that it now functions less as clarification and more as a permanent discount applied to bad news.

Investors who want to price the business rather than the story it tells about itself have one option available to them that requires no new disclosure regime and no regulatory intervention: read the number companies did not choose to lead with.

Wyre's opinion bylines are editorial personas of Floof Digital LLC, not separate members of staff. Essays are produced with AI assistance under human editorial direction. How Wyre works.

More Opinion

From the same desk

Analysis

The Funding Arrived After The Exploit

This week two startups raised money to secure "agentic AI" systems, Reco taking $55 million and Rig Security emerging from stealth with $12 million, both pitched as defences against autonomous AI agents run amok. In the same news cycle, SecurityWeek reported hackers already using ChatGPT's custom GPTs inside ClickFix attacks, and OpenAI itself pulled the launch of GPT-6.1 Astra to publish safety cases rather than ship it. The argument here is simple: the capital raised this week is chasing a threat model that attackers, not analysts, defined first, and the rest of the week's breach reports, from a Pentagon personnel agency losing data on 3 million people to a 1,500-location pizza chain confirming a cyberattack, shows the funding is arriving into an environment where far more basic failures are still going unpatched.

5 min

Analysis

Thirty-Four and a Half

The federal government has cut the 2031 fuel economy target to 34.5 mpg, and the trade coverage has settled on a single, comforting framing: cars will get cheaper because manufacturers no longer have to engineer toward a tougher number. That framing is true and also incomplete. A lower mpg floor lowers the compliance cost baked into a new car's sticker price today, but it does nothing to lower the amount of petrol that car will burn over the ten or so years someone drives it. The saving is real and it is one-time. The cost it locks in is real and it repeats every time you fill the tank, for as long as you own the thing. Nobody in Washington put those two facts in the same sentence this week.

5 min

Analysis

The Roadmap Skips The Present

OpenAI has told the world that 80 to 90 percent of its research already targets GPT-7 and whatever comes after it, even as tens of thousands of security probes suggest its Hugging Face incident was not a one-off but an opening chapter. That gap between where the company's attention sits and where its actual exposure sits is the story this week's reporting keeps returning to. GPT-6 Astra is already wired into a robot cleaning an unfamiliar kitchen and diagnosing botched IKEA shelves, Goldman Sachs expects Big Tech to spend $1.2 trillion on AI infrastructure by 2027, and insurers are already saying AI is raising healthcare costs. None of that pace has been matched by a comparable urgency about the holes in what's already shipped. Speed is winning the argument with security, and OpenAI's own numbers are the evidence.

5 min