Scope: IBM, fiscal years 1993 to 1995, with 1996 shown separately as an early check. This grades decisions and audited results while Louis V. Gerstner Jr. was CEO, using a rubric published below before the verdict. It isn’t a career ranking or a biography, and it doesn’t argue that one person saved IBM. Evidence cutoff: filings available through IBM’s March 1997 10-K, read on November 8, 2026.
Related reading: Is Home Depot Stagnating? 8-Quarter Read
The usual Gerstner story arrives already graded: a company written off, an outsider at the top, a comeback. Boards keep reaching for that story when they hire a turnaround CEO. The trouble is that it skips what a board actually has to decide: how much of the recovery came from the new leader’s decisions, and how much came from cuts already made, a cost base already being reset, a pension plan, currency, and a market that was moving anyway.
So I went back to the four 10-K filings IBM sent the SEC from 1994 to 1997, plus filings from three peers of the same era. Here’s what the numbers support and what they don’t.
Bottom line (provisional): B, 70.6 out of 100, sitting on the B/C line. 85% of the rubric could be assessed. Institutional durability/succession is N/A because this review stops before the 2002 succession. That’s why the grade is provisional and not a career verdict.
What holds up: operating margin before charges went from 0.5% in 1993 to 13.1% in 1995, well ahead of the peer median. Recurring cash after investment turned from negative before 1993 to about $5 billion a year. What drags the grade down: services growth was already running above 30% a year before he arrived, the predecessor had booked $15.3 billion of restructuring charges, gross margin never got back near its 1991 level, a pension swing helped the later years, and “one-time” workforce charges kept coming back. A one-point change on any single dimension flips the letter between B and C.
The Rules Before the Verdict
The rubric is fixed before looking at the case and isn’t tuned to fit it. The weights and anchors below are Stagnation Assassins’ editorial rubric for grading CEO decisions. It isn’t a framework from any of Todd’s books, and nobody at IBM used it. It scores decisions and results within the declared role and window. It never scores wealth, charisma, popularity, pay, or stock-price headlines.
| Dimension | Weight | 0 looks like | 2 looks like | 4 looks like |
|---|---|---|---|---|
| Operating economics / durable customer value | 25% | Comparable economics and observable customer outcomes both fall behind peers | Economics roughly track peers, or improve in only one period; customer evidence mixed | Comparable economics beat peers in at least two annual periods, with corroborating customer outcomes |
| Cash resilience / capital allocation | 25% | Recurring cash shortfall and funding fragility worsen, excluding asset sales and new financing | Liquidity stabilizes; recurring cash generation or investment return mixed | Recurring post-investment cash and funding headroom improve over two annual periods without counting disposals or refinancing as operating cash |
| Strategy / execution against disclosed commitments | 20% | Most material dated commitments missed, with no corrective response | Mixed delivery; at least half of milestones met or openly revised | Most milestones delivered, measurable customer response, documented adaptation |
| Governance / reporting / stakeholder costs | 15% | Material reporting failures persist; stakeholder harm omitted | Costs disclosed and remediation present, outcomes mixed or incomplete | Independently corroborated remediation; stakeholder costs quantified with mitigation outcomes (layoffs are not celebrated as success) |
| Institutional durability / succession | 15% | Gains reverse within two years of a transition, traced to fragile systems | Some gains persist, some reverse | Gains persist for at least two annual periods through a leadership transition |
How the grade works: Scores of 1 and 3 fall between the anchors, and the reason is written next to each one. Missing evidence is N/A, not a 0 and not a “neutral” 2. Coverage is the sum of the weights that can be assessed. The overall grade is withheld below 85% coverage, or if operating economics, cash, or governance is N/A, or if any core dimension has Low confidence. Grade = 100 × Σ(weight × score/4) ÷ covered weight. A ≥85, B 70 to 84.9, C 55 to 69.9, D 40 to 54.9, F below 40.
Evidence labels used below: [Filing] means audited or SEC-filed figures. [Company assertion] means management’s own statements in those filings. [Analyst interpretation] means an independent writer’s view. [Calculation] means arithmetic on filed figures. [Inference] means my judgment, which you can disagree with.
Windows and Sources
- Tenure: IBM’s own archive says Gerstner was CEO and chairman from April 1993 to March 2002, and that Samuel Palmisano succeeded him as CEO on March 1, 2002 (IBM history). That archive confirms the dates. It is not evidence of what caused the recovery.
- Baseline: FY1990 to FY1992, the three years before he was appointed.
- Evaluation, years 1 to 3: FY1993 (a partial year: he started in April, so Q1 belongs to his predecessor), FY1994, and FY1995.
- Early check: FY1996, shown separately. It doesn’t drive the scores.
- Not assessed: years after 1996 and the post-2002 succession. That dimension is N/A.
- Primary filings read: IBM 10-K for FY1993, FY1994, FY1995, and FY1996. Each includes the annual report as Exhibit 13. IBM’s 1997 annual report was checked for restatements: its five-year table repeats the same 1993 to 1996 revenue, net earnings, and total debt. Its per-share figures are restated for a later stock split, so this piece uses no per-share numbers.
- Peer filings: Hewlett-Packard FY1996 10-K, Digital Equipment FY1996 10-K, and Unisys FY1996 10-K.
- Independent context: D. Quinn Mills, “The Decline and Rise of IBM,” MIT Sloan Management Review, July 15, 1996. Mills notes the article is based on extensive interviews with IBM executives and personnel and access to some company files, and he thanks John Akers among others. That is useful access, and it also means his sources had their own stake in the story.
What He Walked Into: The Baseline
| $ millions [Filing] | FY1990 | FY1991 | FY1992 | FY1993 (partial year 1) |
|---|---|---|---|---|
| Revenue | 68,931 | 64,766 | 64,523 | 62,716 |
| Net earnings (loss) | 5,967 | (2,861) | (4,965) | (8,101) |
| Net before accounting changes | 5,967 | (598) | (6,865) | (7,987) |
| Restructuring charges | n/a in filings read | ~3,700 | 11,645 | 8,945 |
| Operating income before restructuring | n/a in filings read | 4,673 | 3,406 | 308 |
| Gross margin | n/a in filings read | 50.5% | 45.6% | 38.5% |
| Total debt (year-end) | 19,545 | 26,947 | 29,320 | 27,342 |
| Cash dividends paid on common | 2,774 | 2,771 | 2,765 | 905 |
Sources: FY1993 10-K five-year comparison and MD&A “Results of Operations” and “Restructuring Charges.” The 1992 net figure includes a $1,900M positive accounting-change effect (income taxes), so the loss before accounting changes was larger, at $6,865M. The 1990 operating detail isn’t in the filings I retrieved, so it’s marked n/a rather than estimated.
Three things in that baseline matter for the grade.
1. The cost reset had already started. [Filing] Restructuring charges were “$3.7 billion in 1991” and $11.6 billion in 1992: about $15.3 billion booked before April 1993. Headcount at IBM and its wholly owned subsidiaries fell from 344,396 at the end of 1991 to 301,542 at the end of 1992, a cut of 42,854, before the new CEO arrived (FY1993 10-K, employee table).
2. The services engine was already running. [Filing] Services revenue, excluding maintenance and the Federal Systems unit, went from $4,144M (1991) to $5,530M (1992) to $7,648M (1993). That’s growth of 33% and 38% [Calculation]. The integrated-services business people credit to the turnaround was already the fastest-growing line before the turnaround began.
3. The margin collapse was a pricing and mix problem, not just a cost problem. [Filing] Gross margin fell from 50.5% to 38.5% between 1991 and 1993. The FY1993 MD&A attributes this to “hardware pricing pressures and the company’s shift to services revenue.”
What Management Knew in Early 1994
This panel uses only what IBM told investors at the time, in the FY1993 10-K filed March 28, 1994. Hindsight comes later, in the results section. Everything here is a [Company assertion].
- Demand: “Industry demand has slowed in recent years, and the industry continues to suffer from excess capacity and sluggish economic growth, particularly in Europe and Japan.”
- Core product: “Mainframe processor revenue and margins declined significantly in the last year, and this trend may continue.”
- Mix: services were “growing rapidly” but at “lower margins than the company’s traditional hardware offerings,” so “it is uncertain when pressure on the company’s cost structure will be diminished.”
- Structure: “The company’s management strategy also shifted in 1993 from the establishment of increasingly autonomous business units–an emerging federation of companies–to remaining an integrated provider of information technology.”
- A commitment you can test: if its demand assumptions held, IBM said it could absorb future productivity costs “without resorting to additional special charges.”
- A second testable commitment: “By 1996, it is expected those actions, begun in 1993, to reduce expense structure will yield approximately $7.0 billion in annual savings.”
Read cold, that’s a company that knew its main profit pool was shrinking, knew its growth engine had thinner margins, and committed to a cost number and a no-more-charges promise. That’s the yardstick for the strategy and execution score.
The Decisions, Dated
| When | Decision or event | What the filings show |
|---|---|---|
| Q2 1993 | Right-sizing program | [Filing] $8,945M restructuring charge in 1993, tied to “restructuring actions announced in the second quarter of 1993.” Q2 1993 net loss: $8,036M. |
| 2H 1993 | Dividend cut | [Filing] Common dividend per share paid: $0.54 in each of Q1 and Q2 1993, then $0.25 in Q3 and Q4. Cash dividends dropped from $2,765M (1992) to $905M (1993). |
| 1993 | Keep IBM integrated | [Company assertion] The strategy shifted from a “federation of companies” to “remaining an integrated provider.” Listed here as one dated decision. Its payoff is tested in the services numbers below. |
| 1993 | Balance-sheet bridge | [Filing] $11,794M of new debt and $1,091M of preferred stock raised. Short-term borrowings cut by $5,247M. |
| 1994 | Sale of Federal Systems Company | [Filing] $1,503M in proceeds, recorded as an investing cash flow, plus a $382M pretax gain in SG&A. |
| July 5, 1995 | Lotus acquisition | [Filing] About $3.2B ($2,880M net cash). $1,840M was written off as purchased in-process R&D, which caused the Q3 1995 loss of $538M. |
| 1995 to 1996 | Share repurchases | [Filing] Common shares repurchased for $4,864M (1995) and $5,810M (1996), plus repurchase of the Series A preferred in 1995. |
| 1996 | Tivoli acquisition | [Filing] $716M net cash. $435M purchased in-process R&D. |
Raw Results, Then the Reconciliation
Statutory numbers first. Adjustments come second and are shown openly, so you can reverse any you don’t accept.
| $ millions | 1991 | 1992 | 1993 | 1994 | 1995 | 1996 (check) |
|---|---|---|---|---|---|---|
| Revenue [Filing] | 64,766 | 64,523 | 62,716 | 64,052 | 71,940 | 75,947 |
| Operating income, as reported [Filing] | 939 | (8,239) | (8,637) | 5,005 | 7,591 | 8,596 |
| + Restructuring charges / purchased in-process R&D [Filing] | ~3,734 | 11,645 | 8,945 | 0 | 1,840 | 435 |
| Operating income before those items [Calculation] | 4,673 | 3,406 | 308 | 5,005 | 9,431 | 9,031 |
| Operating margin before those items [Calculation] | 7.2% | 5.3% | 0.5% | 7.8% | 13.1% | 11.9% |
| Gross margin [Calculation] | 50.5% | 45.6% | 38.5% | 39.5% | 42.2% | 40.2% |
| Net earnings (loss) [Filing] | (2,861) | (4,965) | (8,101) | 3,021 | 4,178 | 5,429 |
1991 restructuring is derived from the FY1993 MD&A ($4,673M operating income before restructuring, less $939M after). 1995 income-statement figures follow the FY1996 10-K’s reclassified presentation. The totals match the FY1995 10-K.
Adjustment 1: the “one-time” items that kept happening
[Filing] SG&A included a $382M gain on the Federal Systems sale in 1994. In 1995 it included a $626M charge for “work force separations and asset reductions,” offset by a $175M one-time gain. In 1996 it included $669M “associated with infrastructure reductions.” Taking those out [Calculation]: 1994 margin goes to 7.2%, 1995 to 13.7%, and 1996 to 12.8%. The direction doesn’t change. But a separation charge in three straight years isn’t really a one-time item. It’s part of what it cost to run the place.
Adjustment 2: pension cost turned into pension income
| Defined-benefit pension cost, $M [Filing] | 1992 | 1993 | 1994 | 1995 | 1996 |
|---|---|---|---|---|---|
| U.S. plan, net periodic pension cost | 261 | 727 | 11 | (252) | (285) |
| Worldwide defined-benefit plans | not reported on a comparable basis | 678 | 165 | (137) | |
Sources: FY1993, FY1995, and FY1996 10-K retirement-plan notes. Negative cost means pension income.
[Calculation] From 1994 to 1996, worldwide pension expense improved by $815M, about 9.5% of 1996 pretax earnings of $8,587M. Holding pension cost at the 1994 level, the 1995 margin before items would be about 12.4% and 1996 about 10.8%, instead of 13.1% and 11.9%. That’s a real tailwind. It doesn’t reverse the recovery. The U.S. plan’s expected return on assets was 9.25% to 9.5% in those years [Filing], so part of this “improvement” came from actuarial assumptions and asset returns, not from operations.
Adjustment 3: currency and perimeter
[Company assertion/Filing] IBM said currency moved revenue by about −3% in 1993, +2% in 1994, and +4% in 1995. In 1996, revenue grew 5.6% as reported and 8.6% excluding currency. The 1993 base includes $2,063M of Federal Systems revenue that was sold in 1994. Lotus (from mid-1995) and Tivoli (1996) added revenue that the filings I read don’t break out separately.
[Calculation, approximate] Reported revenue rose 21.1% from 1993 to 1996, or 25.2% after removing Federal Systems from the base. Backing out the approximate cumulative currency effect leaves roughly 22% before subtracting acquired revenue, which can’t be quantified from these filings. That’s about 6 to 7% a year at best: real growth, but not the story the headline suggests.
Cash: Recurring or Borrowed?
| $ millions | 1991 | 1992 | 1993 | 1994 | 1995 | 1996 (check) |
|---|---|---|---|---|---|---|
| Cash from operations [Filing] | 6,725 | 6,274 | 8,327 | 11,793 | 10,708 | 10,275 |
| − Plant, rental machines, other property [Filing] | 6,497 | 4,751 | 3,154 | 3,078 | 4,744 | 5,883 |
| − Investment in software [Filing] | 2,014 | 1,752 | 1,507 | 1,361 | 823 | 295 |
| = Post-investment cash [Calculation] | (1,786) | (229) | 3,666 | 7,354 | 5,141 | 4,097 |
| Of which: receivables + inventory release [Filing] | (922) | 1,756 | 1,768 | 2,171 | (423) | (454) |
| Post-investment cash excluding working-capital release [Calculation] | (864) | (1,985) | 1,898 | 5,183 | 5,564 | 4,551 |
| Restructuring cash paid, inside operating cash [Filing] | n/a | n/a | n/a | 2,772 | 2,119 | 1,491 |
| Total debt, year-end [Filing] | 26,947 | 29,320 | 27,342 | 22,118 | 21,629 | 22,829 |
The Federal Systems sale proceeds ($1,503M, 1994) and marketable-securities flows are investing items and are excluded above. Restructuring cash paid is the “effect of restructuring charges” line in the cash-flow statements. The FY1993 10-K notes that the 1993 drop in receivables was “primarily attributable to the company’s securitization of assets.”
Read it in three steps. In 1993, liquidity was bought: $11.8B of new debt, $1.1B of preferred stock, a dividend cut, and a securitization-assisted fall in receivables. In 1994 and 1995, recurring cash after investment ran above $5B a year, even after paying out more than $2B a year in cash against old restructuring reserves, and total debt fell by about $5.7B from the end of 1993 to the end of 1995 [Calculation]. In 1995 and 1996, the cash went out again: $10.7B of common buybacks, $3.6B of net acquisition cash, and $1.3B of dividends, against about $9.2B of post-investment cash. Debt rose again in 1996. Whether Lotus earned back its price can’t be judged inside this window.
Peers: Same Era, Same Business, Failures Included
Peer set: Hewlett-Packard, Digital Equipment, and Unisys. All three were U.S.-listed computer-systems vendors filing on EDGAR in the period, and two of them were losing money or shrinking during it. Non-U.S. mainframe rivals were left out because their filings aren’t on EDGAR, which is a gap. One more disclosure: I knew roughly how these companies fared before picking them. That’s a selection risk I can’t fully remove, so the sensitivity test below reruns the comparison with different subsets. Fiscal years differ: HP’s ends in October and DEC’s in late June or early July. Operating margins below exclude separately disclosed restructuring and special charges [Calculation on Filing data].
| Operating margin before charges | FY1992 | FY1993 | FY1994 | FY1995 | FY1996 | Revenue change FY93 to FY96 |
|---|---|---|---|---|---|---|
| IBM | 5.3% | 0.5% | 7.8% | 13.1% | 11.9% | +21.1% |
| Hewlett-Packard | 8.6% | 9.2% | 10.2% | 11.3% | 9.7% | +89.1% |
| Digital Equipment | −4.6% | −1.6% | −3.6% | 0.8% | 3.1% | +1.3% |
| Unisys | 10.2% | 11.4% | 7.5% | 4.5% | 5.1% | +4.3% |
| Peer median | 8.6% | 9.2% | 7.5% | 4.5% | 5.1% | +4.3% |
DEC adds back restructuring of $1,500M (FY92), $1,206M plus a $310M intangible write-down (FY94), and $492M (FY96). Unisys adds back special charges of $185.2M (1994, the operating portion of $186.2M) and $846.6M (1995). Unisys says its 1996 results include reserve reversals offset by new charges, so its 1996 figure is as reported. HP reported no separate restructuring line in this table.
[Calculation] IBM’s margin gain beat the peer median in FY1994 (+7.3 points versus the median peer’s −2.0) and FY1995 (+5.3 versus +1.1). That’s the two annual periods the rubric asks for. By FY1995, IBM’s margin level also passed every peer. On growth, though, IBM’s +21% sat far behind HP’s +89%. The demand environment was strong enough for one same-era hardware company to nearly double, so “the market was terrible” can’t carry IBM’s growth record. Neither can “the market lifted everyone”: DEC and Unisys barely grew.
The Scorecard
| Dimension | Weight | Score | Confidence | The observations behind the score |
|---|---|---|---|---|
| Operating economics / customer value | 25% | 3 | Medium | Margin before charges beat the peer median in FY94 and FY95. It held after the pension and one-time adjustments. Customer evidence: audited services revenue (excluding Federal Systems) more than doubled from $7.6B (1993) to $15.9B (1996), and services gross margin rose from 16.5% (1993 excluding Federal Systems, as IBM reported it) to about 20 to 21%. Held below 4 because services growth was already 33 to 38% a year before 1993, gross margin never recovered (40.2% in 1996 against 50.5% in 1991), and there’s no retention or satisfaction data in the filings, only revenue. |
| Cash resilience / capital allocation | 25% | 3 | Medium | Recurring post-investment cash (excluding working-capital release) went from negative in 1991 and 1992 to $5.2B and $5.6B in 1994 and 1995. Debt fell about $5.7B from the end of 1993 to the end of 1995. Held below 4 because 1993 liquidity came from financing and a dividend cut, 1995 and 1996 buybacks plus acquisitions exceeded post-investment cash, and the return on the $3.2B Lotus deal can’t be assessed in the window. |
| Strategy / execution vs. commitments | 20% | 3 | Medium | Kept: integrated-provider structure (services scaled), profit from Q4 1993 on (except the Q3 1995 Lotus write-off quarter), no restructuring line after 1993. Not cleanly kept: workforce and infrastructure charges of $626M (1995) and $669M (1996) were absorbed in SG&A. Not verifiable: the “$7.0 billion in annual savings” by 1996. The later filings I read don’t report against it. SG&A plus R&D fell only $2.3B from 1993 to 1996, and the rest of the target can’t be separated from mix shifts in cost of sales. |
| Governance / reporting / stakeholder costs | 15% | 2 | Medium | Reporting was transparent: charges, gains, purchased R&D, currency, and pension each disclosed separately, and the 1997 report repeats 1993 to 1996 totals without restatement. The costs were heavy and only partly mitigated in the filings: wholly owned headcount fell 81,703 from 1992 to 1994, R&D was cut 36% from 1992 to 1995 (excluding purchased R&D), and the dividend was cut by more than half. The filings disclose these costs but not outcomes for the affected employees. |
| Institutional durability / succession | 15% | N/A | — | Outside this window. Not scored, and not treated as a 2. |
Coverage: 85% (original weights 25 + 25 + 20 + 15). Calculation: (0.25×3/4 + 0.25×3/4 + 0.20×3/4 + 0.15×2/4) ÷ 0.85 = 0.600 ÷ 0.85 = 70.6, a provisional B. No core dimension is N/A or Low confidence, so the rules allow a provisional grade, and only for 1993 to 1995. It isn’t a career grade.
Disconfirming evidence: the case against crediting the CEO
- About $15.3B of restructuring charges and a 42,854-person headcount reduction came before April 1993.
- Services, the business most associated with the turnaround, grew 33% (1992) and 38% (1993) before the new strategy could have reached the numbers. After that it grew 27%, 31%, and 25%: sustained, not accelerated.
- Gross margin in 1996 (40.2%) was still nearly 10 points below 1991. The margin recovery came from expenses, not price.
- Pension expense improved by $815M from 1994 to 1996, about 9.5% of 1996 pretax earnings.
- Workforce and infrastructure charges showed up again in 1995 and 1996.
- R&D fell from $6.5B (1992) to $4.2B (1995, excluding purchased R&D). Mills asked at the time, “How much of the company’s muscle was cut with the fat?” [Analyst interpretation]
- HP grew revenue 89% over the same stretch, so the market had room that IBM didn’t capture.
The Credible Alternative Explanation
[Analyst interpretation] Writing in 1996, Mills argued that Gerstner “has not taken IBM on a new course as much as he is returning it to its roots.” In his reading, the one-stop, full-service model was IBM’s old strategy, and the company had drifted from it in the 1980s. He also described the sequence plainly: “Initially, Gerstner’s top management team cut costs and downsized. Profitability returned, but not growth.”
The filings support a version of that. The services trend line was inherited. The biggest cost reset was already underway. Pension income and currency helped the later years. If that were the whole story, though, the peers should have recovered too. Unisys’s margin fell from 11.4% to 4.5% over the same years. DEC reached only 3.1% by FY1996. IBM got to 13.1%. [Inference] A fair split: the margin and cash turnaround depended on decisions made in 1993 and 1994, including the second large charge, the dividend cut, the financing bridge, the Federal Systems sale, and keeping the company integrated. The growth that followed owes a lot to strengths IBM already had. These controls can’t separate the CEO’s personal contribution from his team’s, his board’s, or the predecessor’s groundwork. So this grades IBM management’s decisions during his tenure, not the man.
Sensitivity Test: Does the Grade Survive?
| Change the assumption | Effect on scores | Overall |
|---|---|---|
| Peer set = DEC + Unisys only (drop HP) | Operating economics might argue for 4 on margins. Customer evidence still caps it. | Unchanged at 70.6, or 77.9 (B) if raised |
| Peer set = HP only | Margin gain still beats HP in FY94 and FY95. Growth badly lags. Score could fall to 2. | 63.2 (C) |
| Window = FY1992 baseline instead of the 1993 trough | IBM +6.6 margin points to FY1996. DEC +7.7 from its own trough, Unisys −5.1, HP +1.1. IBM still beats the median, but less dramatically. | Unchanged |
| Hold pension at 1994 cost | 1995 margin 12.4%, 1996 10.8%. Still well above peers. | Unchanged |
| Strip all one-time items both ways | Margins move slightly up. The recurring nature of the charges argues against raising execution. | Unchanged |
| Credit capital allocation (score cash 4) | Only if you accept the 1995 and 1996 buybacks and the Lotus price as good uses of cash, which this window can’t show | 77.9 (B) |
| Weight stakeholder costs harder (governance 1) | R&D and headcount cuts read as under-mitigated | 66.2 (C) |
Verdict on the verdict: the direction holds up under every test. IBM’s 1993 to 1995 margin and cash repair beat its peers, and its growth story is mostly inherited. The letter doesn’t hold up. Any single one-point change moves it between B and C. Treat “provisional B” as a range from high C to low B. It isn’t a medal.
What a Board Should Take From This
- Split the charges by who decided them. Put the predecessor’s charges and headcount cuts on a separate line before crediting anyone with the margin rebound.
- Check whether the “new” growth engine was already growing. If the line was compounding at 30%+ before the appointment, the new leader gets credit for scaling it and improving its margin, not for discovering it.
- Normalize pension and currency before celebrating margin. Here, those two items alone were worth more than a point of margin.
- Count recurring “one-time” charges as operating cost. Three straight years of separation charges is a cost of running the business.
- Judge cash after distributions, not before. Buybacks and acquisitions can quietly take back the headroom a turnaround created.
If you’re building the dashboard to run this test inside your own company, start with Measuring Transformation Progress: KPIs, Milestones, and Success Metrics, and add the reconciliation lines above that most dashboards skip. If your board is stuck waiting for perfect attribution before acting, The 70% Rule: Decision Velocity Matrix covers how much certainty a reversible decision actually needs. And if the question is whether your operators can make the shift from cutting to growing, see Why Do Excellent Operational Leaders Fail at Transformation? For a different lens on the same CEO, Todd’s full-career, survival-after-succession view is in The Greatest Turnaround CEOs, Judged Honestly. This piece covers only the first three years and the audited numbers.
Limits
Filings tell you what happened to the numbers, not who thought of what. Mills’s article is interview-based and only partly public. The peer set is small and U.S.-only. Acquired revenue isn’t broken out, so organic growth here is an upper bound. No counterfactual is claimed: nobody knows what a broken-up IBM would have earned. None of this is investment advice.
Run the Test on Your Own Business
- Book Todd to speak: boards and leadership teams can bring this evidence-first approach to turnarounds into the room. toddhagopian.com/speaking
- Read the books: Stagnation Assassin, plus the forthcoming 10-Minute Transformation (Koehler Books, February 23, 2027). toddhagopian.com/book/
- Free tool: find out whether your business is an ATM or a money pit with the ATM Test Scorecard.
- Want to read 10-Minute Transformation before launch? Join the launch team.
Sources
- IBM, Form 10-K for fiscal 1993 (filed March 28, 1994), including Exhibit 13. SEC EDGAR
- IBM, Form 10-K for fiscal 1994 (filed March 28, 1995). SEC EDGAR
- IBM, Form 10-K405 for fiscal 1995 (filed March 26, 1996). SEC EDGAR
- IBM, Form 10-K for fiscal 1996 (filed March 27, 1997). SEC EDGAR
- IBM, 1997 Annual Report, five-year comparison (restatement check). IBM
- Hewlett-Packard, Form 10-K for fiscal 1996 (selected financial data). SEC EDGAR
- Digital Equipment Corporation, Form 10-K for fiscal 1996 (eleven-year financial summary). SEC EDGAR
- Unisys Corporation, Form 10-K405 for fiscal 1996 (six-year summary; Note on special charges). SEC EDGAR
- IBM, “Louis V. Gerstner Jr.,” IBM History (tenure dates only). IBM
- D. Quinn Mills, “The Decline and Rise of IBM,” MIT Sloan Management Review, July 15, 1996. MIT SMR
Calculations are mine, from the filed figures cited. Mills’s 1996 figure of 1995 sales “almost $72 billion, up more than 12 percent” matches the filed $71,940M, up 12.3% from $64,052M. Educational analysis, not financial advice.
About the Author
Todd Hagopian is VP of Product Strategy and Marketing and a business transformation author, speaker, and executive. His research is available as SSRN preprints. He is the author of the Turnaround Code Trilogy: The Unfair Advantage, Stagnation Assassin, and 10-Minute Transformation (February 23, 2027). More about Stagnation Assassins: About.

