The Revlon ERP failure cost the company roughly $64 million in orders its biggest plant could not ship, $53.6 million in cleanup charges, and a material weakness in its financial controls, all from one SAP go-live in Oxford, North Carolina, in February 2018. That much is on every page that ranks for the story. The usual moral is “don’t do a big bang.”
The filings tell a sharper story. So does a federal judge’s 25-page opinion that almost nobody quotes. The plant reportedly failed testing badly enough to delay the launch four times. The chief executive left the month before go-live. And when Revlon finally explained what broke, it didn’t blame SAP at all. It blamed its own risk assessment, its own staffing, and controls nobody owned.
The software is the least interesting thing in this case. Here is the rest of it, from the primary record.
What Revlon Was Trying to Merge Before the Revlon ERP Failure
Revlon didn’t have one legacy system to replace. It had a pile of them. According to the shareholder complaint summarized in the court’s opinion, before 2014 Revlon ran 21 separate enterprise resource planning (ERP) systems across 50 different entities. ERP software is the system of record for a company’s finance, inventory, manufacturing, and orders; 21 of them means 21 versions of the truth.
Then the company kept buying things. In 2013 it acquired the Colomer Group, a Barcelona professional-hair-care business that already ran SAP. Revlon’s own heritage operations leaned on Microsoft Dynamics AX. In 2014, Revlon decided to put everything on one SAP system, and its plan, per the complaint, was to take the template the newly acquired subsidiary had used for its own SAP rollout and apply it to the merged company.
In June 2016 Revlon agreed to buy Elizabeth Arden for an enterprise value of about $870 million, promising $140 million in projected cost savings. Arden ran Oracle Fusion Applications. So by the time the big SAP project was in flight, the systems estate held Dynamics, Oracle, a Spanish SAP template, and the remnants of 21 ERPs, all inside a company that had carried heavy debt since Ron Perelman’s MacAndrews & Forbes bought it in a debt-financed takeover in 1985. MacAndrews still owned about 87% of Revlon’s Class A shares at the end of 2019.
Integration was not a side project. Revlon reorganized into four brand segments (Revlon, Elizabeth Arden, Portfolio, and Fragrances) and booked $82.2 million in Elizabeth Arden integration restructuring charges. The ERP had to land in the middle of all of that. If this sounds like a pattern, it is one; we mapped it across two dozen cases in our pattern library of why ERP implementations fail.
| Date | Event | Source |
|---|---|---|
| Before 2014 | 21 separate ERP systems across 50 entities | Shareholder complaint, per court opinion |
| 2013 | Revlon acquires Colomer, already on SAP | Press coverage; court opinion |
| 2014 | Decision to consolidate onto one SAP system, using the acquired subsidiary’s template | Court opinion |
| June to September 2016 | Elizabeth Arden deal announced (about $870M enterprise value) and closed; Arden runs Oracle Fusion | Press coverage; FY2018 10-K |
| March 2017 | 2016 10-K warns the SAP rollout “may disrupt the Company’s operations” | Court opinion, quoting the 10-K |
| January 2018 | CEO Fabian Garcia departs; interim leadership | Court opinion |
| February 1, 2018 | SAP goes live at the Oxford, N.C. plant | Court opinion; FY2018 10-K |
| March 15, 2018 | 2017 10-K discloses the disruption | FY2017 10-K |
| March 18, 2019 | NT 10-K: annual report late, material weakness | SEC filing |
| March 28, 2019 | FY2018 10-K: $64M lost sales, $53.6M charges, adverse auditor opinion on controls | SEC filing |
| May 2019 | Shareholder class actions filed | Court docket, press |
| March 2020 | FY2019 10-K: material weakness remediated | SEC filing |
| September 17, 2020 | Federal court dismisses the lead securities suit | Lachman v. Revlon (E.D.N.Y.) |
| June 2022 | Chapter 11 filing, Southern District of New York | SEC 8-K |
| May 2023 | Revlon deregisters its stock after reorganizing | SEC Form 15 |
The Oxford Go-Live: How One Plant Became a Single Point of Failure
Revlon’s 10-K has a risk factor whose title tells you most of what you need: “The Company depends on its Oxford, North Carolina facility for production of a substantial portion of its products.” The plant covered about 1,012,000 square feet of manufacturing, warehousing, distribution, and office space, and it served all four business segments. It was also registered with the FDA as a drug manufacturing establishment, because sunscreen and antiperspirant count as over-the-counter drugs.
That is where SAP went live first. Not a pilot site. The plant.
The shareholder complaint, as Judge Rachel Kovner summarized it in Lachman v. Revlon, alleged a sequence page one never mentions. These are allegations the court accepted as true only for the purpose of ruling on a motion, and the suit itself was dismissed (more on that below), so read them as a plaintiff’s account, not findings:
- Revlon built a team of about 100 people, including IT consultants and subject matter experts, to run the transition. The team “did not have experience using SAP in the retail context,” nor at a manufacturing plant as large and complex as Oxford.
- Testing at Oxford surfaced problems that led Revlon to delay the company-wide rollout four times.
- Officials considered pushing the date back again. They went live instead, on February 1, 2018.
- Warehouse management and production systems were not properly integrated, and employees had not been adequately trained.
- Orders went unfilled, items were lost in the production process, and the plant’s operations were set back by two months.
- Executives convened “crisis management meetings,” some employees worked seven days a week, and the company brought back a former head of supply chain to help dig out.
Our corpus has an independent line on the same point: an ERP-support vendor’s post-mortem lists among the contributing factors that “Revlon didn’t have experience with SAP.” Four delays is the detail that matters. A testing program that stops the launch four times is working. The fifth decision is the one that failed.
A plant that fails testing four times is sending a message. The question is who in the room was paid to listen to it.
Who Was in the Room
The court opinion lists the defendants’ tenures, and laid end to end they read like a relay race with nobody holding the baton at the handoff:
- CEO Fabian Garcia served from April 2016 to January 2018. He left the month before go-live.
- Paul Meister was interim principal executive officer from January to May 2018, the launch window.
- Debra Perelman was chief operating officer from January to November 2018 and became CEO in May 2018.
- Christopher Peterson was COO from April 2017 to July 2018 and CFO from June 2017 to March 2018.
- Victoria Dolan became CFO in March 2018, one month after the go-live she would later have to explain to the SEC.
- The chief accounting officer seat changed hands in 2017 and again at the end of 2018.
None of that proves causation. It does explain how a fifth delay could go uncalled: the people who had lived through the first four were mostly leaving or moving seats as the date arrived.
“Back to Normal” by April
Revlon’s first-quarter 2018 earnings release, filed May 10, 2018, put a $20 million figure on lost net sales from the Oxford disruptions and then said this: “as of early April, the facility is back to normal production capacity and the Company is re-filling its pipeline orders.”
The second quarter lost about another $30 million in net sales to the same disruption, mostly in international markets where the pipeline still hadn’t refilled. Production capacity had recovered. The business hadn’t.
That gap is the most transferable fact in the case. A plant can be “back” on the operations dashboard while customers on the other side of the world are still waiting for product that was never made. Hershey learned the same lesson in 1999 with different candy and a different calendar; our Hershey ERP failure autopsy shows how a timing decision, not the software, did the damage there too.
What the Revlon ERP Failure Cost, Quarter by Quarter
Every ranking page quotes $64 million. None shows how it built up, and the build-up is where the lesson lives. Here is the ledger from Revlon’s own quarterly earnings releases and its 10-K.
| Period | Lost net sales (Oxford) | Disruption charges | What Revlon said at the time |
|---|---|---|---|
| Q1 2018 | ~$20M | $10.0M | Plant “back to normal production capacity” as of early April |
| Q2 2018 | ~$30M | $23.1M | Declines “predominantly in the international market” |
| Q3 2018 | Not quantified in the release | $16.5M | Revlon-brand sales up in North America as retailers restocked; international still down |
| Q4 2018 | Not quantified in the release | ~$4.0M (derived: full year less Q1 to Q3) | $1.4M of Oxford costs netted against Q4 sales |
| Full year 2018 | ~$64M | $53.6M | Plant “operating at pre-SAP levels” by December 31 |
Two more numbers from the 10-K sharpen the picture. The Revlon brand segment absorbed about $54 million of the $64 million in lost sales. And the extra costs from the Oxford disruptions cut the company’s gross margin by 1.9 percentage points for the year.
Now the fact-hygiene part. Several write-ups present the $70.3 million fourth-quarter 2018 net loss as the price of the ERP failure. Read the release. That Q4 loss was an 8.6% improvement on the $76.9 million loss a year earlier, and Oxford touched Q4 sales for $1.4 million. The full-year 2018 net loss of $294.2 million (versus $183.2 million in 2017) had plenty of authors: $176.6 million of interest expense, currency losses, an $18.0 million goodwill impairment, and lower sales. The ERP was one line in a bad year at a heavily indebted company. It was not the whole year.
You’ll also find a “$54 million total migration cost” and a “$32 million in 2018 migration spending” circulating. I couldn’t find either figure in Revlon’s FY2017 or FY2018 10-K or in any of its four 2018 earnings releases. The $54 million looks like a rounding of the $53.6 million remediation charge, which is a different thing entirely. The $10 million of “unplanned expenses” that also gets quoted does reconcile: it matches the first-quarter charge.
The Paper Trail: What Revlon Actually Told the SEC
Every page on the first page of Google paraphrases press coverage of these filings. The filings themselves are short, public, and more candid than the coverage.
| Filing | Date | What it said |
|---|---|---|
| 2016 Form 10-K | March 3, 2017 | The SAP rollout “may not result in improvements that outweigh its costs and may disrupt the Company’s operations” |
| 2017 Form 10-K | March 15, 2018 | The new ERP system had “disrupted the Company’s business operations”; no assurance the issues would be fixed in time to recover lost sales |
| Q1 2018 earnings release | May 10, 2018 | ~$20M of lost net sales; plant “back to normal production capacity” as of early April |
| Form 12b-25 (NT 10-K) | March 18, 2019 | Annual report late; material weakness “primarily related to the lack of design and maintenance of effective controls” tied to the U.S. ERP implementation |
| 2018 Form 10-K | March 28, 2019 | $64M lost net sales, $53.6M charges, three control deficiencies, auditor’s adverse opinion, nine-step remediation plan |
| 2019 Form 10-K | March 12, 2020 | Material weakness remediated; controls effective as of December 31, 2019 |
The NT 10-K, the form a company files when it can’t get its annual report in on time, is two paragraphs of substance, signed by CFO Victoria Dolan. “The principal reason for the delay is the recent identification of a material weakness.” A material weakness, in plain English, is a gap in financial controls serious enough that a material error in the books might not be caught in time.
The 2018 Form 10-K, Item 9A, then names three deficiencies. I’m quoting the key phrases because they are the actual root-cause analysis, written by the company, and they read nothing like “SAP broke”:
- Revlon “did not perform an effective continuous risk assessment process” that identified the risks the new ERP system posed to its financial reporting controls.
- It “did not maintain a sufficient number of knowledgeable, trained personnel” in the U.S. operations hit by the rollout, people who “understood and were held accountable for” their control responsibilities.
- As a result, it did not design and consistently operate process-level controls over four things: recording inventory, receivables, net sales, and cost of goods sold; reconciling balance sheet accounts; reviewing and approving the complete population of manual journal entries; and using complete, accurate information in manual controls.
Read deficiency three again. Inventory, receivables, sales, and cost of goods sold are the core of a manufacturer’s books. Manual journal entries are what people type when the system can’t do it for them. A company leaning on manual entries during a broken go-live, without anyone reviewing all of them, is a company whose auditor will notice.
The auditor did. KPMG, which audited the 2018 statements, issued an adverse opinion on Revlon’s internal control over financial reporting. The financial statements themselves were not restated; the deficiencies “did not result in any material misstatement.” The books were right. Nobody could prove they would stay right.
The nine-step remediation plan is worth a skim for one line in particular: Revlon committed to “enhance the Company’s review and sign-off procedures for IT implementations.” That is a company writing, in a federal filing, that its go/no-go process for system launches needed fixing. It also committed to “supplement internal resources with third-party consultants.” The cure for an implementation that went live without enough trained people was, in part, more outside people.
The Stock Drop and the Shareholder Lawsuit That Lost
The most quoted number in this story is a 6.9% stock drop “within 24 hours.” A widely shared Medium piece says 6.4% instead, and at least one other write-up agrees with it. Both numbers are real, and the court record explains why. Revlon’s share price fell approximately 6.87% when the company announced the late filing and material weakness on March 18, 2019. It fell an additional 6.4% on March 28, the day the 10-K landed with the details. Each page on Google is quoting one leg of a two-step fall.
Shareholder class actions followed in May 2019. Several plaintiffs’ lawyers filed; the case that went forward in the Eastern District of New York was Lachman v. Revlon, No. 19-cv-2859. It claimed Revlon and eight current and former executives had downplayed the SAP risks before launch and the damage afterward.
It lost. On September 17, 2020, Judge Kovner granted the motion to dismiss, finding the plaintiffs had not adequately pleaded any material misstatement or a strong inference of intent to deceive. The dismissal was without prejudice, with leave to file an amended complaint, and Revlon’s 2020 10-K lists no such suit among its legal proceedings. Nobody on page one reports the ending.
The reason it lost is the uncomfortable part. Revlon had warned investors in its 2016 10-K that the SAP rollout could disrupt operations, and it disclosed the disruption in its 2017 10-K a few weeks after go-live. Disclosure didn’t save the plant. It saved the company in court.
The Aftermath: Remediation in 2019, Chapter 11 in 2022
The operational recovery took most of 2018. By December 31, 2018, the 10-K says, Oxford was “operating at pre-SAP levels,” while the company was still refilling inventory at retail partners, particularly overseas.
The controls recovery took 2019. Revlon’s 2019 10-K reports that it “completed a series of actions and measures that effectively remediated the previously-disclosed material weakness,” and that its internal control over financial reporting was effective as of December 31, 2019. Roughly two years from go-live to a clean bill on controls.
Then, on June 15 and 16, 2022, Revlon and its subsidiaries filed for Chapter 11 in the Southern District of New York (Case No. 22-10760), with $575 million in debtor-in-possession financing lined up. Readers ask whether SAP bankrupted Revlon. By Revlon’s own account, no: its press release blamed “liquidity constraints brought on by continued global challenges, including supply chain disruption and rising inflation, as well as obligations to its lenders,” and CEO Debra Perelman pointed to “our challenging capital structure.” The company reorganized and deregistered its public stock in May 2023.
The honest read sits between the two camps. The ERP failure did not cause the bankruptcy four years later. It did take $64 million of sales and $53.6 million of cash out of a company that had very little room for either, in the same year it was paying $176.6 million in interest.
The story has also hardened into curriculum. In 2020 the IBS Center for Management Research published a 14-page teaching case, “ERP Implementation Failure at Revlon”, distributed through The Case Centre for postgraduate courses. Its closing question is the one this post keeps returning to: what Revlon could have done to understand and mitigate the implementation risks before go-live, not after.
Bottom line: Revlon’s own root-cause analysis blames risk assessment, staffing, and control ownership, not the software. Those are buyer-side decisions, which means they are also buyer-side fixes.
What the Revlon ERP Failure Teaches a Buyer in 2026
Strip away the cosmetics and this is the case that every S/4HANA consolidation after an acquisition is quietly repeating: several inherited ERPs, one template borrowed from whichever acquired unit already ran SAP, a single plant that matters more than the rest, and leadership that changes while the project doesn’t. The page-one lessons (“manage risk,” “do change management”) are true and useless. Revlon’s 10-K gives you a better checklist, because it is the company’s own list of what it didn’t do.
Assess Risk Continuously, Not Once
Deficiency one was a “continuous” risk assessment that wasn’t. A risk register written at blueprint and reviewed at steering committee is not continuous. Each test cycle that fails should change the risk picture, and somebody outside the delivery team should own that call. For the go-live gate itself, one corpus implementation guide we track uses concrete thresholds, such as 90% or better test-case completion and a go-live readiness assessment treated as a formal, signed milestone. Pick your own numbers. Write them down before the date gets political.
Rehearse the Cutover in Something That Looks Like Production
SAP’s own methodology says it plainly. As Denecken, Musil and Santhanam write in SAP Activate (SAP PRESS, 2025): “Before performing the cutover to production, it’s important to rehearse the cutover plan entirely in a test system that reflects the future state of the production system.” The same chapter adds: “There may be a need to postpone the go-live date if significant changes and critical items are raised as a result of this rehearsal.”
That is a publisher’s methodology book, not a critique of anyone. It is also the discipline the complaint describes breaking down: testing at Oxford kept surfacing problems, and the launch went ahead anyway. The method expects postponement. The question is whether your governance does.
Don’t Make Your Biggest Plant the Pilot
Oxford produced a substantial portion of everything Revlon sold, for all four segments, and it went first. If you have one site that can stop the company, it should go live after the template has survived somewhere smaller, or with a fallback that lets you ship from the old system for a defined window. LeasePlan’s failed program made the opposite bet on a larger scale, consolidating 35 systems in a big-bang approach, and wrote off €92 million. Revlon was consolidating 21.
Staff the Controls, Not Just the Configuration
Deficiency two is about people who understood and were “held accountable for” financial controls in the operations the system touched. Implementation partners staff configuration, data migration, and testing. Your controllership has to staff the controls, and the partner’s plan should show where those people sit in hypercare. If the plan has no line for it, ask why.
Decide Now Who Calls the Delay
Here is the part an implementation partner can’t print: Revlon’s integrator, whoever it was, appears in none of Revlon’s filings, and that is normal. The company whose name is on the 10-K carries the risk. A delivery team paid by the month and told four times that the plant isn’t ready has no commercial reason to call a fifth delay, and an interim executive who arrived in January has no history to call it with. Name the person who can stop the go-live, give them the test results directly, and do it before the first slip, not the fourth.
The other honest point: a partner who has run a go-live at a plant your size, in your industry, is worth more than one who has run many go-lives somewhere else. Revlon’s team, as alleged, had SAP experience but not retail or large-plant experience. That’s a matching problem, and it is the one we built the engine for. Describe your consolidation and your plant, and watch it rank SAP partners on the evidence, free, no signup. If you’d rather run the selection yourself, our guide on how to choose an SAP implementation partner covers the reference checks that would have surfaced Revlon’s experience gap.
Frequently Asked Questions
What happened with Revlon?
In February 2018 Revlon switched its North American operations onto SAP S/4HANA, starting at its Oxford, North Carolina plant, and the plant could not make or ship enough product. That cost about $64 million in net sales and $53.6 million in remediation charges in 2018, and in March 2019 it forced a late 10-K and a material weakness disclosure. Revlon’s own filings blamed no continuous risk assessment, too few trained people owning financial controls, and weak controls over inventory, sales and journal entries. The company fixed the controls by the end of 2019, then filed for Chapter 11 in June 2022 over debt, not SAP.
Is Revlon still a public company?
No. Revlon filed for Chapter 11 in June 2022, reorganized, and deregistered its public stock with a Form 15 in May 2023. It still sells its brands, but it no longer files 10-Ks, which is why the SEC paper trail on the SAP rollout stops with the filings cited in this post.
What are some examples of ERP failures?
Revlon’s 2018 SAP go-live is one of the best documented, because it produced SEC filings and a federal court opinion. Other well-known cases: Hershey’s 1999 rollout, which left about $100 million in candy undelivered before Halloween; Nike’s i2 demand-planning rollout, which cost about $100 million in lost sales; MillerCoors, which sued its integrator for $100 million in 2017; and Lidl, which wrote off about €500 million. Our hub on why ERP implementations fail tabulates them side by side.
What percentage of ERP projects fail?
It depends on how you define failure, which is why the widely repeated 55% to 75% figures rarely cite a primary source. Gartner’s current research predicts that by 2027, more than 70% of recently implemented ERP initiatives will fail to fully meet their original business goals. Our hub walks through the verified numbers and where the recycled ones came from.
What is replacing ERP?
Nothing replaces ERP itself; companies replace one ERP with another, or move the same one to the cloud. Revlon’s project is the pattern: SAP S/4HANA was brought in to replace Microsoft Dynamics AX, Elizabeth Arden’s Oracle applications, and a sprawl of smaller systems. For SAP customers the live version of that question is ECC, whose mainstream maintenance SAP set to end in 2027, with S/4HANA as the successor. That is a replacement on a deadline, which is exactly the condition under which Revlon went live at its most important plant.
What ERP does Coca-Cola use?
Coca-Cola is an SAP customer. When SAP and Microsoft announced S/4HANA on Azure in November 2017, Coca-Cola was named as a customer of both, and its CIO Barry Simpson called the deal “an extremely important development for the Coca-Cola System.” Like Revlon, it is a consumer goods business running many entities on one SAP estate, which is why the Revlon lessons on staging and control ownership apply well beyond cosmetics.
Sources
- Revlon, Inc., Form 10-K for fiscal 2018, filed March 28, 2019: Oxford disruption figures, Item 9A deficiencies, auditor opinion, remediation plan, facility details.
- Revlon, Inc., Form 12b-25 (NT 10-K), filed March 18, 2019: late filing and material weakness disclosure.
- Revlon, Inc., fourth-quarter and full-year 2018 earnings release, March 18, 2019: Q4 net loss and Q4 Oxford costs.
- Revlon, Inc., 2018 earnings releases for Q1, Q2, and Q3: quarterly lost sales and disruption charges.
- Revlon, Inc., Form 10-K for fiscal 2019: remediation of the material weakness.
- Lachman v. Revlon, Inc., No. 19-cv-2859 (E.D.N.Y. Sept. 17, 2020): complaint allegations, executive tenures, stock-price moves, dismissal.
- Revlon, Inc., Chapter 11 press release, June 16, 2022.
- Sven Denecken, Jan Musil and Srivatsan Santhanam, SAP Activate, 3rd ed. (SAP PRESS, 2025): quoted on cutover dress rehearsals.
- Syeda Maseeha Qumer and Debapratim Purkayastha, “ERP Implementation Failure at Revlon”, IBS Center for Management Research teaching case (2020), via The Case Centre.
- Cosmetics Business: Elizabeth Arden deal value and projected savings.
Revlon is usually told as a cautionary tale about SAP. Its own filings tell it as a cautionary tale about who was accountable when the test results came back. The software will go live on whatever date you choose. Whether your business goes live with it depends on who has the authority to say “not yet,” and whether you gave it to them before you needed it. When you’re ready to pick the team, browse SAP partners by module, industry, and plant-floor experience.
