Related

Electric Engine Autonomous Manufacturing Plant With Robotic Arms: Diverse Specialists Talking About Modern Machinery Production. Industrial Engineer, Designer, And Assembler Discussing Project.

The ROI of Digital Product Passports: benefits beyond compliance

The ROI of Digital Product Passports: benefits beyond compliance
Sen. Warren and Rep. Omar Challenge U.S. OSHA on Evidence for Respirator Rollbacks 

Sen. Warren and Rep. Omar Challenge U.S. OSHA on Evidence for Respirator Rollbacks 

Sen. Warren and Rep. Omar Challenge U.S. OSHA on Evidence for Respirator Rollbacks 
EU 2027 Work Program Pulls Back on Sustainability

EU 2027 Work Program Pulls Back on Sustainability

EU 2027 Work Program Pulls Back on Sustainability
Latin American warehouse employee taking inventory of packages using a bar code reader and a digital tablet

ESPR Digital Product Passport requirements: a checklist for businesses

ESPR Digital Product Passport requirements: a checklist for businesses

SDS data quality is the accuracy, completeness and currency of the information in your safety data sheets (SDSs), and getting it wrong is expensive. Research by 3E and Verdantix found that every type of SDS data quality failure averages more than $1 million per event, across five risk areas: regulatory fines, worker safety, market access, product loss and AI built on poor data. In 3E’s on-demand SDS data quality webinar, 3E and Verdantix walked through the findings and what leading teams do differently. Here are the takeaways.

One wrong classification, a $10 million problem

A pharmaceutical company had everything lined up for a new product launch: shipping routes, packaging, a launch date and the celebration to go with it. Then a final check turned up a problem. The product’s classification was wrong for the destination jurisdiction.

Fixing it set off a chain reaction. Shipping had to be rerouted, packaging and labels redone, and the launch pushed back. Across the research, rerouting averaged about $2.8 million, repackaging about $2.5 million, and the opportunity cost of a delayed launch topped $3 million. Alan Johnson, managing director of Chemical and Workplace Safety at 3E, put the total at roughly $10 million. Checking classification up front would have avoided it.

That story came out of a joint study by 3E and Verdantix on the total value of quality SDS data, which measured what poor SDS data quality really costs. This article walks through the five risk areas the study covered, what leading teams do differently, and where to start in your own organization.

How 3E and Verdantix measured the cost of poor SDS data quality

SDS quality problems rarely show up as one line item. A wrong hazard classification touches regulatory, logistics, sales and safety budgets at once, so nobody sees the full bill. 3E teamed up with Verdantix, an independent research firm, to price it.

The study surveyed more than 50 EHS and product compliance leaders across three revenue bands, from under $50 million to $5 billion and above. It also included in-depth interviews with practitioners in chemical safety, hazard communication and product stewardship. Each interview ran more than an hour, often over several sessions.

Respondents estimated the direct and indirect cost of each failure type. Verdantix added desk research on costs disclosed by regulators and by companies that had been through an incident. The survey figures are self-reported midpoints, so treat them as directional rather than exact.

Every failure category averaged more than $1 million per event. The lowest, audit and regulatory risk, still came in around $1.1 million. As Verdantix’s Nathan Goldstein put it, no category is small enough to ignore.

Risk 1: Regulatory citations and fines

The fine is the smallest part of the bill. Hazard communication was the second most frequently cited OSHA standard in fiscal year 2025, and the two most common failures were SDS-related: no sheet on file, and no worker access to it. A serious violation is capped at roughly $20,000, but citations can repeat and cluster, and that figure can quickly reach six or even seven figures.

The indirect costs are larger. Respondents reported that insurance impacts alone could add almost $1.6 million on top of the direct costs. Customer audits are another drain. They can arrive unannounced, and one customer’s request can turn into five or six, pulling your team away from the work of staying compliant.

Teams that handle this well are inspection-ready every day:

  • Every product and site holds a current, local-language, GHS-compliant SDS.
  • Supplier and regulatory checks keep those sheets up to date.
  • An audit trail shows what changed, when, and which vendor version it replaced.
  • When an inspector asks for a sheet, they can answer in minutes, not days.

The audit trail matters most after an incident, when you need to produce the exact SDS that was on hand at the time.

Risk 2: Worker safety and emergency response

Human harm ranked as the top financial risk among category leaders, with environmental harm a close second. This is also where an SDS failure can be irreversible.

Two outside data points show the stakes. The United Nations puts toxic work exposure deaths at nearly 2.8 million a year, mostly from long-term illness rather than accidents. And the BlueGreen Alliance found that 30% of the SDSs it tested carried wrong hazard warnings.

One pharmaceutical interviewee described working with neurological drugs where responders have minutes, sometimes seconds, to decide on treatment, protective equipment and next steps. That decision depends on having the right sheet in hand. In the study, 47% of leaders expected delayed or incorrect emergency response.

Scale raises the exposure sharply. Among companies with $5 billion or more in revenue, average human harm costs were about 4.5 times those of the smallest firms, roughly $5.8 million against $1.3 million. Environmental harm rose from $0.8 million to $6.9 million across the same bands.

What leading teams do:

  • Write SDSs so the first aid, PPE and spill sections are easy to find and read in a hurry.
  • Translate key information into the languages their workers speak, and post it as safety cards or placards near the point of use.
  • Keep one validated record that flows to every site and system, instead of duplicate copies that drift apart.

Risk 3: Market access and trade disruption

Here, SDS quality hits revenue directly. These problems tend to surface late, just as a product is ready to ship, which is exactly when they cost the most. Classification errors found at the end of development mean re-planning routes, repackaging and sometimes finding a new logistics partner.

Line stoppages were rare: only one in five respondents expected them. But those who did put the direct cost at $6.1 million on average, before lost and late sales, rebooking and the extra calls and emails. By company size, the cost of a shipping disruption rose from $0.4 million at the smallest firms to $6.4 million at the largest.

The wider data is sobering. ECHA found that 44% of hazardous mixtures were misclassified. IATA counts more than 1.25 million dangerous goods air shipments a year that carry rejection risk, before you consider sea, rail or road.

One interviewee described missing an order deadline for a new market. The customer walked away from a multi-million dollar opportunity.

How leaders reduce the risk:

  • Classify for each destination market, not once for the product.
  • Generate country-specific SDSs and labels on demand.
  • Bring classification into R&D, so new formulations are built with downstream jurisdictions in mind.
  • Check dangerous goods classification for every transport mode, including water, road and rail.

This matters especially for companies expanding across regions. Johnson noted that North American companies entering Europe often struggle with the differences between the two regulatory scopes, and that this challenge still comes up with prospects.

Risk 4: Product loss, recall and held stock

These costs arrive after the product is made, when you have already sunk money into it. The most common failure respondents reported was wrong hazard communication labeling, and government studies back that up, with 44% showing incorrect hazard information. Held or quarantined stock happened less often, but its indirect cost was higher, around $2.6 million. A hold stops everything: nothing ships, and nobody touches the product until the problem is fixed.

Recalls can end companies. Goldstein pointed to the Takata airbag recall, which eventually bankrupted the manufacturer.

Goldstein named stock on hold as the cost companies most often overlook. It sits out of sight in a warehouse or chemical closet, but its labels and storage placement still depend entirely on the SDS.

Scale is the other challenge. One interviewee managed 5,500 SDSs worldwide, and Johnson said some companies hold more than a million. No team can re-read that many documents by hand, so leading companies take a different approach:

  • Profile their SDSs by risk and review the highest-risk records first, without ignoring the rest.
  • Track every change, including supplier updates and missing downstream-jurisdiction versions.
  • Get the SDS right first, then build labels and everything else from it.

Still, 38% of respondents expected some relabeling or rework as products went to market. Moving those decisions upstream would change that number.

Risk 5: AI built on poor data

The fifth risk is the newest and the least visible. AI tools and agents read the SDS data underneath them and make decisions from it. If a sheet is wrong, the error is copied and amplified at scale. Johnson put it plainly: AI exposes bad data faster, and the errors compound faster.

The study’s finding was that most AI failures trace back to the data, not the model. Goldstein compared it to building a house on a flawed foundation. Whatever you build on top will lean.

In compliance, near-correct is not enough. Johnson said a system that is 98% accurate can still mean someone is exposed to a hazard they didn’t know about.

Two things make SDS data fit for AI:

  • Structured, quality-controlled data: wrong classifications, CAS numbers or substance names become serious problems once they are repeated at scale.
  • Governance: you need to trace how data flows and what outcomes it produces, and control who can see which data.

Both point to one trusted source for SDS data. Johnson also noted that spreading SDS data across several providers raises total cost of ownership and makes quality harder to guarantee.

Five SDS data quality habits of teams that manage these risks well

Incidents cost far more than prevention. Across the interviews, five operating habits covered every risk the study identified:

  • Be audit-ready every day: check that SDSs are current, in the right language, and carry the right product and code numbers, not just that they are filed.
  • Put the right data in workers’ hands: anyone on the shop floor should be able to find the right SDS in seconds.
  • Clear products before they ship: check against destination rules before the order is placed.
  • Stop problems at the front door: verify SDSs and labels, and scan QR codes to confirm they lead where they should, before anything leaves your site.
  • Trust one source: feed every label, report and decision from a single trusted SDS record before layering on AI.

That single record reaches well beyond hazard communication. It supports emergency response, classification, market entry, and customer and regulatory inquiries. It also supports sustainability reporting, since for most companies the SDS is still the best place to find substances of very high concern, which reporting under CSRD and similar standards needs you to identify.

Quality also has to be monitored continuously. Formulations, ingredients and regulations keep changing, so ask whether your data is accurate, complete, accessible, current and defensible.

Where to start improving SDS data quality

You don’t need a full audit to see where you stand. Three steps give you a baseline:

  • Score your own risk: note where you have seen problems before, or where you believe the exposure is highest, across the product value chain.
  • Put a cost on it: include direct and indirect costs. Each function may see only its own slice, so look at the combined figure and use it as a benchmark to revisit.
  • Test defensibility: if an inspector arrived today, could you show that your SDS data is accurate, current and validated, and produce it in seconds or minutes rather than hours or days? Any gap becomes part of your risk profile, which takes you back to step one.

The full Verdantix and 3E report on SDS data quality goes deeper into how the numbers vary by industry and company size. If you want to talk through how your own SDS data holds up, 3E Protect gives you a single, quality-controlled source for SDS management.

Hear the complete discussion: watch the SDS data quality webinar on demand.

Frequently asked questions

  • What is SDS data quality?

    SDS data quality describes how accurate, complete, current and accessible the information in a company’s safety data sheets is. It covers hazard classification, composition, first aid, PPE and transport data, and whether each SDS is available in the right language for every market and site.

  • How much does poor SDS data quality cost?

    In the 3E and Verdantix study of more than 50 EHS and product compliance leaders, every SDS failure category averaged more than $1 million per event. One misclassification at a pharmaceutical company cost roughly $10 million in rerouting, repackaging and a delayed launch.

  • What are the five risks of bad SDS data?

    The 3E and Verdantix research identified five risk areas: regulatory citations and fines, worker safety and emergency response, market access and trade disruption, product loss, recall and held stock, and AI built on poor data.

  • Why does SDS data quality matter for AI?

    AI tools and agents make decisions from the SDS data underneath them. If that data is wrong, the error is copied and amplified at scale. According to the 3E and Verdantix study, most AI failures trace back to the data rather than the model, so structured, governed SDS data is a prerequisite for using AI in compliance.

  • How can companies improve SDS data quality?

    Start by scoring where your SDS risk is highest, putting a combined cost on it, and testing whether you could produce accurate, current SDSs for an inspector in minutes. Leading teams then keep one trusted SDS record that feeds every label, report and decision.

Related Resources

Electric Engine Autonomous Manufacturing Plant With Robotic Arms: Diverse Specialists Talking About Modern Machinery Production. Industrial Engineer, Designer, And Assembler Discussing Project.

News

The ROI of Digital Product Passports: benefits beyond compliance
The ROI of Digital Product Passports: benefits beyond compliance
Sen. Warren and Rep. Omar Challenge U.S. OSHA on Evidence for Respirator Rollbacks 

News

Sen. Warren and Rep. Omar Challenge U.S. OSHA on Evidence for Respirator Rollbacks 
Sen. Warren and Rep. Omar Challenge U.S. OSHA on Evidence for Respirator Rollbacks 
EU 2027 Work Program Pulls Back on Sustainability

News

EU 2027 Work Program Pulls Back on Sustainability
EU 2027 Work Program Pulls Back on Sustainability
Latin American warehouse employee taking inventory of packages using a bar code reader and a digital tablet

News

ESPR Digital Product Passport requirements: a checklist for businesses
ESPR Digital Product Passport requirements: a checklist for businesses

View All 3E Resources

View All 3E Resources