This guide offers a scorecard to help you evaluate any tool according to your specific needs, whether you use spreadsheets, outsourced authors, or software. It is designed for US small and mid-size chemical manufacturers, where one or two people often manage the entire SDS program.
This article is informational, not legal advice. Check the primary sources linked below and confirm with your regulatory professional.
What an SDS authoring tool actually has to produce
An SDS follows a standardized 16-section format. In the US, OSHA’s Hazard Communication Standard (29 CFR 1910.1200(g)(2)) specifies these sections: identification; hazard(s) identification; composition/information on ingredients; first-aid measures; fire-fighting measures; accidental release measures; handling and storage; exposure controls/personal protection; physical and chemical properties; stability and reactivity; toxicological information; ecological information; disposal considerations; transport information; regulatory information; and other information, including the date of preparation or last revision. The minimum required content for Sections 1–11 and 16 is detailed in Appendix D. The GHS guidance (GHS Rev. 11, Annex 4, A4.2.3.1) and REACH Article 31(6) both list the same 16 headings, with minor differences in wording for the EU.
The structure is straightforward, but content accuracy relies on three key factors.
Classification. Section 2 is a result. Someone or something has to work out the hazard classes and categories first. Under the HCS, chemical manufacturers and importers must evaluate their chemicals and classify them (paragraph (d)(1)). Employers are not required to classify chemicals unless they choose not to rely on the classification performed by the chemical manufacturer or importer. For a mixture, the HCS points to the procedures in Appendices A and B. For health hazards, Appendix A sets a sequence: use test data on the mixture if you have it, then bridging principles, then calculation methods such as concentration cut-offs. Carcinogenicity, germ cell mutagenicity and reproductive toxicity are an exception: mixtures are classified from the ingredient information unless a case-by-case justification supports classifying the mixture as a whole (A.0.4.2). Every other section that mentions hazards depends on that result.
Regulatory data. Section 3, Section 8 and Section 15 depend on data about the ingredients: harmonized classifications, exposure limits, and lists such as REACH or Prop 65. That data changes, and it differs by region.
Regional rules. Canada, the EU, Great Britain and Mexico each specify their own format, required content and language. In the US, OSHA does not enforce the information content of Sections 12 to 15 (Appendix D: they "may be included in the SDS, but are not mandatory"), but all 16 headings must still appear in order (29 CFR 1910.1200(g)(2)(xii)-(xv)). The EU and Great Britain list all 16 headings in Article 31(6).
The manufacturer, importer, or supplier remains responsible for the published SDS (29 CFR 1910.1200(g)(5)). While tools can assist, your company is accountable, so ensure a person can review and verify the tool’s output.
Five ways to produce SDSs, and when each fits
There are five main approaches to producing SDSs, which can be used individually or in combination. This comparison focuses on categories rather than specific brands.
| Approach | How it works | Where it fits | Where it may strain |
|---|---|---|---|
| Spreadsheets and Word templates | Your team writes each SDS by hand in a template. | A handful of simple products in one region. | All classifications and phrases are manual. Consistency relies on individuals, and updates require manually searching through files. |
| Enterprise EHS or product-compliance suites | SDS authoring is one module of a large platform. | Larger companies with many sites and a dedicated system team. | Implementation is typically a larger project. Assess whether it aligns with the needs of a small team. |
| Point solutions | A focused authoring tool with a phrase library and templates. | Teams that want to author in-house, one tool for one job. | Depth varies: how classification is calculated, how the regulatory data is kept current, how many regions. |
| AI-native tools | AI reads supplier documents; some tools also use it to draft text. | Teams that want speed on extraction and drafting. | The main consideration is which decisions are made by AI. Refer to the classification criterion below. |
| Outsourced authors | A firm writes SDSs for you. | No in-house expertise, or overflow work. | You depend on their turnaround. Costs and control differ by contract. Classification decisions can end up outside your team unless you review them. |
Many companies combine these approaches. The scorecard below applies to all methods. For a detailed outsourcing comparison, see SDS authoring services vs. software.
The scorecard: ten criteria for comparing SDS authoring software
Score each option from 0 to 3 for each criterion (0 = no, 1 = partly, 2 = yes, 3 = yes and demonstrated on your own data). Assign higher weights to the criteria most important to your business. For example, companies shipping only within the US may assign lower weights to regions and languages.
1. Data extraction
Most SDS work starts with supplier SDSs: ingredient identities, concentrations and hazard data. Ask:
Does the tool read supplier PDFs, or do you re-type them?
Does every extracted field carry a confidence score?
Can you review the extracted values beside the source document?
Is accuracy stated per field, and what does the vendor say it covers?
2. Classification method and auditability
This criterion has the most impact. Ask:
Is the classification calculated from composition and ingredient data, or entered by a person, or suggested by an AI model?
If it is calculated: which rules, for which jurisdiction? Additivity formulas, concentration cut-offs, bridging principles?
Can you see the rule behind each result?
Which classifications does the tool calculate by fixed rules, which does it leave to a person, and how is that decision recorded?
What happens when a concentration range straddles a cut-off? A good tool flags it. A weak one picks a value silently.
Can you export the audit trail for an inspector or a customer?
Hazard classes are assigned under published criteria. For some hazard classes, the criteria call for expert judgment or weight of evidence (29 CFR 1910.1200 App. A, A.0.1.2 and A.0.3). If a language model produces the classification, ask how the result can be checked. See why classification cannot be a guess.
3. Regulatory database management
In SDS authoring software, regulatory database management means what data sits behind the drafts and who keeps it current. Ask:
Which lists and data sets are included: harmonized classifications, exposure limits, Prop 65, REACH lists, toxicological and physical-chemical data?
How many substances does the database cover, and how does that compare with your ingredient list? Ask them to run your own top 50 ingredients.
How fast do updates arrive after a regulator publishes a change?
When a listed substance changes, does the tool tell you which of your SDSs are affected?
4. Regions and languages
Which regions can the tool author today, as a named list? Which are still on the roadmap rather than live?
Are language variants generated from the same product record as regional variants?
Do standard phrases come from a controlled library, one approved translation per phrase?
Are region-specific formats handled, not just translated text?
Language rules vary. The US HCS requires an English SDS. Canada's Hazardous Products Regulations require both English and French. EU REACH requires an official language of each Member State where the product is placed on the market, unless that Member State provides otherwise (Art. 31(5)). Mexico's NOM-018 requires Spanish. Our multilingual SDS authoring guide lays out each rule with its source.
5. Review workflow
Can any SDS publish without a person's approval?
Does the tool flag what needs a decision (low-confidence values, unresolved classifications, sections affected by a rule change) instead of asking you to reread everything?
Is there a record of who approved what, and when?
6. Migration
Ask:
How does your existing library get in? PDFs, spreadsheets, an export from another system?
Who does the work: you, the vendor, a consultant?
How long from contract to first published SDS? Ask for the answer in weeks, and ask what it depends on.
7. Integration and export
Can you export all your data in a structured form at any time? Is there a fee?
Is there an API to connect your ERP, PLM or EHS system?
What happens to your data if you leave?
8. Data privacy
Are your compositions used to train the vendor's models, in any form?
Where is your data stored, and who can see it?
Get the answers in the contract.
9. Support model
Who answers a regulatory question: a support agent or a regulatory specialist?
What is included in support, and what is extra?
10. Total effort
Evaluate total effort, not just licensing costs. For each option, estimate the hours required per new SDS, per revision, and per new region, including setup, migration, and review. Tools that automate routine tasks and flag exceptions can save significant time. Ask each vendor to demonstrate their process using your own documents.
A scoring sheet you can copy
| Criterion | Weight (1-3) | Option A (0-3) | Option B (0-3) | Option C (0-3) | Evidence / notes |
|---|---|---|---|---|---|
| 1. Data extraction | |||||
| 2. Classification method and auditability | |||||
| 3. Regulatory database management | |||||
| 4. Regions and languages | |||||
| 5. Review workflow | |||||
| 6. Migration | |||||
| 7. Integration and export | |||||
| 8. Data privacy | |||||
| 9. Support model | |||||
| 10. Total effort |
Multiply each score by its weight and total each column. If any option scores 0 on classification, review, or privacy, consider it a disqualifier regardless of the overall total.
Small chemical companies: what changes when you pick SDS authoring software
Several factors are particularly important for small and mid-size US manufacturers.
Your specific ingredient list is more important than the overall database size. A tool may cover many substances but still miss those you use. Always test with your own ingredients.
Consider setup time as a direct cost. Projects that take months can delay other essential work.
Often, a single person manages the program. A review queue that highlights exceptions can save more time than any individual feature.
Customers may request justification. An audit trail allows you to quickly explain classification decisions.
Start with a pilot. Test a subset of products before migrating your entire SDS library.
Questions to ask in a demo
For a meaningful demo, bring a selection of your own supplier SDSs (ideally including a mixture with a concentration range) and a list of the regions where you distribute products. Ask the vendor to demonstrate the following:
Importing a supplier SDS, showing field-level confidence scores and the original source for each extracted value.
How the tool classifies your mixture, with visibility into the rules or logic behind each result.
The system’s handling when a concentration range overlaps a regulatory cut-off.
Generating SDSs for the same product in multiple regions and languages, all from a single record.
The review and approval process: can an SDS be published without manual review?
The audit trail for a published SDS, tracing from source data to final output.
Notifications or flags when a listed substance changes and impacts your SDSs.
Exporting your data: what is the process, how long does it take, and are there any costs?
Data privacy terms: will your formulations be used to train any model?
The migration plan for your SDS library, including estimated timeline in weeks.
Who provides regulatory support and how accessible are they?
If a vendor will not demonstrate with your own data, factor this into your evaluation score.
How Valenc approaches it
Valenc is a purpose-built software platform, not an authoring service, that works with third-party partners who provide SDS authoring services. The foundation is a robust product record. Valenc’s AI-powered agents extract data from supplier SDS PDFs, assigning a confidence score to each field for transparent review alongside the source document. This data is then enriched using Valenc’s database of approximately 55,000 chemicals, including global regulatory lists, toxicological and physical-chemical data, classifications, exposure limits, and Section 15 data such as Prop 65 and REACH.
Valenc delivers classification through a transparent, deterministic rules engine, never a black-box language model. For each jurisdiction, the platform applies GHS additivity, concentration cut-offs, and bridging principles, ensuring every classification decision is both visible and auditable. All 16 SDS sections are drafted using a curated phrase library in over 25 languages, enabling efficient, region-specific authoring from a single product record. Valenc currently supports the US, Canada, Mexico, the EU, the UK, and the UN GHS module, with new regions being added. GHS-compliant labels are generated seamlessly alongside SDSs.
No SDS is ever published without human approval. A dedicated review agent and a comprehensive audit trail provide full transparency: every change is tracked from the original PDF through to the final document. Valenc’s intelligent monitoring agents proactively track regulatory updates from OSHA, EPA, REACH, and CLP, clearly flagging affected SDSs in your library. Your proprietary data remains fully confidential and is never used for training external AI models.
Migration is seamless: you can go live within weeks. Your data remains fully portable, with no export fees, and integration is simple via REST API to your existing ERP or PLM systems. For more on key capabilities, see SDS authoring software, GHS classification, and SDS data extraction.
To evaluate fit, take advantage of free SDS indexing. Through December 31, 2026, chemical manufacturers can submit up to 100 SDSs for complimentary indexing through a secure portal and receive an Excel index within three business days, no obligation required.
Frequently asked questions
What is the best SDS authoring software for small chemical companies?
There is no universal best solution. The optimal choice is the platform that aligns with your strategic priorities and scores highest on your tailored evaluation scorecard. For most small companies, this means prioritizing auditable classification, up-to-date regulatory intelligence, rapid migration, and a review workflow designed to surface only true exceptions.
Do I need SDS authoring software, or can I use templates?
Templates suffice for a handful of simple products in a single region. As your product portfolio and geographic footprint expand, manual processes quickly become a bottleneck and a primary source of risk and rework. Evaluate your resource investment per SDS and per revision before committing to a manual approach.
What is the difference between SDS authoring and SDS management?
Authoring refers to the creation of SDSs for the products you manufacture. Management encompasses the storage, distribution, and tracking of SDSs you receive or publish. Some solutions address both needs. Ask each vendor which area their tool is purpose-built to support. For further detail, see the SDS management software guide.
Can AI write my SDS?
AI excels at extracting information from supplier documents and supporting draft reviews. However, hazard classification must follow published regulatory criteria. Some classes still require expert human judgment. For transparency, ask every vendor to clarify which steps are AI-driven, which are rule-based, and which require human expertise and approval.
How do I move my existing SDS library into a new tool?
Understand the migration process in detail: how your PDFs and spreadsheets are imported, who manages the workload, and the expected timeline. Always request a demonstration of both import and export capabilities before making a commitment.
Does the tool make me compliant?
No software guarantees compliance on its own. Tools empower your team to meet regulatory obligations, but ultimate responsibility for each SDS remains with your organization. Ensure a qualified professional reviews and approves every SDS, and consult primary sources or regulatory advisors for final guidance.
See it on your own supplier SDSs
Valenc sets up a walkthrough on your own supplier SDSs. You see extraction, classification and a review-ready draft on chemistry you already know. Request a demo.
Related guides: multilingual SDS authoring for regulatory teams, OSHA HazCom standard: what it means for SDS and label authoring, the HCS 2024 checklist, and the SDS management software guide.