Most SDS software demos look alike for the first ten minutes: a search box, a list of documents, a PDF viewer. The differences show up when you hand the tool your own documents: a supplier's new revision, an old scanned sheet, a product your team calls something other than the manufacturer does. This guide gives you a vendor-neutral set of criteria, a demo scorecard you can print and fill in, and the test tasks behind it. What NeoTech's platform has been confirmed to do is kept in its own labelled box, so you can tell evaluation advice apart from product facts.
Start with the requirement the software has to support
OSHA's Hazard Communication Standard requires an employer to have a safety data sheet in the workplace for each hazardous chemical it uses, and to keep those sheets readily accessible during each work shift to employees in their work areas. Electronic access is allowed, as long as it creates no barriers to immediate employee access in each workplace.
That sets the first test for any SDS tool: can an employee in their work area, on any shift, reach the right sheet without help? Search speed, extraction and reporting features matter, but they come second. If the tool needs a login that floor staff don't have, or depends on a network that drops, plan and test the fallback (a printed copy, a kiosk device) as part of the evaluation, and see SDS binder requirements for a fallback access checklist.
Source29 CFR 1910.1200: OSHA Hazard Communication StandardPrepare before the first demo
Vendors demo on clean sample data. Bring your own, and use the same set with every vendor so the results are comparable.
Bring to every demo
- Three to five current supplier SDS PDFs you actually use
- A newer revision of one product you already hold, plus the older version
- One older or scanned document (for example, a legacy MSDS-format sheet)
- A mixture SDS with ingredient ranges and, if you have one, a trade-secret ingredient
- A product your site names differently from the manufacturer's product name
- A sample of your inventory list (anonymized if needed) to test import
- A list of who needs access: roles, sites, shifts, contractors and temporary staff
- Your current cost baseline: subscriptions, staff time, binder printing
Evaluation criteria
Use these criteria to structure questions before and during the demo. Ask to see each one working on your documents; a feature that is only described has not been evaluated yet.
| Criterion | What to ask | Evidence to look for |
|---|---|---|
| Worker access | How does an employee on the floor reach a sheet? What happens if the login or network fails? | Access shown from a shared or floor device, and a fallback you can maintain |
| Search | Can you search by manufacturer, product name, synonym and product code? How are near-matches shown? | Your differently named product found during the demo |
| Supplier revisions | When a supplier issues a new revision, how does it get into the library, and who decides it replaces the old one? | The steps shown on your revised PDF, with the earlier version still retrievable |
| Uploaded versions | What happens to a PDF you upload? Is data extracted, and is it reviewed before it goes live? | A review step with a visible status before approval |
| Migration | How do existing documents and inventory lists come in? Which file types, size limits and column rules apply? | Your sample inventory imported, with problems reported rather than dropped |
| Permissions | Which roles exist? Can someone view without editing? Who can approve an SDS? | A role list and a view-only user demonstrated |
| Review controls | Can a record be rejected or sent back? Can a reviewer correct extracted values before approval? | Statuses and the correction step shown on a real upload |
| Inventory linkage | How is an SDS tied to inventory rows at each location? How are unmatched rows found? | Matched and unmatched rows identified on your data |
| Audit and export | Can you export activity, and get your original PDFs and data back out? | An exported file opened during the demo |
| Cost | What is included, and what is billed separately (users, sites, documents, setup)? | A written quote or published price covering your user and site count |
Weight each criterion 1–3 for your site before the demos, so the scoring reflects your priorities rather than the order vendors present in.
A demo scorecard you can fill in
Print one copy per vendor. For each task, score 2 if it was shown working on your document, 1 if it was shown only on the vendor's sample data or partly worked, and 0 if it was only described or not available. Multiply by the weight of the matching criterion.
| Task | Source document | Expected result | Evidence to capture | Questions to ask |
|---|---|---|---|---|
| 1. Find an existing product | A product you hold, searched by your site's name for it | Correct manufacturer and product returned; near-matches distinguishable | Screenshot of the results | How are duplicates and near-matches handled? |
| 2. Upload and review a supplier PDF | A current supplier SDS (PDF) | Extracted fields shown for review before the record goes live | Record status before and after approval | Which fields are extracted, and how is one corrected? |
| 3. Replace a revised SDS | The newer revision of a product already loaded | New revision becomes current; earlier one retrievable | Revision dates of both versions | Who decides that a revision replaces the old one? |
| 4. Reach an SDS as a worker | Any approved SDS | Sheet opened from a floor device or fallback within your target time | Time taken and login steps needed | What happens if the network or login fails? |
| 5. Import your inventory | Your sample inventory file | Rows imported; unmatched columns or bad rows reported | Import summary or error list | What are the file type, size and row limits? |
| 6. Link SDSs to inventory rows | The rows just imported | Matched and unmatched rows identified | Count of matched vs unmatched rows | How is an unmatched row resolved? |
| 7. Restrict a user | A view-only test account | User can view records but not change them | The refused edit attempt | Which roles exist, and who can approve an SDS? |
| 8. Export and audit | Activity from this demo session | Export shows who did what and when; original PDF downloadable | The exported file | How long is activity kept, and in what format? |
| 9. Retire an SDS | A product you no longer use | Record removed or archived, with a warning about dependent items | What the tool shows before removal | How do we keep a record of superseded sheets? |
Maximum score per task is 2 × the criterion weight. Note any task the vendor asked to skip. That is evidence too.
On task 9: OSHA's records-access rule lets employers stop keeping a specific SDS as long as some record of the substance's identity, where it was used and when it was used is kept for at least 30 years. Ask how the tool supports that record before you delete anything.
Source29 CFR 1910.1020: Access to employee exposure and medical recordsRunning the extraction and review test
Task 2 deserves the most time, because extracted data that nobody checks can flow into inventory totals and reports. Use a mixture SDS with ranges, and a clean PDF as well as a scanned one. If NeoTech is on your shortlist, the AI SDS extraction page describes its upload-and-review path; run the same steps on it as on every other tool.
- Upload the supplier PDF and note how long it takes for the record to reach a reviewable state.
- Compare at least these fields against the PDF: product identifier and code, manufacturer and emergency phone, revision date, each ingredient's CAS number and concentration range, hazard classes, signal word, and density or specific gravity with its unit.
- Deliberately correct one field, save, and confirm the corrected value, not the original extraction, is what appears wherever the record is used.
- Reject or cancel a second upload and confirm it does not appear in the approved library.
- Repeat with the scanned document and record which fields came back blank or wrong.
- Ask who can approve a record, and whether approval can be bypassed or validation loosened.
For a field-by-field review method with a worked example, see how to review AI-extracted SDS data.
Cost: compare what you pay for, not just the list price
Two quotes with similar monthly figures can cost very different amounts once setup and internal time are included. Ask each vendor to price the same scenario: your number of sites, users and documents.
- Billing basis: per user, per site, per document, or a plan tier
- Setup and migration: who loads your existing documents and inventory, and at what cost
- Internal time: reviewing extracted data and approving records takes staff hours in any tool
- Training and support: what is included and what is billed separately
- Fallback access: printing or devices needed so workers can reach sheets
- Contract terms: renewal, price changes, and getting your documents and data out if you leave
For a NeoTech comparison point, NeoTech pricing is published by month, with a 30-day free trial that needs no credit card, which is enough time to run the scorecard tasks on your own documents.
Score and decide
Total the weighted scores, then look past the total at any task scored 0 on a criterion you weighted 3. A tool that wins on search but cannot show how a revised sheet replaces the old one may cost you more in manual work than it saves.
Ask vendors to confirm unclear answers in writing, and pilot the leading option in one work area before rolling it out. For the operating workflow after you choose, see how to manage safety data sheets.
What NeoTech can demonstrate
The box below is limited to functions confirmed in NeoTech's in-app help center and deployed build. Anything not listed, ask to see working in the demo like any other vendor's claim.
Regulatory statements in this article were checked against the official sources linked above on .