You already have a QMS. The question is whether you can defend it.
Every regulated site I have worked with already has a quality management system: the quality unit, the procedures, the batch records, the people who sign them. "Do we need a QMS or an eQMS?" is therefore the wrong question, and it wastes the meeting it is asked in. The useful question is narrower and more uncomfortable: can anyone here defend the system we have without dragging three people into a room and rebuilding the story from memory?
The pattern that produces a "no" is familiar and it starts harmless. Folders. A shared drive. Approval in email. One extra tracker "just to be safe". Then production gets busier: a procedure revises and operators still pull last month's print from a drawer; training looks complete on a sheet that does not know about the revision; a deviation needs follow-up and three people argue which file owns it; change control becomes a round of reminders. Nobody in that room was looking for software. They were looking for the record.
What is an eQMS? is the short vocabulary table. This is the pharma version: what the regulations name, what software does and does not buy, and what has to change in the daily work before the label "eQMS" means anything. How to score vendors is the buyer guide. Read this first if the argument in the room is still "QMS versus eQMS".
A QMS is not software
ICH Q10 (Step 4, 4 June 2008) defines a Pharmaceutical Quality System as a "management system to direct and control a pharmaceutical company with regard to quality." The definition is based on ISO 9000. It does not mention a product, a database, or a vendor.
Q10 is explicit that it is "not intended to create any new expectations beyond current regulatory requirements." Regional GMP already asks for written procedures, records, a quality unit, CAPA-shaped investigation, and control of change. Q10 names four elements that sit on top of that foundation:
- process performance and product quality monitoring
- a CAPA system
- a change management system
- management review of process performance and product quality
Those elements are the QMS. They can live in binders, in a tangle of workbooks, or in a governed system. The medium is not the system.
21 CFR Part 211 does not require an eQMS. It requires the quality work. Part 11 arrives only when those records are electronic. Annex 11 arrives when a computerised system is used as part of GMP-regulated activities: "The application should be validated; IT infrastructure should be qualified." Neither text invents a category called eQMS. Industry did that, as a name for the software that runs the processes Q10 already listed.
So the useful split is:
| What it is | What it is not | |
|---|---|---|
| QMS | The management system: procedures, records, responsibilities, CAPA, change, review | A licence. A module list. A scan of last year's SOPs |
| Document system | Storage and versioning of controlled documents | CAPA, training-on-change, deviation follow-up |
| eQMS | A computerised system used to run QMS processes, with the records and the connections in one place | Automatic compliance. A finished validation. A smaller quality unit |
If you only digitised documents, you still have a QMS. You also have a document system. You do not yet have an eQMS. That three-way split is in the guide. The rest of this post is what fails when people skip it.
Cosmetic digitization is still a traditional QMS
The failure mode is not "they used paper." Paper can be a controlled QMS. The failure mode is calling it electronic while the process still splits.
A drive full of signed PDFs. A spreadsheet that is the real CAPA list. Change requests in email. Training on a sheet that does not know which revision is in force. Each piece can look tidy. Together they are the same reconstruction problem the opening scene described.
Hybrid paper-digital systems are the next post in this wave: the seam where one half of the record is a binder and the other half is a screen. This post stops one step earlier. Even a fully on-screen setup is a traditional QMS if the four Q10 elements do not share a path.
Cosmetic signs:
- two places that both look current
- approval that lives in a thread, not on the item
- training completion that cannot name the procedure in force
- an inspector question that takes a room, not a search
The pixels are not the upgrade. The connection is.
What changes in the daily work
| Area | Traditional QMS (paper, drive, or cosmetic eQMS) | eQMS worth the name |
|---|---|---|
| Document control | Latest copy is a hunt; superseded prints linger | Current version at point of use; history readable, not competing |
| Audit trail | Sequence rebuilt after the fact from emails and "who remembers" | Activity stays on the record, in a form a reviewer can open |
| CAPA | Separate tracker, reminders, overdue chasing | Action, owner and effectiveness still attached to the source event |
| Change control | Implementation path hidden in side conversations | Impact, documents, training and verification on the same change |
| Training | Completion sheets drift from the procedure in use | Evidence tied to the revision that actually governed the work |
| Inspection support | Fire drill: exports, screenshots, a narrator | Retrieve the story from the workflow |
That table is not a feature list. It is Q10's CAPA and change-management elements, plus the documents and people those elements already implied, tested the way an inspector tests them.
For QA, the daily change is fewer handoffs to chase and less stitching. For the quality unit that has to defend a record, the change is that approvals, history and the linked change can be shown without assembling them. For operators, the change is that the procedure in their hand matches the one in the system — the same gap document control walks from the line.
If none of those burdens move, you changed the screen.
The first question that tells you which one you have
Can you show the path from deviation to CAPA to change without opening four systems?
If the answer is export files and explanation, the system has already leaked control. Deviation, CAPA and change is what that leak looks like after you already bought software. Catch the shape here, while the argument is still "do we need an eQMS?"
A second question, for sites that think they already moved: where does the unofficial original live? If a workbook, a personal folder or a printout is still the thing people trust, the QMS never left. The software is a copy.
What a poor eQMS choice leaves behind
A weak selection does not solve the old problem. It repackages it.
- the same inspection findings, in a digital wrapper
- more QA time spent explaining the tool
- duplicate trackers kept "just in case" — two originals
- training, CAPA and change still living in separate screens
- a re-validation project when the team has to switch again
If the software does not change the workflow, it only changes the login. Connected quality records is the post-purchase version. The buyer-guide test still applies: retrieval without reconstruction, or you are not done.
What a bad migration looks like
Moving from "QMS" to "eQMS" on a project plan does not automatically improve control.
A bad migration gives you old process habits inside new software, unused screens, a side tracker because nobody trusts the new one, confusion over where the real record lives, and audit prep that still depends on exports and memory.
That is why the software decision matters less than the workflow design. Map the four Q10 elements onto the processes that hurt in the last inspection. Those are the processes that have to move as a path, not as five separate go-lives. A document-only first phase is a document system. Call it that. Do not call it the eQMS until CAPA, change and training share the record.
Validation does not vanish because the vendor sold a pack. Annex 11 still asks the regulated user to assess the supplier and to validate the application. The CSA versus CSV post is the method; the buyer guide is the bake-off. Here the point is narrower: buying the licence is not implementing Q10.
What still does not change
The quality unit still owns the judgements. Procedures still have to be followed. Product quality still sits with the site, not with the vendor.
ICH Q10's three objectives — product realisation, a state of control, continual improvement — do not transfer to a SaaS agreement. Management review still has to happen. Knowledge management and quality risk management are still enablers, not modules you tick.
Part 11 does not get you a QMS. It tells you what an electronic record has to carry if the predicate rule already required the record. A site can have a thorough paper QMS and no Part 11 scope. A site can have Part 11-shaped software and a QMS that still splits. Those are different conversations.
How to tell which you need
You already have a QMS if you manufacture under Part 211 (or the QMSR / ISO 13485 for devices). The live question is whether that QMS can still be defended at the current volume of work.
Walk one recent inspection question.
- Which Q10 element did it land on — CAPA, change, monitoring, review?
- How many places did you open to answer it?
- Was the unofficial original in a file the system does not know about?
- How long did the reconstruction take?
- If that question arrived unannounced tomorrow, would the time get shorter?
If 2 is more than one governed record, and 4 is measured in hours, the QMS is leaking. Software that digitises the same split will not close it. Software that keeps the path on the work might. That is the only "versus" that earns an eQMS project.
Download the eQMS RFP scorecard for the vendor bake-off that follows. Do not start the bake-off until this walk has a written answer.
How Complere holds the process together
The QMS stays the QMS. Procedures, judgements and release decisions remain yours. What changes is that the work those judgements sit on does not scatter when it moves.
CAPA, change and training can be read with the event. You are not assembling the inspection answer from a tracker, a mailbox and a completion sheet.
History is on the item. Who did what, when, and what the record said before, is something a reviewer opens, not something a room reconstructs.
The current procedure is the one in use. Superseded versions stay readable. They do not compete for the operator's attention.
What good looks like is still your team's call. The software's job is to make the record behind that call cheap to produce and hard to lose, which is the only sense in which an eQMS is worth paying for.
