How to Choose an Automation Vendor for Medical Device Manufacturing

MEPSCo GAMP

Choosing an automation vendor for medical device manufacturing requires evaluating capability, GAMP documentation support, track record on similar projects, and financial stability — not just price. A structured scoring methodology applied before selection prevents the change orders, schedule slippage, and underperforming equipment that result from choosing the lowest bid on an incomplete specification.

TL;DR

Vendor selection is a qualification decision, not a price decision
A properly written URS is required before any vendor is contacted — without it, quotes aren’t comparable
The Kepner-Tregoe (KT) analysis produces a weighted, defensible selection recommendation management can stand behind
On two projects in 25 years, management overrode the KT recommendation and selected the lowest bid — both came in late, over budget, and underperforming

Two projects. Both times the same thing happened.

The KT analysis identified the right vendor. The scoring was clear — capability, GAMP documentation depth, track record, responsiveness, financial stability. One vendor came out ahead. It wasn’t the lowest bid.

Management selected the lowest bid both times.

Both projects came in late. Both came in over budget. Both produced equipment that underperformed against specification. Both required significant additional engineering time and cost to recover — more, in both cases, than the price difference between the recommended vendor and the one that was selected.

I’m not telling these stories to complain about management. I’m telling them because they’re the clearest illustration of why vendor selection in medical device automation is not a price decision. It’s a qualification decision. And the discipline to make it a qualification decision — consistently, with documentation that defends the recommendation — is exactly what the process described here is designed to provide.

This is also how Equipment Procurement & Vendor Management works at MEPSCo. The process didn’t emerge from a framework. It came from 96+ equipment procurement projects in medical device manufacturing, and from the two projects where the right framework was ignored.

Before a Vendor Is Contacted: The URS Comes First

The most common vendor selection mistake isn’t how vendors are evaluated. It’s that they’re contacted before the requirements are defined.

When an RFQ goes out before a proper URS exists, every vendor quotes against their assumptions. One assumes a basic HMI. Another quotes a full data acquisition system. One assumes manual changeover. Another quotes automated tooling. The price spread on a mid-size filling line can run $700k+ between vendors quoting the same general request against completely different assumptions.

The manufacturer receives three quotes it can’t compare. Leadership asks which one to pick. The answer — because the scope was never defined — is that nobody knows.

This is where vendor selection goes wrong before it starts. A Budgetary URS — complete enough to produce realistic, comparable quotes — must exist before any vendor is contacted. That URS defines what the vendor is quoting against. When every vendor receives the same complete requirements, they quote to the same scope. The price differences become capability differences, not assumption differences. Now the selection decision has something to work with.

The URS Development process MEPSCo uses builds the Budgetary URS from stakeholder interviews — every person who touches the equipment has a requirement. Operations, maintenance, validation, quality, management. The URS captures all of them before a vendor sees it.

What a Vendor Quote Actually Tells You

Once the URS exists and quotes come in, the instinct is to compare prices. That instinct is wrong — or at least incomplete.

A quote tells you what the vendor believes the machine will cost to build to their interpretation of the requirements. It doesn’t tell you whether they can actually build it. It doesn’t tell you what their change order history looks like on similar projects. It doesn’t tell you whether their GAMP documentation staff will still be there in six months when validation starts. It doesn’t tell you whether they’ve built a 200 PPM filling machine before or whether this is their first one.

Price tells you the number on the bottom of the page. The number on the bottom of the page is one input into the decision. It’s not the decision.

Three questions a quote doesn’t answer, but the selection process must:

Can they build this specific machine? Not machines in general. This machine. If the URS specifies a 200 PPM continuous filling line with 100% vision inspection and automated pump calibration, the question isn’t “do they build filling lines” but “have they built this kind of filling line, and do they have the references to prove it?”

Who handles the GAMP documentation? Vendors build equipment. GAMP documentation isn’t their core competency — it’s a service they offer because customers ask for it. Ask who at the vendor writes the documentation. Is it a dedicated person? An engineer who does it as a secondary responsibility? What’s their turnover rate in that role? Validation happens six or more months after machine delivery. The documentation you receive at FAT may be written by someone who wasn’t there for the build.

What does their change order history look like? Ask references specifically about change orders — whether they happened, what drove them, and how the vendor handled them. A vendor with a pattern of change orders on similar projects will generate change orders on this one. The URS eliminates gaps. The vendor’s behavior determines whether they exploit the gaps that remain.

The Kepner-Tregoe Methodology Applied to Vendor Selection

The Kepner-Tregoe (KT) methodology, developed by Charles H. Kepner and Benjamin B. Tregoe at the RAND Corporation in 1958, is a structured decision-making framework that evaluates options against weighted criteria. It was designed specifically for high-stakes decisions where price alone isn’t sufficient and where the selection has to be defensible after the fact.

Applied to automation vendor selection, it works like this.

Define the must-haves. Before scoring begins, define the requirements a vendor must meet to be considered at all. These are binary — pass or fail. A vendor who can’t provide references on similar projects in FDA-regulated manufacturing doesn’t get scored; they get disqualified. A vendor who can’t demonstrate GAMP documentation capability doesn’t get scored; they get disqualified. Musts are not negotiable.

Define the weighted criteria. For vendors who pass the must-haves, score them against weighted criteria. Each criterion gets a weight reflecting its relative importance to this project. The weights aren’t fixed — they shift based on project complexity, timeline, and the specific documentation requirements.

The criteria MEPSCo uses across equipment procurement projects:

Technical capability (weight: high). Has the vendor built equipment of this type, at this scale, for this process? Not general automation capability — specific capability for this URS. References from customers who can speak to the specific equipment type are required. Vendor claims without references are discounted.

GAMP documentation capability (weight: high). Does the vendor have a dedicated GAMP documentation function or does documentation fall to engineers as a secondary responsibility? Who writes the protocols? Have they produced IQ/OQ/PQ packages for equipment in FDA-regulated environments? What’s their documentation staff turnover? This criterion gets high weight because documentation is what qualification depends on, and documentation staff at vendors come and go.

Track record: on-time delivery (weight: medium-high). Ask references whether the equipment was delivered on time and whether the timeline changed after contract signing. Schedule slippage on the vendor’s prior projects is the strongest predictor of schedule slippage on this one.

Track record: change orders (weight: medium-high). Ask references specifically about change orders. A vendor who consistently issues change orders on similar projects will issue them again. The URS reduces this risk significantly. It doesn’t eliminate it entirely — the vendor’s behavior in ambiguous situations still matters.

Financial stability (weight: medium). An automation project typically runs 12–18 months from contract to validation. A vendor who has financial problems during that period creates risk that has nothing to do with technical capability. A D&B report, financial references from their bank or major customers, and a check on their ownership structure takes half a day and can prevent a significant problem.

Responsiveness during the quote process (weight: medium). How a vendor behaves during quoting is how they’ll behave during the build. A vendor who takes three weeks to respond to a technical question during the quote process will take three weeks to respond to a FAT punch list item. Responsiveness during quoting is a leading indicator of project management quality during the build.

Score each vendor against each criterion, weighted. The KT analysis produces a total score for each vendor. The vendor with the highest score is the recommended selection.

The recommendation goes to management with the scoring methodology and the underlying data — not just a name. When management has the KT analysis in front of them, the selection discussion becomes a conversation about the tradeoffs, not a comparison of gut instincts.

And when management overrides the recommendation anyway and selects the lowest bid, there’s a documented record of the recommendation that was made and the evidence behind it. That record matters when the project runs into trouble.

What a Facility Tour Actually Tells You

The original post described the facility tour as part of the vendor selection process. It is. But what you’re looking for during a facility tour isn’t obvious.

The shop floor tells you about production discipline. Is the floor organized? Are materials labeled and stored correctly? Are work orders visible? A shop floor that looks chaotic produces equipment built in a chaotic environment. Not always — but often enough to be a signal.

The documentation area tells you about GAMP capability. Ask to see examples of IQ/OQ/PQ packages they’ve produced for other customers. Ask where in the building the documentation function lives. Is there a dedicated documentation team or does documentation happen in the corner of an engineer’s desk? Documentation quality is often visible in the documentation environment.

The engineering team tells you about depth. Meet the engineer who will run this project — not just the sales rep or the VP of sales. The person presenting in the conference room is not always the person who will be building the machine. Ask who the project engineer is and what their background is.

The reference calls validate what you saw. After the facility tour, call the references. Ask specifically: “The person you met on the tour — was that the person who actually ran your project?” The answer tells you whether the presentation team and the execution team are the same people.

After Selection: The Contract Is Not Protection

After the KT analysis identifies the preferred vendor and management agrees, the contract gets written. The instinct is to believe the contract protects the manufacturer if something goes wrong.

It does. But contracts protect against the vendor’s failure to deliver what the contract specifies. They don’t protect against a vendor who delivers exactly what the contract specifies — which may not be what the manufacturer actually needs, if the URS was incomplete when the contract was written. Change orders are contractually valid. That’s the mechanism.

The protection against change orders isn’t legal language. It’s a complete URS before the contract is signed and an independent engineer on the manufacturer’s side reviewing every line of the contract against the URS.

Reza sits on the manufacturer’s side of the table. Not the vendor’s. The vendor wants to win the contract and build the machine. MEPSCo’s job is to make sure the machine that gets built is the machine the manufacturer actually needs — and that every requirement is traceable from the URS through the contract through FAT through IQ/OQ/PQ.

Common Questions

How many vendors should be solicited for quotes?

Three is the standard minimum for a contested procurement. Three vendors provide enough comparison to identify outliers — a quote that’s significantly below the others is often an indication of scope gaps or assumptions that favor the vendor. More than five vendors creates a process burden that rarely adds information gain proportional to the effort. Three to four is the practical range for most medical device automation procurement projects.

Should existing vendor relationships be evaluated differently?

Not scored differently, but weighted differently where relevant. If a vendor has an existing relationship with the manufacturer and there’s documented performance data from prior projects, that data feeds directly into the track record criteria — and it’s better data than references can provide. The KT analysis still runs. The relationship history replaces or supplements the reference check. If the relationship history is positive, the existing vendor will likely score well on the criteria that matter. If it isn’t, the KT analysis surfaces that clearly.

What if the vendor with the highest KT score is also the most expensive?

That’s the situation the KT analysis is designed to handle. The recommendation goes to management with the scoring and the price comparison. Management can make an informed decision about whether the capability difference justifies the price difference — or whether to negotiate with the preferred vendor before selection is final. What the KT analysis prevents is management making that decision without the capability data in front of them. Both projects where management overrode the KT recommendation had the capability data available. They chose not to weight it against price.

Can the vendor handle GAMP documentation themselves?

Some can. Most can’t — not at the level a medical device manufacturer needs, and not reliably. The documentation staff at automation vendors come and go. Validation happens six or more months after delivery. The person who understood the build may not be there when validation starts. And even when the vendor’s documentation is technically complete, it was written by someone whose interests include getting the machine accepted, not by someone whose only job is to make sure the documentation accurately reflects the manufacturer’s requirements. That’s a structural conflict. The manufacturer needs their own person writing documentation from the manufacturer’s side.

How long does the vendor selection process take?

From initial RFQ to vendor selection, typically four to eight weeks — longer for more complex equipment. The Budgetary URS development (2–4 weeks) plus quote period (2–3 weeks) plus KT scoring and management approval (1–2 weeks). Rushing the vendor selection process to save two weeks at the beginning of a 12–18 month project is one of the more reliable ways to add months to the back end.

What happens when no vendor passes the must-haves?

The vendor list gets expanded. This happens on projects involving non-standard equipment — specialized reagent handling, unusual container formats, novel processes. If the three vendors on the initial list don’t have the capability, a fourth and fifth vendor are solicited. The must-haves don’t flex. The vendor list does.