Buyer resource
How to Select an Industrial Automation Integrator
Select an automation integrator by verifying four things: real engineering evidence on projects like yours, a documented process from specification through acceptance testing, contractual clarity on who owns code and drawings, and a support model you can live with after commissioning. Price matters only after those four are satisfied.
Published · Last reviewed by Geek Engineering senior engineering review.
Why integrator selection goes wrong
Most bad automation outcomes are selection failures, not engineering failures. The winning bid was chosen on price, the scope was vague enough that everyone imagined a different machine, and by the time the gaps surfaced, switching vendors cost more than pushing forward.
The method below is designed to surface those gaps before the purchase order, when they are still cheap to fix.
Step 1: Verify capability with evidence, not adjectives
Every integrator's website says roughly the same thing. Evidence separates them:
- Ask for sanitized examples of projects at similar complexity — drawings, screen layouts, code structure — not just logos and testimonials.
- Ask who, specifically, would engineer your project and what they have personally delivered.
- Ask which platforms they work in weekly, versus occasionally. Platform depth shows up in commissioning speed.
- Ask what they would subcontract. Subcontracting is normal; not disclosing it is not.
Step 2: Probe the engineering process
A capable integrator can describe, unprompted, how a project moves from your requirements to an accepted system. Listen for these artifacts:
- A functional specification you approve before programming starts
- Design reviews at defined points, with your team in the room
- A written factory acceptance test (FAT) tied to the specification
- A commissioning and site acceptance plan with defined roles
- As-built documentation as a listed deliverable, not a favor
Step 3: Settle ownership and access before signing
If the answer to the last question is no, you are not buying a system; you are subscribing to a vendor. That can be acceptable, but it should be a conscious decision priced accordingly.
- Who owns the PLC and HMI source code — and is it delivered unlocked?
- Who owns drawings, and in what format (native CAD versus PDF)?
- Are there runtime licenses, and what do they cost to maintain?
- Can another qualified engineer support the system without the original vendor?
Step 4: Evaluate the support model
Ask what happens in the first ninety days after acceptance: who answers, through what channel, at what cost. Ask how remote access is handled and secured. Ask what a small change (one new recipe, one screen edit) costs and how quickly it typically happens. Vague answers here predict your future accurately.
Step 5: Compare quotes on scope, not totals
When quotes differ significantly, the cheaper one has usually excluded something: commissioning days, documentation, training, spares, or contingency for the unknowns in your legacy equipment. Build a comparison table by scope line, force the exclusions into the open, and re-compare. The 'expensive' quote is frequently the honest one.
What to ask, and how to read the answer
| Question to ask | A good answer | An answer that should worry you |
|---|---|---|
| Who specifically will do the engineering? | Names the engineers, describes their comparable work, and offers a technical conversation with them | “Our team”, with no one available to talk before the purchase order |
| What do you need from us to quote accurately? | A specific list — drawings, backups, constraints, downtime windows — and a refusal to fix-price without them | A firm price offered immediately, from a one-page description |
| Who owns the code and drawings afterwards? | You do, in native editable form, stated in the contract | Vague reassurance, or source held “for support reasons” |
| What is subcontracted? | Named up front with who is accountable for the interface | “We handle everything in-house” on a scope no single firm covers |
| How are changes handled? | A written change process with cost and schedule impact shown before you decide | “We are flexible” — which means absorbed silently, then argued later |
| What happens if the acceptance test fails? | A defined punch-list process with severities, owners and a retest | Surprise that you asked |
| What does support look like after commissioning? | A named model with response expectations and what it excludes | “Call us any time”, with nothing written down |
Key takeaways
- Choose on verified capability, process, ownership, and support — then price.
- A written functional specification and FAT separate professionals from optimists.
- Unlocked code and native drawings determine your independence for the next decade.
- Quote differences are usually scope differences in disguise.
Related questions
Should we always choose a local integrator?
Local presence matters most for hardware-heavy work with frequent site visits and for emergency response expectations. For controls, software, and data work, much of the engineering happens remotely regardless of address. Weigh locality as one factor — travel cost and response time — rather than a filter that excludes better-fit teams.
How many integrators should we ask to quote?
Two or three serious candidates is usually right. More than that spreads your evaluation attention too thin to do the evidence-checking that actually predicts outcomes, and good integrators invest less in proposals they are unlikely to win.
What is a red flag in an automation proposal?
The biggest ones: no functional specification step, acceptance criteria left undefined, code ownership unaddressed, commissioning days suspiciously low for the scope, and an inability to name who will do the engineering. Any one of these predicts change orders and disputes.
Does a certification or partner logo guarantee quality?
No. Platform partnerships and industry memberships indicate investment and access to vendor support, which has real value, but they do not substitute for evidence of similar delivered work and a sound process. Evaluate the engineering, not the badge wall.
Continue your project research
Prefer to talk this through?
Send a short brief — objective, equipment, timeline — and a senior engineer will respond with questions and an honest fit assessment.
Get a Quote