Choosing an industrial IoT vendor is one of the most consequential technology decisions a manufacturing organization will make. Get it right, and you have a foundation for operational improvement for years. Get it wrong, and you're stuck with a system that doesn't fit, doesn't scale, or doesn't deliver the promised value.

Having been on both sides of vendor evaluations, here's a practical framework for making this decision well.

Start With Your Requirements, Not Their Capabilities

The biggest mistake I see is letting vendor demos drive the evaluation. Instead of understanding what they can do, start by defining what you need:

  • What specific problems are you trying to solve? Reduced downtime? Better quality? Compliance documentation?
  • What assets need monitoring? Start with the most critical, not everything.
  • What data do you actually need? More isn't better. Define the decisions you'll make with this data.
  • What integrations are required? CMMS, ERP, historians, SCADA systems.
  • What are your constraints? Network limitations, security requirements, regulatory compliance.

Write this down before talking to any vendors. Share it with them. Evaluate their responses against your requirements, not against each other's feature lists.

Technical Evaluation: Beyond the Demo

Vendor demos are carefully choreographed to show the product at its best. Here's how to get past the polish:

Architecture Questions

  • Where does data processing happen? Edge, cloud, or hybrid?
  • How does the system handle network outages?
  • What's the data retention policy and cost structure?
  • How are firmware updates and security patches deployed?
  • What happens if we want to leave? Data export and portability.

Integration Questions

  • What APIs are available? REST, MQTT, OPC UA?
  • Have you integrated with [our specific systems] before?
  • Who does the integration work? Vendor, partner, or us?
  • What documentation and support is available?

Sensor and Hardware Questions

  • What sensor types do you support?
  • Can we use third-party sensors?
  • What's the expected hardware lifespan?
  • How is hardware replaced when it fails?

Red Flags to Watch For

Years of vendor interactions have taught me what to avoid:

  • Vague answers about data ownership: If they can't clearly explain who owns your data and what they do with it, walk away.
  • No reference customers in your industry: Generic case studies aren't enough. Talk to real customers with similar use cases.
  • Pricing that requires a spreadsheet to understand: Per-sensor, per-data-point, per-user pricing models get expensive fast.
  • Proprietary lock-in everywhere: Proprietary sensors, proprietary protocols, proprietary data formats. Each one is a dependency.
  • The "roadmap" answer: If critical functionality is "coming soon" or "on the roadmap," evaluate what they have today.
  • Implementation partners do everything: If the vendor doesn't have their own implementation capability, you're adding another party and another point of failure.

The Reference Call Framework

Reference customers are your best source of truth. When you get access to them, ask:

  • How long did implementation actually take compared to the estimate?
  • What surprised you after go-live?
  • What would you do differently?
  • How responsive is support when things break?
  • What's the total cost of ownership compared to the initial quote?
  • Would you choose them again?

Listen for hesitation. Happy customers are enthusiastic. Unhappy customers who've been asked to be references give careful, qualified answers.

Proof of Concept: Doing It Right

A pilot or proof of concept should prove value, not just prove the technology works. Structure it with:

  • Clear success criteria: What specific outcomes will demonstrate value?
  • Realistic scope: 5-10 assets, not the whole facility.
  • Representative conditions: Include challenging network areas, older equipment, real operators.
  • Time for stabilization: At least 3 months to get past installation noise.
  • Exit clause: What happens if the pilot doesn't work?

Total Cost of Ownership

The purchase price is just the beginning. Build a complete picture:

  • Hardware: Sensors, gateways, edge devices
  • Software: Licenses, subscriptions, data storage
  • Implementation: Installation, configuration, integration
  • Training: Initial and ongoing
  • Support: Maintenance contracts, premium support tiers
  • Internal resources: Your team's time to manage and optimize
  • Scaling costs: What happens when you expand to more assets?

Ask vendors to provide a 3-year TCO estimate with all these components. Compare apples to apples.

The Decision Framework

When you have evaluated multiple vendors, structure your decision around:

  1. Fit: How well does the solution match your specific requirements?
  2. Capability: Can they deliver what you need today and grow with you?
  3. Risk: What's the downside if this doesn't work out?
  4. Value: What's the realistic ROI given honest implementation timelines?
  5. Relationship: Can you work with these people for years?

No vendor is perfect. The goal is finding the best fit for your specific situation, not the objectively "best" product.

Final Advice

Take your time. This decision will affect your operations for years. Rushing to meet a budget deadline or executive pressure often leads to choices you'll regret.

Involve the right people. Maintenance teams, IT, operations, and leadership all have perspectives that matter. A solution that looks great to IT but doesn't work for the maintenance team will fail.

Trust your instincts. If something feels off about a vendor or their answers, it probably is. The best technology from a company you don't trust is still a bad choice.

And remember: the best time to negotiate contract terms, exit clauses, and SLAs is before you sign, not after.