Before You Let a Vendor Tell You What You Need, Read This
- Marcus Blair
- May 8
- 4 min read
When most people hear "digital transformation," they immediately think of robots on the production line or pricey AI systems designed for companies ten times their size. But for small and mid-sized manufacturers, transformation looks very different. It's not about bleeding-edge technology. It's about getting the most from what you already have—and turning scattered data into actionable insight.
I recently had the opportunity to present to a group of manufacturing leaders about what this transformation can actually look like for companies doing $5M to $50M a year. This post is a recap of the major takeaways and key tools that resonated with the audience—folks juggling multiple facilities, legacy systems, and yes, plenty of spreadsheets.
Main Message: Don’t Start with AI. Start with Clean Data.
Every big goal—whether it’s predictive analytics or AI-enabled inspections—depends on one thing: clean, reliable, structured data.
"A perfect engine can’t run on bad fuel."
The first half of the presentation focused on getting to a trustworthy dataset that works across departments. It’s not glamorous, but it pays off immediately. If you walk into your plant and ask “what happened on Line 1 last night?” and the answer is “I’ll go find Jimmy,” you don’t need AI. You need a system.
The Analytics Maturity Curve: Know Where You Stand
The Analytics Maturity Curve has been around for a while, at least 10 years or more. It is still a useful paradigm, even if its evolved over time.
Foundational – Where is the data?
Descriptive – What happened?
Diagnostic – Why did it happen?
Predictive – What is likely to happen?
Prescriptive/Cognitive – What should we do next?
Every company doesn't need to aim for level 5; many times, its not feasible or practical. Most small manufacturers should focus on reaching level 3. If you can diagnose issues without pulling in three departments and chasing down paper, you’re ahead of the game. From there, you can decide what advanced technology would work best for you and how your business functions.
Step 1: Make the Most of What You Already Have
You likely already own everything you need to get started.
Most attendees used Microsoft 365. That means:
Power BI for real-time dashboards
Power Automate for workflow automation
SharePoint, Fabric or Dataverse for shared data storage
I shared a case study from a boat manufacturer. We digitized inspection forms using Microsoft Forms and SharePoint, created dashboards in Power BI, and used Power Automate to route repairs and store birth certificates for warranty claims. Power Apps provided integration for the entire thing.
Total implementation time: 8 weeks. New software purchases: zero.
Step 2: Choose the Right Strategic Next Step
After creating a clean, centralized a single source of truth and automating away the spreadsheets, manufacturers face a fork in the road:
Does Communication Need to Improve? → Go for cross-company automation
Do Decisions Need to Happen Faster → Explore predictive analytics
For example, a client of ours with field sales reps had no way to check real-time inventory. Sales kept overpromising, ops kept underdelivering, and accounting had no visibility.
We automated their sales-to-shipping pipeline using the tools they already owned. The result? Sales could instantly see what was in stock and when the rest would be built, which meant they could confidently commit to the client.
Step 3: Build a Competitive Edge with Local AI
This is where AI can shine, but only after the foundation is laid. Without a single source of truth and agreed upon processes and definitions, the project will fail.
Locally hosted AI offers real benefits over public tools like ChatGPT:
No cloud subscription fees
Better data security
Faster performance
Custom models trained on your internal data
We use a private AI in-house. It reviews sales transcripts, summarizes meetings, and even gives feedback on proposals. Because it’s trained on our business, it understands our context and goals better than any generic tool.
A great starter project? Build a chatbot trained on your HR manual. It’s low-risk, low-effort, and builds comfort with how LLMs work.
Visual Recap: Your First Stack Might Look Like This
With these, you’ve already started automating, visualizing, and creating better operational decisions—without installing a new ERP or hiring a dev team.
What the Room Responded To
The idea that “AI isn’t the first step” struck a chord.
Several attendees shared that Power BI alone could save hours every week.
Many liked the breakdown of where bottlenecks form: not in effort, but in handoffs and data visibility.
There was relief that transformation doesn’t have to be expensive or disruptive.
People want guidance, not a 60-slide pitch deck from a vendor who doesn’t understand them.
Final Thoughts: Practical Over Perfect
The most important thing I told the room is this:
“Start with what you have. Clean it up. Use it well. Then, and only then, layer in automation, AI, or prediction.”
Digital transformation is a journey, not a shopping list. The tools are ready. Your people are probably more open to change than you think. You just need the right first step.
If you're a small manufacturer sitting on mountains of paper or siloed spreadsheets, you don't need to buy new tech. You need a partner who will help you put your existing tech to work.
And that’s exactly what we do.
Want a copy of the slides or access to the full inspection case study?
