How to drive continuous improvement in manufacturing without disrupting the floor

In this episode of the Factory Ops Exchange, Chris Sass, Quality Manager at a Pennsylvania-based manufacturer, joins Lauren Dunford, CEO and Co-Founder of Guidewheel, to discuss practical approaches to change management, data-driven decision-making, and grassroots continuous improvement.

Top 5 takeaways

Best practices and key learnings

Stop making decisions based on what you saw on the floor walk

One of the most common traps in factory operations is trusting anecdotal evidence over actual data. A plant manager walks the floor, sees a line down, and assumes that is the biggest problem. But the data often tells a completely different story.

Chris described this pattern directly. His team used Guidewheel to capture downtime data down to the second, then ran a Pareto analysis to isolate the critical few issues driving the majority of lost time. The result contradicted the prevailing assumptions about which production line was the real problem.

"So many leaders these days, they trust what their eyes can see. They're doing their walk out on the floor... and then they make assumptions based on things that aren't the actual data."

Chris Sass, Quality Manager

This is not a technology problem. It is a behavior problem. Most manufacturers already collect data. They store it on a server somewhere. The breakdown happens between collection and action. Leaders either do not analyze the data, do not trust it, or get paralyzed when the data contradicts their gut.

The fix is straightforward. Define realistic KPIs. Build a reliable measurement system. Then actually take action based on what the numbers say, not what you think you saw at 7:15 this morning.

AI tools and data analytics platforms like Guidewheel can accelerate this by automatically capturing machine-level data and surfacing the insights that matter. But the cultural shift of trusting data over instinct has to come first.

Drive change so slowly they do not realize it is happening

The biggest barrier to improvement in most plants is not technical. It is the phrase "we've been doing this for 20 years." Chris's approach to overcoming this resistance is deliberately incremental. He introduces changes in such small portions, over such extended timelines, that the team experiences the results before they register the change.

This is not about being sneaky. It is about being smart. When you roll out a massive new process or system all at once, you trigger resistance. People feel disrupted. They push back. But when you introduce a small tweak, let it settle, measure the impact, and then introduce the next one, you build momentum without friction.

The outcome Chris described: an 8% OEE increase that the team celebrated as their own win. That is the goal. Not credit. Impact.

The method follows a clear pattern. Start on the shop floor. Talk to operators about what is actually broken. Take those real problems and translate them into the KPIs that leadership tracks. Then work the gap between the two, one small improvement at a time.

"There's plenty of businesses where there's a disconnect between leadership and people that are working out there on the floor... it's going to be the people down on the shop floor that know the true of what's going on."

Chris Sass, Quality Manager

This approach aligns directly with how AI automation and modern manufacturing platforms should be deployed. Not as a rip-and-replace overhaul, but as a layer that installs fast, proves value quickly, and scales from there.

Build a culture where improvement is the default, not the exception

Chris drew a sharp distinction between two models of continuous improvement. The first: a company hires a consultant, runs a week-long kaizen, trains a group of managers, and then nothing happens afterward. The second: someone on the team wakes up every morning on their drive to work thinking, "What am I going to improve today?"

The first model fails because it has no owner after the consultant leaves. The second model succeeds because it is embedded in how people work, not bolted on as a special event.

Chris outlined his own evolution as a practitioner in six stages: reporting problems to your boss, offering solutions alongside problems, having solutions ready to deploy, having solutions already in progress before telling your boss, inviting others to co-develop solutions, and finally finding things that are not even problems yet and improving them anyway.

That progression is a practical framework for any quality manager, production supervisor, or operations leader who wants to move from reactive to proactive.

The key ingredient is not a specific tool or methodology. It is repetition. Small chunks over long periods of time. Daily engagement. Inviting others in so they have ownership of the outcome.

How to put these insights into practice

These are not abstract principles. Here is how to start applying them this week.

1. Audit your decision-making inputs. Pick one recurring production decision your team made last month based on observation or gut feel. Pull the actual data for that same period. Compare. If they do not match, you have found your first opportunity to let data lead.

2. Identify one small, low-risk improvement to deploy quietly. Talk to operators about their biggest daily frustration. Find the smallest possible fix. Implement it without fanfare. Measure the result. Use that result to build credibility for the next change.

3. Translate floor problems into leadership language. Take the top three complaints from your operators and map each one to a KPI your leadership team reviews. Downtime minutes become lost throughput dollars. Changeover delays become missed shipment risk. Speak the language that unlocks resources.

4. Establish a daily improvement habit. Before your next shift, ask yourself one question: what is one thing I can make 1% better today? Do that for 30 days. Invite one person to join you in week two. By month two, you have a movement, not a program.

5. Get real-time visibility on your machines. If your team is still relying on manual logs or end-of-shift reports to understand what happened, you are making decisions on stale data. Platforms like Guidewheel can be installed in minutes per machine with no IT involvement, giving you second-by-second data that reveals the true picture of your operation. That is the foundation for everything else.

Moving from assumptions to action on the factory floor

The gap between what we think is happening on the factory floor and what is actually happening is where margin leaks, quality issues, and missed targets live. Closing that gap does not require a multi-year digital transformation or a team of consultants. It requires a commitment to trusting data, making changes incrementally, and empowering the people closest to the work to lead improvement every day.

AI in manufacturing is not about replacing people. It is about giving them better information, faster, so they can do what they already do well with more confidence and precision. That is the practical promise of data analytics and AI applied to real factory operations.

If you are ready to see what is actually happening on your machines and start turning that visibility into measurable gains, Book a Demo.

Watch and listen

Watch the webinar now:

Listen on Apple Podcasts