Most AI implementations in public health don’t fail because of the technology. They fail because organizations skip the most crucial step: understanding what they’re actually trying to fix.
You can’t automate what you don’t understand. And you definitely shouldn’t automate what’s already broken.
This is where process mapping comes in. Process mapping is the practice of documenting a workflow from start to finish, breaking it down into individual steps so you can see exactly what’s happening, who’s doing it, and where bottlenecks or inefficiencies exist. It’s not a bureaucratic checkbox. It’s the foundation that separates successful AI adoption from expensive failures.
At F&T Labs, we’ve seen this firsthand. When we worked with Sauk County Public Health on their environmental health workflows, we didn’t start by talking about AI. We started by mapping their processes. The result? An 83% reduction in workload on routine tasks, freeing their team to focus on work that actually required their expertise.
The Problem: AI Implementations Are Failing at Alarming Rates
Here’s the uncomfortable truth: by some estimates, 80% of AI projects fail, and 95% of generative AI pilots fail to deliver measurable business impact. These aren’t minor setbacks. These are complete failures that waste time, money, and erode trust in technology solutions.
Why? Part of the problem is investment bias. Budgets favor visible, topline functions over high-ROI back office improvements. Organizations focus on technology instead of understanding their actual problems. They jump straight to “AI solutions” without first mapping where those solutions should actually go.
And here’s what makes it worse: research shows that external partnerships see twice the success rate of internal builds, yet organizations keep trying to go it alone.
Public health departments implementing AI are racing toward solutions without taking the time to understand their own workflows. And until that changes, they’re setting themselves up to become another failure statistic.
To allow AI to augment your staff, you need to take a moment and identify where the real problems are before you can solve them. That likely involves working on automating some of the less glamorous aspects, those back office functions that are less likely to be invested in but offer the highest return. It means mapping our processes first and choosing technology second.
Let’s look at why AI implementations fail, and how process mapping prevents each of these expensive mistakes.
Truth #1: Organizations Jump Straight to “AI Solutions” Without Understanding Their Processes
A RAND Corporation report based on 65 interviews with experienced AI practitioners found that AI projects fail at twice the rate of non-AI IT projects. The report found that “AI projects often fail when they focus on the technology being employed instead of focusing on solving real problems for their intended end users.”
The research found that organizations often choose technology first and find problems to solve second. This backwards approach is a recipe for failure.
The problem isn’t that public health departments are rushing into AI adoption. The problem is that so few departments are actually implementing AI at all. And when they do start exploring it, they often focus on how to get their teams access to chatbots like ChatGPT or Claude rather than thinking about what processes could actually be augmented. They’re not thinking about the bigger picture and how AI could really help them.
The antidote? Identify the pain points your department is having. Don’t focus so much on the technological constraints. A consultant or implementation partner can help you think about that. But really identify where the pain points are in your workflows and processes. Some of which could be fixed by improving people and processes, but some of which could be fixed through automation.
Truth #2: You Can’t Automate What You Can’t Articulate
Even if you know something is broken, that doesn’t mean you understand why it’s broken or where the breakdown actually happens.
Here’s a sobering statistic: more than 60% of organizations are trying to implement automation without having clearly defined, standardized processes. No wonder it’s likely to fail. And AI is slightly different from traditional automation, but you’re still trying to automate something. The same principles apply.
It’s like trying to give someone directions to your house when you don’t actually know your own address.
In public health, this plays out as “everything takes too long” or “we’re drowning in paperwork.” These are real problems, but they’re not actionable because you can’t identify the source. Without mapping the actual process, you can’t pinpoint whether the problem is:
- Too many handoffs between departments
- Redundant data entry across multiple systems
- Approval bottlenecks that shouldn’t exist
- Manual tasks that could be eliminated entirely
When we work with health departments, we consistently find that staff know something is inefficient but can’t articulate exactly what’s happening at each step. That’s not their fault. It’s just what happens when processes evolve organically over years without documentation.
You can’t fix what you can’t see. And you can’t automate what you can’t articulate.
Truth #3: You Risk Automating Broken Processes
Here’s where things get really expensive: automating a bad process just makes it fail faster.
Gartner warns that “automation is not meant to make up for failures in systems or defer system replacement; using automation in that way simply extends the life of suboptimal legacy applications by creating savings that mask underlying inefficiencies.”
This is the “paving the cow path” problem. Just because cows have always walked that winding path through the field doesn’t mean it’s the most efficient route. It just means it’s worn into the ground. Paving it makes the cows walk faster on a still-inefficient path.
McKinsey research found that “too many organizations fail to consider how automating certain steps in a business or customer-facing process will affect upstream or downstream handoffs and connections, which can introduce new inefficiencies, capping the value delivered by automation.”
In our work with public health departments, we’ve seen teams eager to automate their environmental health complaint intake process. That sounds great until you map it out and realize the actual problem isn’t the intake. It’s that the complaint categorization system hasn’t been updated in 15 years and doesn’t match current health code. Automating the intake would just speed up routing complaints to the wrong departments.
The fix isn’t faster automation. It’s redesigning the categorization system first, then automating the improved process.
What Is Process Mapping? (And Why a PB&J Matters)
We’ve covered that you need to understand your processes. So let’s talk about one way that you can do that, and that is a practice called process mapping.
Process mapping breaks down a workflow into its individual steps so you can see what’s actually happening. Think about making a peanut butter and jelly sandwich:
- Gather your ingredients and tools
- Prepare your workspace
- Wash your hands
- Lay out two slices of bread
- Apply peanut butter to one slice
- Apply jelly to the other slice
- Press the slices together
- Slice the sandwich diagonally (or straight, we don’t judge)
- Serve it
- Clean up your workspace
PB&J Sandwich Process Map
Cross-functional workflow with swim lanes
Prep Team
Assembly Team
Quality & Service
Cleanup Team
Simple, right? But here’s what happens when you map it out: you can suddenly see inefficiencies, unnecessary steps, and opportunities for improvement. Maybe you realize you don’t need to dirty two knives. You could use one. Maybe you discover that laying out everything first actually slows you down.
In public health, most processes are invisible until you map them out. They involve coordination across departments, legacy workarounds that became “just how we do it,” and steps that made sense ten years ago but don’t anymore. Process mapping makes the invisible visible. And once you can see a process clearly, you can identify where AI can augment your team’s work and where it absolutely shouldn’t.
This is where you can really begin to identify where you can augment and automate parts of your processes. Many of your processes within a local public health department are invisible or span multiple departments. They’re really helped by mapping them out so everyone can see the full picture.
Real Results: How Process Mapping Led to Workload Reduction
Let’s talk about Sauk County Public Health.
When we began working with Sauk County Public Health, their environmental health staff were drowning in routine inquiries and paperwork. They knew they were spending too much time on administrative tasks, but they couldn’t pinpoint exactly where the bottlenecks were or which tasks were eating up their days.
We didn’t start by proposing an AI solution. We started with process mapping.
Our approach:
- Led a series of facilitated workshops with the entire environmental health team
- Brainstormed and listed all the major functions and workflows in the department
- Selected 3-5 high-pain, high-volume processes to map in detail
- Walked through every single step. No shortcuts, no “obvious” steps skipped
- For each step, asked: Does this require expertise (head) or relationships (heart)?
What we discovered: Huge portions of their daily work were administrative and repetitive, but required looking up a lot of information. We were focusing on environmental health work, and it involved things like answering the same questions over and over, looking up the food code, routing routine inquiries to the right staff member, generating standard form letters, and tracking follow-up dates. This work did require their environmental health expertise, but it was primarily looking stuff up. They needed a way to augment the lookup process and use their brains for more complex analysis and decision-making.
The outcome:
By identifying which tasks truly required their expertise versus which could be automated or streamlined, we helped them achieve an 83% reduction in workload on certain processes. That’s not 83% fewer staff. It’s 83% more time for environmental health specialists to do environmental health work.
The insight wasn’t that they needed to work harder. The insight was that they needed to work differently. And we couldn’t have gotten there without mapping their processes first.
Your Step-by-Step Process Mapping Framework
A proven methodology for identifying where AI can augment your team’s work
Brainstorm Your Process Inventory
Gather your team and list all major functions and workflows. Look for high-frustration areas, high-volume tasks, and time-consuming processes.
Prioritize & Select Processes to Map
Pick 3-5 processes based on pain level, volume, impact, and improvement potential. Start where high pain meets high volume.
Map the Current State
Bring together everyone who touches the process. Walk through every single step—don’t skip “obvious” steps. Those often hide inefficiencies.
Apply the Head/Heart Filter
For each step, ask: Does this require expertise (head) or relationships (heart)? If no to both, it’s a candidate for automation or simplification.
Redesign Before You Automate
Ask: Can this be eliminated? Simplified? Then automated? The order matters—elimination beats simplification beats automation.
Test & Measure
Pilot the redesigned process with a small team. Track time savings, quality improvements, staff satisfaction, and error rates. Then scale what works.
Step 1: Brainstorm Your Process Inventory
Start broad. Gather your team and list out all the major functions and workflows in your department. Don’t filter yet. Just capture everything.
Look for:
- High-frustration areas (“I hate doing X”)
- High-volume tasks (“We do this 50 times a week”)
- Time-consuming processes (“This always takes forever”)
- Bottlenecks (“Everything waits on this approval”)
Step 2: Prioritize & Select Processes to Map
You can’t map everything at once. Pick 3-5 processes to start with based on:
- Pain level: Which processes cause the most frustration?
- Volume: Which happen most frequently?
- Impact: Which consume disproportionate time relative to their value?
- Improvement potential: Which have the most room for optimization?
Start with processes where high pain meets high volume. Those are your quick wins.
Step 3: Map the Current State
This is where the real work happens. Bring together everyone who touches the process. Not just supervisors, but the people doing the actual work every day.
Consider using an online whiteboarding tool like Mural, Figma, or similar platforms to collaboratively document the process in real time.
Walk through every single step like you’re explaining it to someone who’s never seen it before:
- What happens first?
- Then what?
- Who does that?
- What do they need to do it?
- How long does it take?
- What happens if something goes wrong?
Don’t skip “obvious” steps. Those often hide the inefficiencies.
Use the sandwich example as your guide. If you can break down making a PB&J into 10 steps, you can break down processing environmental health complaints into its component parts.
Step 4: Apply the Head/Heart Filter
Now the magic happens. Go through each step and ask:
Does this step require my HEAD? (expertise, judgment, specialized knowledge, analysis)
Does this step require my HEART? (relationships, empathy, trust-building, cultural competence)
If the answer to both is no, if it’s routine, repetitive, or administrative, that’s a candidate for automation, elimination, or simplification.
Mark these steps clearly. They’re your opportunity zones.
Step 5: Redesign Before You Automate
Before rushing to implement AI, ask three questions about each “opportunity zone” step:
- Can this step be eliminated entirely? (Do we even need to do this?)
- Can it be simplified? (Could we do this in fewer sub-steps?)
- Can it be automated with AI? (Would a tool handle this better than a person?)
The order matters. Elimination is better than simplification, and simplification is better than automation. Automating an unnecessary step just makes you do something unnecessary faster.
Step 6: Test & Measure
Pick one redesigned process and pilot it with a small team or for a limited time period.
Track metrics:
- Time savings (How much faster is the new process?)
- Quality improvements (Are outcomes better?)
- Staff satisfaction (Do people actually prefer this?)
- Error rates (Are we catching more/fewer mistakes?)
Adjust based on real-world feedback. Then scale what works.
From Mapping to Implementation: The Need for Secure Tools
Here’s where process mapping connects to AI adoption: once you know what to automate, you need the right tools to automate it.
This is critical in public health. You can’t just plug into ChatGPT and call it a day. You need HIPAA-compliant, secure AI tools built specifically for your workflows.
That’s why we built PH360. It provides the AI capabilities you need after you’ve done the process mapping work, ensuring you’re automating the right things in a way that protects your community’s data.
But the tool comes second. The mapping comes first.
The Partnership Advantage
One more thing: you don’t have to figure this out alone.
Research from MIT shows that purchasing AI tools from specialized vendors and building partnerships succeed about 67% of the time, while internal builds succeed only 33% of the time. That’s a 2:1 advantage for partnered implementations.
McKinsey research confirms that more than two-thirds of AI implementation leaders use external partnerships to develop solutions, recognizing their internal limitations and seeing the value in bringing in external expertise.
Why does this matter? Because effective AI implementation requires both process expertise (which you have) and AI implementation expertise (which specialists have). The combination is more powerful than either alone.
Let’s Map Your Processes Together
You don’t have to figure this out alone.
At F&T Labs, we’ve developed a proven process mapping methodology specifically for public health departments. We’ve walked dozens of health departments through this exact framework, and we’ve seen what works and what doesn’t.
We bring the structure. You bring the expertise about your work. Together, we identify where secure AI tools like PH360 can augment your team’s intelligence.
Ready to start? Schedule a conversation with our team. Let’s talk about your workflows, your pain points, and how process mapping can lay the foundation for successful, strategic AI adoption.
Because the path to AI success doesn’t start with technology. It starts with understanding.
And that starts with mapping.
