Establishing Action Through Data Analysis
Prioritizing Data from Hundreds of Conversations
Note: This methodology applies to analyzing feedback for any organizational change initiative, not just quality programs.
Starter Questions
How do you analyze input from across an entire organization?
Should you use AI to help analyze the data?
How do you develop the right plan for your company?
What if the data tells you your idea won’t work?
Analyzing Conversational Feedback
When gathering feedback at scale, specific themes will have revealed themselves. Potential examples include:
“I was never trained on the program.”
“We need more templates to guide us.”
“The checklists are too long.”
“I don’t know who to call for help.”
“We don’t have time for this.”
“The process doesn’t match how we actually work.”
For each comment, develop a solution.
Comment: ”I was never trained on the program.”
Solution: Provide regular training.
Comment: “We need more templates to guide us.”
Solution: Create additional templates.
Comment: “The checklists are too long.”
Solution: Simplify current documentation.
Comment: “I don’t know who to call for help.”
Solution: Clarify support structure.
Comment: “We don’t have time for this.”
Solution: Identify and remove redundant steps.
Comment: “The process doesn’t match actual work.”
Solution: Redesign process with field input.
Continue this for all conversations. In my situation, roughly 600 comments from 300 interviews yielded 24 distinct solutions.
Quantifying the Trends
Sum up the number of times each solution appears across all interviews. Create a simple bar chart showing frequency.
Simplify current program: ████████████████████ 87 mentions
Provide regular training: ███████████████ 64 mentions
Create additional templates: ████████ 31 mentions
Clarify support structure: ██████ 23 mentions
Remove redundant steps: █████ 19 mentionsThese solutions become the actions for the business case. However, you’ll need to identify which solutions to focus on first. You could choose the top five, or top ten. It depends on your capacity, your team’s capacity, and your organization’s capacity for change.
Addressing Executive Concerns Simultaneously
The solutions (actions) must address both:
The most frequent concerns from user interviews
The specific concerns executives raised
Example: One executive stated we can’t add any new tasks without removing two other steps (a constraint on complexity). Looking at our sample data above:
“Simplify current program” was mentioned 87 times
“Remove redundant steps” was mentioned 19 times
By simplifying the program and removing redundancy, we address both the superintendent feedback and the executive constraint. This makes approval far more likely.
Another example: An executive was concerned about training effectiveness. The data shows:
“Provide regular training” mentioned 64 times
“Simplify current program” mentioned 87 times
We can pose the following argument: “Simplifying the program makes training easier because there’s less to learn. Combined with regular training sessions, we solve both the competency gap and the complexity problem.”
When your actions solve problems at multiple levels simultaneously, approval becomes inevitable.
On the Use of AI: Why You Shouldn’t
Analyzing subjective feedback from conversations is difficult. There are more robust statistical methods than what I used, and AI tools could theoretically help. Why didn’t I just record these conversations in Microsoft Teams and load them into an AI, asking for a summary? That could have saved enormous time. Having said that, I encourage you not to use AI for this analysis (at least at first), except to validate your findings after you’ve done the thinking. A few things to consider:
You can’t defend AI-generated conclusions. When you present your business case, executives will ask hard questions: “Why should we approve this? Why these actions and not others? How do you know this will work?” If you used AI to analyze the data, your answer is: “Because the AI said so.” That’s not compelling. You’ll get demolished by follow-up questions.
You lose the context. AI can identify themes, but it can’t understand the emotion, frustration, or resignation in someone’s voice when they say: “We don’t have time for this.” That context matters when you’re deciding which problems to solve first.
You can’t articulate the nuance. When an executive says “But I’m very concerned about X,” and you need to respond with “I understand your concern. I talked with 150 people and that’s not what I discovered. Tell me more about your perspective,” you can only do that if you did the thinking, not the AI.
Critical thinking is the point. Building a business function isn’t about replicating data. It’s about demonstrating your critical thinking. Executives are betting on your judgment, not an AI’s analysis.
By analyzing the feedback yourself, you prepare for the challenging conversations that inevitably arise. You become the best-prepared person to lead the effort because you internalized all the context during analysis. This is how you become the right person with the right plan: Using data from your organization to develop actions for your business plan. You’ll have the context to back up your intentions and articulate them under pressure.
You can use AI to check your analysis for themes you missed, help format your presentation, validate your conclusions. But don’t outsource the thinking.
Prioritizing Actions
You’ve identified 10-24 potential solutions. You can’t do everything at once. How do you prioritize? Use these criteria:
Impact: Which actions solve the biggest problems? (Look at frequency in your data).
Executive Alignment: Which actions directly address executive concerns?
Dependencies: Which actions must happen before others can succeed?
Stage 3 thinking: Which actions raise competency vs. which create dependency?
To expand on the sample data above:
Simplify current program (87 mentions) → HIGH impact, executive concern, enables training
Provide regular training (64 mentions) → HIGH impact, depends on simplified program
Create additional templates (31 mentions) → MEDIUM impact, quick win
Clarify support structure (23 mentions) → MEDIUM impact, quick win
Remove redundant steps (19 mentions) → Covered by #1 (simplification)
Priority 1: Simplify the program (enables everything else, addresses most concerns).
Priority 2: Provide training on simplified program (raises competency, not dependency).
Priority 3: Clarify support structure (quick win, prevents confusion during rollout).
Priority 4: Create templates as needed based on training feedback (iterative, not upfront).
This gives you a logical sequence that builds momentum while addressing root causes.
The Stage 2 vs. Stage 3 Lens
While analyzing solutions, ask for each one: “Does this raise competency or create dependency?”
Raises Competency (Stage 3):
Simplify processes so people can use them
Train people to do the work themselves
Provide tools and templates as aids
Create clear standards and expectations
Hold people accountable for results
Creates Dependency (Stage 2):
Hire staff to do work for project teams
Add support roles that take over tasks
Create processes so complex only SMEs can navigate them
Position the function as a service provided to projects instead of by projects
Every action should pass this test. If the data is full of dependency-creating solutions, you’re building expensive overhead that won’t scale.
Showing Your Work
When presenting the business case, it’s essential to be transparent about how you came to your conclusions.
Summary data: Bar charts showing solution frequency
Sample quotes: Anonymous representative comments (with role/tenure for context)
Demographic coverage: Table showing you interviewed across regions, roles, tenure
Analysis methodology: Brief explanation of how you coded and analyzed feedback
The act of preparing these materials forces you to ensure your analysis is actually rigorous, not just confirmation bias. Most executives won't want to see this detail, though asking ahead of time about their preferred level of depth can be helpful. Some will ask “How do you know this is right?” and you’ll need to show your work, or at least be able to briefly and effectively articulate it.
Data is the Truth
After analyzing 300+ comments and identifying 10-20 solutions, ask yourself:
“If we do these things, will project teams become more competent, or more dependent?”
If the honest answer is “more dependent,” you’re building a business case for expensive overhead. Either redesign your actions or admit you’re choosing the Stage 2 knowingly. If it’s “more competent,” you’re building a business case for sustainable capability. That’s what scales. The data tells you the truth. Your job is to listen and respond honestly, even if it contradicts what you wanted to build initially. Even if it creates tension at the leadership level. How the company navigates that tension will help the functional leader understand if the company will accept the direction the data is instructing them to go.