What Pokemon Taught Me About Workforce Analytics
I analyzed 1,025 Pokemon like they were employees. Here's what I learned.
Why Pokemon?
As I was experimenting with Claude Code over the holiday, I wanted to practice workforce analytics—the kind of analysis HR teams use to understand talent pipelines, identify skill gaps, and spot high performers. The problem? Real workforce data is sensitive. You can't exactly publish an analysis of your company's performance reviews.
So I needed a proxy dataset. Something with enough complexity to be interesting, enough structure to apply real frameworks, and enough public familiarity that anyone could follow along.
Pokemon, a personal hobby that I love, fit perfectly.
Think about it: every Pokemon has stats (skills), types (departments), abilities (behavioral traits), evolution stages (career levels), and competitive usage data (performance ratings). There's 25 years of design decisions baked into over a thousand creatures. It's basically a fictional HR database.
What started as a methodology exercise turned into genuine discoveries about game design patterns—and a few lessons about analytics pitfalls that apply far beyond Pokemon.
The Setup
I mapped Pokemon concepts to workforce equivalents:
- Base Stats → Core Competencies
- Type → Department/Function
- Evolution Stage → Career Level
- Competitive Usage → Performance Rating
- Generation → Tenure Cohort
With 1,025 Pokemon spanning 9 generations, I had a rich dataset to cluster, analyze, and stress-test. Here's what stood out.
5 Things I Found
1. Natural Roles Emerge on Their Own
I ran a clustering algorithm on Pokemon stats without telling it anything about types, names, or game lore. It found 5 distinct groups:
- Defensive Anchors (10%): High defense, low speed. The compliance and QA of the Pokemon world.
- Emerging Talent (36%): Low stats, early evolutions. Your junior associates.
- Swift Strikers (21%): Fast and balanced. Sales, agile teams.
- Technical Specialists (15%): High special attack, deep expertise. Senior individual contributors.
- Workhorse Executors (18%): High HP and attack. Operations leads.
The algorithm had no idea what a "tank" or "sweeper" was in Pokemon terms. It just found them from the numbers.
The HR insight: Most organizations define job families top-down. Leadership decides there should be "analysts" and "managers" and "specialists," then slots people into those boxes. But what if you let the data tell you what roles actually exist?
Cluster your workforce on competency assessments, skills inventories, or even project contribution patterns. You'll often find natural groupings that don't match your org chart—and that's valuable information. Maybe your "Senior Analysts" actually split into two distinct profiles: one group that's deeply technical, another that's client-facing. They have the same title but fundamentally different strengths.
This approach also reveals when your job architecture is fighting reality. If your clusters don't map to your levels, your leveling system might be measuring the wrong things. The Pokemon data found 5 clean archetypes from 6 simple stats. Your competency data is richer than that—imagine what it could surface.
2. Game Freak Designs to Stereotypes, Not Gaps
I expected to find "talent gaps"—type and role combinations that should exist but don't. Rock-type speedsters. Electric-type tanks. Interesting design spaces left unexplored.
After running the analysis, I found 11 apparent gaps. Exciting, right?
Then I applied a statistical correction for testing so many combinations at once. When you check 90 different cells, some will look significant just by chance. After accounting for that, zero gaps survived.
What did survive? Five patterns where Game Freak over-indexes:
- Rock types are almost always defensive
- Steel types are almost always defensive
- Electric types are almost always fast
- Psychic types are almost always special attackers
- Fairy types are almost always special attackers
Game Freak doesn't fill gaps. They lean into archetypes. Rock means slow and sturdy. Electric means fast. They design to the fantasy, not to balance.
The HR insight: Organizations do this constantly, and usually without realizing it. Finance hires analytical introverts. Sales hires charismatic extroverts. These patterns feel natural because they match our mental models of what those roles should be.
But mental models aren't strategy. When you cross-tabulate department against competency profile in your own workforce, you'll likely find the same over-indexing I found in Pokemon. Your sales team might be 80% one archetype when optimal performance might come from a 60/40 mix. Your engineering org might lack the communication-heavy profiles that would improve cross-functional collaboration.
The fix isn't to eliminate archetypes. Some clustering is natural and even useful. The fix is to make it visible. Run the analysis to see where you're over-indexed, and ask whether that's intentional strategy or unconscious pattern-matching. Game Freak designed Rock types to be slow because that serves the fantasy. Most organizations don't have that excuse. They're just hiring what they think they need.
3. Speed Correlates with Flexibility
Which Pokemon can do the most things? I built a versatility score combining move pool diversity, stat flexibility, and ability options.
Swift Strikers—the fast ones—came out on top. They have the broadest skill sets and dominate the list of "T-shaped talent" (specialists who can also flex into other roles).
The HR insight: This finding made me think about which roles in an organization tend to develop the most versatile employees. In Pokemon, speed-focused creatures have broader move pools—they're designed to adapt, pivot, and respond quickly. The same pattern likely exists in your workforce.
Roles that require rapid context-switching like product management, consulting, operations, even analytics - tend to either attract or develop employees with broader skill sets. They have to learn a wide variety of skills to successfully do their jobs. Meanwhile, deep specialist roles (your "Technical Specialists" and "Defensive Anchors") build depth at the cost of breadth. Neither is wrong, but they're very different development paths.
If you're designing rotation programs or identifying future leaders, look at where your most versatile employees came from. My guess: it's not your most specialized functions. It's the roles that forced them to move fast and touch everything. The Swift Strikers of your org chart.
4. High Performers Have Depth AND Breadth
Using competitive usage data from Smogon (where uber-competitive Pokemon players track what's actually good), I built a talent grid mapping performance against flexibility.
The top performers weren't pure specialists or pure generalists. They were both—deep expertise in their role, plus enough versatility to adapt.
Dragonite, Zapdos, Clefable, Tyranitar. All of them score high on role fit and high on cross-functional capability.
The HR insight: The "T-shaped professional" is a common framework used in the workforce. Organizations today hire for and measure depth (technical assessments, certifications, years of experience) or breadth (number of projects, cross-functional exposure) but rarely both together.
The Pokemon data made this concrete. When I plotted performance against flexibility, the star performers weren't clustered at the extremes. They sat in the quadrant that required both: high role fit (they're excellent at their specific job) plus high versatility (they can adapt when conditions change). Dragonite isn't just powerful—it's powerful and flexible. That combination is rare and valuable.
In talent reviews, this means asking two questions instead of one. Not just "how good are they at their current role?" but also "how well could they adapt if that role changed?" The pure specialist who can't flex is a single point of failure. The pure generalist who hasn't mastered anything is a permanent floater. The T-shaped employee who has both? That's your bench strength.
Build your 9-box grids to capture this. Performance on one axis, flexibility on the other. Your high-potential population might look different than you expected.
5. My Analysis Was Wrong (And How I Fixed It)
Here's where I have to be honest.
My first network analysis identified Magikarp as the most "central" Pokemon, the one that bridges the most connections across the game world. This seemed like a fascinating discovery. The weakest, most mocked Pokemon is secretly the most important connector?
It was wrong.
Magikarp appeared central because it shows up everywhere. It's in almost 200 locations, intentionally designed as the most common Pokemon in the games. My "centrality" metric was really just measuring "how often does this thing appear?"
Once I controlled for location count, the actual interesting connector emerged: Audino. It appears in a moderate number of places but uniquely bridges two otherwise disconnected game regions.
The HR insight: This is maybe the most important lesson, because it's the easiest mistake to make. Network analysis has become aspirational in people analytics, mapping who collaborates with whom, who's central to information flow, who bridges different teams.
But raw network metrics are dangerously easy to misinterpret. In my analysis, "centrality" was really just "shows up a lot." In your organization, the employee with highest betweenness centrality might not be your most important connector. They might just be your longest-tenured employee who's touched the most projects over time. Or they might sit on a lot of email threads without adding value. The metric can't tell the difference.
Before you identify "key connectors" or "flight risks" based on network position, ask what's confounding the measurement. Tenure? Role scope? Meeting culture? I had to residualize my centrality scores against location count to find the actually interesting connectors. You might need to control for tenure, level, or function to find yours.
Wrapping Up
What surprised me most about this project wasn't any single finding, it was how well the frameworks actually transferred.
Workforce analytics concepts like competency clustering, talent grids, and network analysis weren't designed for fictional creatures with six stats and elemental types. But they worked. They surfaced patterns in 25 years of game design that I haven't seen articulated elsewhere: the way Game Freak leans into archetypes rather than filling gaps, the connection between speed-focused design and versatility, the T-shaped profile of competitively successful Pokemon.
These aren't just cute parallels. They're the same dynamics that play out in real organizations. Companies tend to hire to stereotype. Fast-paced roles develop flexible employees. High performers combine depth with adaptability. Network metrics get confounded by tenure. The Pokemon data just let me show it in a way that's easy to explore and verify.
A year ago, this project would have stayed on my "someday" list. The data pipeline, the clustering, the visualizations. It felt like too much to take on. But the tools have changed, and so has what's possible for someone willing to learn by building.
If you work in people analytics, I hope something here sparked an idea. If you're a Pokemon fan, I hope you enjoyed seeing the games through a different lens. And if you're somewhere in between, welcome to the club.
Built with Python, pandas, scikit-learn, and Claude Code. Data from PokeAPI and Smogon.