The machinery

Most coaches sell you a typology. Here is the real data structure underneath mine — and where I'm honest that it isn't fully mapped yet.

The Seven Patterns are recognition language, not measurement. Underneath them sit sixty-two recurring structures of limiting belief, derived from five hundred and fifteen coded instances from real (anonymized) coaching work — not invented, and not willing to sort into seven boxes: the best possible partition of their co-occurrence network is weak (Q = 0.27), and that resistance is the finding. What looks like a type is a blend.

Beneath the structures sit nineteen bipolar axes — Control↔Trust, Speed↔Thoroughness, and seventeen more — logged live from the coaching room. A pattern is a position across several axes at once, not a category. The axes are measured; where each person sits on them is mapped only in paid, coached work — and parts of the system aren't mapped at all yet. That's stated here because the alternative is pretending otherwise.

The Periodic System

19 bipolar axes, measured from live coaching data — not designed, not imported. Every pattern below is a position across several axes at once, never a single box.

How to read this

The number on each tile is how many of 62 measured coaching patterns touch that axis — not how important it is. High numbers (Control↔Trust, Performance↔Learning) are load-bearing but generic; almost everyone has some position on them. Low numbers (Speed↔Thoroughness, Individual↔Collective) are rare, and rarity is what makes an axis distinguishing. A real pattern typically activates 3–9 axes at once — never just one.

Not shown: pole direction.Which end of each axis a pattern leans toward isn't measured in the graph (INDICATES edges are unpopulated). An embedding-similarity estimate was tried and rejected: several axes came back completely one-sided (Confidence 17–0, Visibility 16–0, Performance 21–0) — implausible for a real corpus, more consistent with a wording bias in the test phrases than a real signal. Excluded rather than shown unreliably.

Not shown: healthier versions.Per Jay Earley's model, each pole has a strained form and a healthy one (people-pleasing vs. cooperation — same lean, different grip). We track grip (defended → loosening) per instance already, but nothing currently joins it to pole direction. No data, so nothing shown.

Tested and empty:whether patterns sharing an axis-lean co-occur more in real clients than patterns with opposing leans — the periodic-table "bonding" question. Measured directly against real co-occurrence data: no difference (same-lean pairs avg. 0.130, opposite-lean pairs avg. 0.127). A real result — just not a positive one.

Not shown: which axis a trigger event activates. Promotion, 360, reorg, and open-ended career planning correlate with different axes in single-practice data — but that's a different corpus and vocabulary than this page, with cells as thin as n=14. See the Ledger of Proof report for the numbers and the caveats.

The free two-minute preview names your likely dominant pattern and always computes all twelve underneath — it surfaces ties instead of forcing one answer, and it is a starting point, never a diagnosis. The full assessment, inside coached engagements, works the same structures at four depths. The books each sit with one belief. The coaching is where the actual work happens. Everything else on this ladder exists to tell you, honestly, whether it's yours.