Skip to content

Explained

Performix, explained.

This is a four-part document. It is not a marketing site, a pitch deck, or a product tour. It is an attempt to make Performix unambiguous to the people who need to understand it precisely: early adopters considering whether the diagnostic will work in their organization, investors evaluating defensibility, employees and partners deciding whether to commit time to the project.

Performix is a protected-feedback diagnostic that finds the one condition blocking a team's performance. It is grounded in a claim about the world — that capable people don't produce performance unless alignment, motivation, and support are present — and built on a research substrate that precomputes the diagnostic from real psychometrics rather than generating it on demand from a language model.

Read in order if you have an hour. Skip if you have a specific question — each section is named so you can find it. Nothing on these pages is intended to persuade. The goal is alignment.

Concept deep-dives

Product concepts the explainer references get their own page. Start with The Three Controls for the operating picture — the three levers on performance and the one number that ranks them — then drill into any piece: the measurement model, the value math, how the list gets built, and how you ask it questions.

  1. The Three Controls

    The operating picture — and why CAMS is the spine of one lever, not the whole product.

    Attraction, Activation, Attrition: three controls on firm performance, each measured, rolling up into one executive number (Activation Index / NAV) that ranks where to focus. Plus how one universal model adapts to the nature of the work through the research library.

  2. The Triple-A Model

    What the diagnostic produces — and the dollar behind the percent.

    The three talent doors (Attraction, Activation, Attrition) read by one engine, and the value stack that turns a performance percentage into a CFO-fundable number: ELV, Net Activated %, NAV, Opportunity.

  3. Protected Feedback

    Why people tell the truth — anonymity as a primitive, not a checkbox.

    Deterministic tokenization plus a hard minimum-group-size gate, shown working on the most sensitive signal there is: from-to attrition, with a real cell suppressed below the floor rather than exposed.

  4. Leadership Quality

    One 0–100 score for whether a leader owns the conditions a team needs.

    A composite of composites — like a credit score you can open. The headline decomposes into compensation stewardship, performance-program goodness, and activation conditions, then drills to the single weakest sub-condition to fix.

  5. CAMS

    Why capable people still don't perform — and which of four conditions binds.

    Capability, Alignment, Motivation, Support are conjunctive, not additive. The diagnostic finds the single binding constraint — the lowest-scoring dimension — so you fix the bottleneck instead of the whole laundry list.

  6. Attraction

    Brand pull and the recruiting funnel — the select-in door of Triple-A.

    A 0–100 brand-consideration index paired with stage-by-stage funnel yield. Tells you whether the hiring problem is weak pull upstream or a leaky process downstream — two different budgets.

  7. The Value Stack

    How a performance percentage becomes a dollar a CFO can fund.

    ELV → Net Activated % → NAV → Opportunity. Walk the chain for Engineering ($555,660 unrealized in the demo), rank segments by Opportunity, and fund the gap that pays back first.

  8. The Insight Player

    Precomputed insight cards that play back like a playlist.

    Each card names one binding constraint, its evidence, a Min-N safety status, and a recommended action — computed ahead of time, instant on playback, never generated on demand by a language model.

  9. How Your List Is Built

    Why this ranks what you see instead of handing you a dashboard to assemble.

    The prioritization story: a gate, a shared 0–100 index, per-person personalization, and one rule a recommender doesn't have — a severe finding is never buried because you keep dismissing it. Google-ranked results, not a Yahoo directory.

  10. Ask Your Data

    Plain-language questions, answered with an admission of how much we know.

    Every answer is labeled measured (from your data), inferred (a range from priors + partial data), or an honest gap with the work plan to close it — retrieved from the precomputed index, never a number invented on the spot.

  11. Recipes

    Pick a recipe for your situation instead of configuring an analysis from a blank page.

    A recipe is a packaged analytical path — inputs, analysis, and outputs deployed as one decision. The same engine underneath every one; the recipe is how it specializes to the nature of your work, drawing what “good” means from the research library.

  12. Adaptive Narrowing

    Start at CAMS minimum; go deeper only when VOI says it's worth it.

    IRT-selected next questions, probabilistic narrowing through dimensions and subconstructs, and Monte Carlo/EVPI-EVSI gating on expensive analytic tracks — a system of learning, not a fixed survey.

Or see it applied

This document explains the instrument in general. If you came with a specific situation, the diagnostic is shown working against three of them — same engine, different entry point.