Three doors, one engine — and the dollar behind the percent.

Water flows in, water sits in the tub, water drains out. The Triple-A Model is three questions, one for each part of the tub — Attraction (are the right people flowing in?), Activation (are the people already in the tub actually producing what they could?), and Attrition (who's draining out, and did we want to lose them?). Most companies put a meter on one tap and ignore the rest.

Every dollar you spend on talent flows through one of these three doors. Pour money into recruiting while your best engineers quietly leak out the back, and you're bailing a tub with the drain open. Obsess over retention while the people you keep are only half-producing, and you're paying full price for partial output. The model forces the uncomfortable question most dashboards dodge: which door is actually losing you the most money right now — so you fund the leak that matters instead of the one that's loudest.

Door 1 — Attraction asks whether the right people are flowing in. The 72 is a brand-consideration index: a 0–100 score for how attractive this team looks to the people it's trying to hire. 72 reads healthy. Beneath it, the funnel — 400 applied, 17 hired, a strong 75% offer-accept rate. The front tap is working.

Door 2 — Activation asks whether the people already here are producing their potential. 70% have every condition in place to perform; the other 30% is the gap. The money math: this segment's lifetime value (ELV ≈ $1.85M) multiplied by 70% gives NAV ≈ $1.30M actually captured — leaving ≈ $556K of Opportunity on the table. Across the demo portfolio that untapped Opportunity totals ≈ $1.0M.

Door 3 — Attrition asks who is leaving and whether you regret it. Nine key-talent exits; six of those nine (67%) were regretted — people you didn't want to lose. The from-to read shows where they went: six left for Competitor-A, citing pay and growth. Not just a number — a named competitor and a reason you can act on.

The point isn't three separate dashboards. It's one measurement engine answering three questions for the same team — the brand index, the activation share, and the regretted-exit rate all trace back to the same psychometric measurements and the same from-to survey signal, not three tools bolted together.

A capability score of “72 out of 100” doesn't move a budget. A line that says “$556K of value sitting unrealized in Engineering” does. The value stack is the bridge — it converts a soft performance percentage into the kind of number a CFO can put next to Revenue and Headcount and act on.

Walk the chain for Engineering. ELV = $1,852,200: each engineer is worth about $1.85M in lifetime profit, rolling up from a Human Capital ROI of 3.1 (profit per $1 of pay), a fully-loaded annual cost of $210,000, and a 4.2-year average tenure-at-exit — together a value multiple of roughly 8.8× annual cost. Net Activated % = 70%, read from the CAMS diagnostic. NAV = 0.70 × $1,852,200 = $1,296,540, the value truly captured. Opportunity = $1,852,200 − $1,296,540 = $555,660 — the dollars left on the table, and the action signal.

Compare segments and the priority falls out. Sales has a lower ELV ($686,000) but only 40% activated, leaving $411,600. Customer Support ($49,140 ELV, 30% activated) leaves just $34,398. Sorted by Opportunity, Engineering is where the first dollar of effort goes. Across all three, the portfolio totals ELV $2,587,340 → NAV $1,585,682 → Opportunity $1,001,658 — about $1M of realizable upside, ranked.

The activation rate isn't a vibe or a generated guess — it comes from the psychometric CAMS diagnostic, and every dollar figure carries its source and its sample size: the 70% rests on n=240 heads, the 4.2-year tenure on 38 leavers. ELV follows a published framework (People Analytics for Dummies, Ch. 6–7), not an ad-hoc formula.

The math is real measurement, not a number this page invented — the figure on screen is the one Performix's measurement engine produced. Every headline is a 0–100 index you can drill from number → sub-measures → underlying data and where it came from, so a skeptic can challenge the 70 rather than be told to trust it.

The diagnostic engine and the index-with-drill-down surfaces are built and rendering on seeded demo data today. The Triple-A measures — segment-level ELV, Net Activated % over CAMS, the from-to attrition read, and the attraction funnel — are surfaced in the product; as your data is connected, the surface lights up on real data, no UI rebuild. That demo-to-live path is the same one the diagnostic itself already walked.