Token ROI Overview

Live attribution across 6 processes · 487 active profiles · updated continuously

New: Return on AI SpendEvery dollar of AI spend, tied to work that shipped — see which people, teams, and models return the most above cost

+129%

7.1x

Organizational Token ROI

North Star · vs. W1 baseline 3.1x

+18%

$1.49M

Annual token spend

Across 6 integrated processes

+64%

$12.9M

AI-attributable value

Marginal value, trailing 30 days

95%

487

Users with ROI profiles

of 512 AI-active employees

Organizational Token ROI

Marginal value per $1 of token spend · 26-week deployment

Spend vs. attributed value

By business process · trailing 30 days ($K)

Token spend Attributed value

Token ROI profiles

Per-user efficiency · click a row for the full profile

User	Team					Percentile
Dana Okafor Staff Software Engineer	Search Platform	214K	$45K	$5.2M	13.3x	P99
Priya Raman Principal Engineer	Payments	188K	$39K	$4.0M	11.4x	P97
Marcus Feld Senior Counsel	Legal & Compliance	96K	$20K	$1.1M	8.9x	P92
Ivy Chen Data Scientist	Growth Analytics	142K	$30K	$1.5M	8.6x	P90
Tom Belunderscore Support Lead	Customer Success	168K	$35K	$940K	5.3x	P74
Lena Sørensen Product Manager	Core Product	88K	$18K	$320K	2.9x	P58
Raj Malhotra Software Engineer	Internal Tools	74K	$15K	$96K	0.6x	P22
Chris Vance Marketing Associate	Demand Gen	52K	$11K	$61K	0.4x	P11

Efficiency dimensions

Top decile vs. median across the six UEP dimensions

Top decile Median

Weekly resource signals

Automated reallocation intelligence from the OLS

Reallocation opportunity

$280K/yr recoverable

Engineering: 3 members in the bottom ROI quartile despite top-20% token spend.

Concentration risk

Succession risk

85% of AI-attributable value in Search Platform is generated by 2 individuals.

Budget expansion case

+$612K unlockable

4 Data Science users are token-constrained by team cap; marginal ROI averages 34x.

Extracted playbooks

Patterns propagated from top-decile users

Front-loaded context framing

Engineering

2.4x

avg. lift

Spend 30–40% of context on structured problem framing before requesting output. Reduces iteration cycles.

42% adopted

Staged decomposition with checkpoints

Engineering

1.9x

avg. lift

Break complex tasks into staged prompts with intermediate validation. Each stage's output becomes the next stage's context.

37% adopted

Escalation triage template

Support

2.1x

avg. lift

Operationalize a top agent's triage prompt into a team-wide template for faster first-contact resolution.

54% adopted

Contract risk extraction pass

Legal

3.0x

avg. lift

Two-pass review: structured risk-flag extraction, then targeted deep-dive. Compresses review cycles.

28% adopted