v1v2
role
trust
Loopmason·Northwind Technologies/Whole organization

Token Efficiency

Where the tokens go, where they're wasted, and exactly how to save them

Tokens used (MTD)
31.4B tokens
of 37.5B tokens allowance
Forecast 32.4B tokens
Recoverable now
5.5B tokens
18% of tokens recoverable
Act now · policy
2.1B tokens
1 config change — model right-sizing
Enforceable here
Coach to recover
3.5B tokens
4 behavioral levers — coached in Craft
Recovered per-engineer in Craft

Act now — policy levers

Admin · one config change · 2.1B tokens/mo recoverable — set it here and the system complies instantly

Set the policy
  • Route trivial tasks off OpusModel right-sizingPOLICY

    340 short, high-acceptance sessions ran on Opus where Haiku/Sonnet would suffice — Opus bills 5× the quota for identical output.

    340 sessionsEffort: lowConfidence: high
    2.1B tokens/mo

Coach to recover — behavioral levers

Per-engineer · via Craft · 3.5B tokens/mo recoverable — quantified here, but there's no switch to flip: each is fixed by coaching how people work, in Craft

Coach in Craft
  • Trim bloated contextContext bloatCOACHING

    120 sessions exceeded 120k tokens of context without a /clear — most of it unused but re-billed every turn.

    120 sessionsEffort: lowConfidence: high
  • Raise prompt-cache hit rateCache efficiencyCOACHING

    Cache hit is 61% vs a 75% target: restarting sessions and re-pasting files throws away the warm cache, so context gets re-billed at full rate.

    org-wideEffort: mediumConfidence: medium
  • Reduce abandoned sessionsAbandoned workCOACHING

    210 sessions burned tokens but ended with no accepted diff — usually unscoped work that should have started in plan mode.

    210 sessionsEffort: mediumConfidence: high
  • Cut repeated failed tool-call loopsRetry loopsCOACHING

    Sessions retried the same failing command 5+ times before changing approach, re-billing the whole context each loop.

    90 sessionsEffort: lowConfidence: medium

Realized savings — audited

Projected vs what actually landed after you acted. Realized deltas are MEASURED (before/after token series keyed to each acted move); projections are PROXY counterfactuals. 4 moves still inside their verification window.

1.6B tokens realizedof 2.2B tokens projected projected realized
  • Sonnet downshift · Checkout routing policyRealized
    Jun 28
    Projected
    725M tokensRealized
    778M tokens(107%)

    Tracking 107% of projection — verification window closes Jul 12

  • Thrash coaching · GrowthRealized
    Jun 21
    Projected
    230M tokensRealized
    185M tokens(80%)

    Verified Jul 3

  • Cache config fix · SRE CorePartial
    Jun 14
    Projected
    525M tokensRealized
    425M tokens(81%)

    Variance: 3 engineers switched models manually before rollout

  • Onboarding nudge · 11 idle seatsNo effect
    Jun 09
    Projected
    578M tokensRealized
    0(0%)

    NO EFFECT: 9 of 11 seats still idle — re-minted with escalation

  • Context→CLAUDE.md playbook · DevExRealized
    Jun 02
    Projected
    160M tokensRealized
    173M tokens(108%)

    Verified Jun 18

The one move that saved nothing is shown too — that's the point. Attribution is causal inference, so no-effect rows stay visible with their cause notes rather than being quietly dropped.

Plan fit · sizing signal, not waste

Unused allowance can't be recovered — you've already paid for it. This informs the next renewal.

Used 31.4B tokensForecast 32.4B tokens · 86%Allowance 37.5B tokens
Headroom
5.2B tokens

On plan — tracking to 86% of allowance; 5.2B tokens headroom. This is a sizing signal for renewal, not recoverable waste.

Efficiency signals

Diagnostic vital-signs — how waste is detected. The recoverable tokens and the fixes live in the lever sections above.

Cache hit rate
61%target 75%
Feeds: COACHING
Tokens / completed session
2.5M tokens
Feeds: COACHING
Tokens / accepted line
8.0K tokens
Feeds: COACHING

Tokens by model

Daily · last 30 days — model mix is the biggest lever, and it feeds the policy lever above

Opus 4.8Sonnet 4.6Haiku 4.5
Opus 4.8
19.4B tokens3.6M tokens/sess
Sonnet 4.6
9.4B tokens1.0M tokens/sess
Haiku 4.5
2.5B tokens400K tokens/sess

Usage anomalies

Token spikes vs 30-day baseline

  • Analytics+62%

    Large data-dump prompts spiked Opus usage

    Jun 28 – Jul 1+600M tokens
  • SRE Core team+41%

    Long unscoped sessions, no /clear

    Jun 26 – Jul 2+58M tokens
  • search-svc repo+33%

    Repeated full-repo reads, low cache hit

    Jul 1 – 2+48M tokens
Signal integrity · team-level
  • zero-rejection window14 sessions
    SRE Core · team-level
  • edits with no prior read9 sessions
    Vendor Pod Chennai · team-level
  • token spike, flat tool diversity6 sessions
    Analytics · team-level
  • abandoned-restart chain5 sessions
    Growth · team-level

Conversation starters — faking work costs nearly as much as doing it. Full context in Guard → Alerts.

Token breakdown

Group token usage and optimization headroom by any dimension

Repo / projectTokens / moShareOptimizable
web-storefront5.3B tokens
583M tokens(11%)
payments-svc4.6B tokens
1.0B tokens(22%)
ml-pipeline4.2B tokens
1.1B tokens(27%)
mobile-app3.1B tokens
469M tokens(15%)
analytics-etl3.0B tokens
744M tokens(25%)
api-graph2.6B tokens
312M tokens(12%)
auth-gateway2.4B tokens
218M tokens(9%)
search-svc2.1B tokens
409M tokens(19%)
billing-core2.1B tokens
409M tokens(19%)
infra-terraform1.8B tokens
256M tokens(14%)

Recover by team — vs each team's own baseline

Not a leaderboard. Each team is compared only to its own 90-day baseline, ordered by recoverable tokens — a work queue, top = biggest opportunity to coach.

ML Platform-3% vs own 90d
Recoverable / mo
116M tokens(14%)
825M tokens/mo totalCoach in Craft
Growth+1% vs own 90d
Recoverable / mo
104M tokens(16%)
650M tokens/mo totalCoach in Craft
Mobile-4% vs own 90d
Recoverable / mo
102M tokens(12%)
850M tokens/mo totalCoach in Craft
SRE Core+2% vs own 90d
Recoverable / mo
102M tokens(17%)
600M tokens/mo totalCoach in Craft
Analyticsflat vs own 90d
Recoverable / mo
100M tokens(16%)
625M tokens/mo totalCoach in Craft
Checkout-2% vs own 90d
Recoverable / mo
85M tokens(11%)
775M tokens/mo totalCoach in Craft
Cloud Foundations-1% vs own 90d
Recoverable / mo
83M tokens(15%)
550M tokens/mo totalCoach in Craft
API Platform-2% vs own 90d
Recoverable / mo
78M tokens(8%)
975M tokens/mo totalCoach in Craft
Developer Experience-1% vs own 90d
Recoverable / mo
44M tokens(6%)
725M tokens/mo totalCoach in Craft