Value — what the fleet is doing for you

What the fleet is used for, what that is worth, and what it costs to get. Every figure is derived from run data on the connected instance; the assumptions behind it are open at the bottom of the page and editable.

Adoption Live + seeded

22 of 23 running

25,754 runs in the last 30 days

17 earning measured value74%

3 with an unclassified actionnot valued

1 deployed but idleSocial Listening

Realised value · 30d Derived

1,687 hours

12.1 people-months · $101,227 at $60/h

123 h lost to failed runs7%

8,544 scheduled checksno human time assumed

171 h still spent reviewinggated actions

Return Derived

4.9× on what it costs

$80,576 net after $20,651 of cost

$101,227 value delivered

$20,651 total cost to runitemised below

$3,503 net per automationaverage

What it costs to get that Derived

Automation is not free. Four costs, each from the same run data as the value above.

$10,268$7,365$2,760

Human review of gated actions · 171 h$10,268

Failed runs and redoing them · 123 h$7,365

Upkeep · 2 h per automation$2,760

Platform execution · 25,754 runs$258

85% of this cost is human time, not compute. Cutting the failure rate moves it more than any change of plan.

By team Fleet

Same arithmetic, grouped by who is accountable.

Customer Care

606 h

RevOps

364 h

Marketing

282 h

Finance

263 h

IT Platform

119 h

People

53.5 h

Legal & Compliance

0.0 h

unassigned — nobody has claimed this

0.0 h

Teams come from Fleet’s own ownership record — n8n has no column for an accountable human, so this grouping cannot be read off the instance. A row reading unassigned — nobody has claimed this is a governance finding, not a formatting gap: the hours and the cost on that line have nobody to defend them.

Where the value is Derived

Every row is delivered runs × times the action fires × minutes a person would have spent.

Support Triage Agent

4,038 clean runs · 4,038 × 9min draft

606 h$36,342

Lead Router

2,208 clean runs · 2,208 × 3min record + 2,208 × 2min notify

184 h$11,040

Onboarding Email Sequence

5,292 clean runs · 5,292 × 2min notify

176 h$10,584

Refund Agent

604 clean runs · 604 × 12min money + 604 × 3min record

151 h$9,060

Lead Enrichment Agent

1,974 clean runs · 1,974 × 4min lookup

132 h$7,896

Data Warehouse ETL

29 clean runs · 1,160 × 4min lookup

77.3 h$4,640

Content Repurpose Agent

114 clean runs · 114 × 20min publish + 114 × 20min publish

76.0 h$4,560

Expense Approval Agent

854 clean runs · 854 × 3min approve + 854 × 2min notify

71.2 h$4,270

Support Triage Agent alone is 36% of the total. Change one minute figure in the panel below and the headline moves with it.

Where value is leaking Derived

Hours at stake first, then what cannot be put in hours.

29.7 h90% of its value

Invoice Sync

31% of 720 runs fail. What this is meant to handle lands on a person instead.

24.6 h4% of its value

Support Triage Agent

Only 2% fail — but on 4,120 runs that is 82 a month. A low rate on a big number is still someone's week.

16.8 h13% of its value

Lead Enrichment Agent

6% of 2,100 runs fail — 126 handoffs back to a person, plus the cleanup.

15.3 h8% of its value

Lead Router

4% of 2,300 runs fail — 92 handoffs back to a person, plus the cleanup.

8.0 h11% of its value

Content Repurpose Agent

5% of 120 runs fail — 6 handoffs back to a person, plus the cleanup.

Unpriced

Churn Risk Scorer

10% of 30 runs fail — 3 of them. No action here has a handling class, so Fleet names the loss rather than putting an hours figure on it.

high risk

Churn Risk Scorer

Carries high risk and “Write churn score to Salesforce” has no handling class, so Fleet will not put a number on it. Worth classifying or retiring.

Unpriced

Contract Review Agent

8% of 40 runs fail — 3 of them. No action here has a handling class, so Fleet names the loss rather than putting an hours figure on it.

high risk

Contract Review Agent

Carries high risk and “Comment on contract” has no handling class, so Fleet will not put a number on it. Worth classifying or retiring.

Assumptions19 inputs · 12 handling classes in play · every figure above recomputes as you change them

Read an incoming item and decide where it goesmin

Compose a written reply for a person · in usemin

Create or update a record by hand in another system · in usemin

Look data up in one system and copy it into another · in usemin

Move, refund or reconcile money · in usemin

Grant or revoke access to a system · in usemin

Write and publish something customer-facing · in usemin

Compile a summary and circulate it · in usemin

Arrange a time with a person · in usemin

Check a claim against a rule and approve it · in usemin

Tell a named person something happened · in usemin

A scheduled check — 0 on the assumption nobody was doing it by hand · in usemin

Hands work to a person — 0, it creates human work rather than displacing it · in usemin

Deducted when a human already reviews a gated actionmin

Cleanup multiplier on a failed run×

Upkeep per automation per monthh

Platform cost per execution$

Hours in a person-monthh

Loaded hourly rate$

Hours and money are derived from run data, never recorded Derived. The minutes, the fires-per-run figures and the six inputs above are authored in Fleet Fleet — n8n has nowhere to store them. Period-over-period comparison begins once Fleet retains history.

The classification table. 28 action classifications ship with this build; 27of them key workflows from the prototype’s demo estate, which is not on this instance Seeded. The remaining one keys this instance’s workflow under an action label the live sync does not produce. Nothing here is matched fuzzily: a near-miss on a label is treated as no classification at all.

Not valued. Fleet will not guess at an action nobody has classified, so these contribute zero:

Churn Risk ScorerWrite churn score to Salesforce

Contract Review AgentComment on contract

SEO/AEO Growth Loop (Slack -> Claude Code routine relay)Fire Routine, Post Session URL to Slack, Post Fire Failure to Slack, Post Completion to Slack, Publish Report