103 pilots · through 2026-06 ·
$68,000 of hardware per unit on loan · synthetic data
Pilots started
103
30 still open
Conversion rate
32%
of decided pilots
Median time to production
6 mo
True CAC
$580,489
5.3x the naive $109,630
Hardware at prospect sites
$3.7M
earning nothing
The number that changes the conversation:
naive CAC counts only the pilots you won — $109,630. True CAC
divides all pilot cost by the wins:
$580,489, or 5.3×.
The lost and stalled pilots are not overhead; they are the cost of acquiring
the customers you did win. At an average $191,585 ACV and
68% deployment gross margin, a won customer
produces about $130K of gross profit
a year — so a three-year contract recovers roughly
$391K against
$580K of CAC, leaving
$190K
unrecovered. The minimum term that clears the hurdle is about
4.5 years — longer than customers will sign.
The blunt conclusion: as currently run, this acquisition motion does
not pay back on any realistic contract term. (An earlier version of
this page called the three-year term "workable" off a revenue-basis ratio;
an external review correctly called the contradiction — gross profit, not
revenue, repays CAC.)
The counterfactual is what makes this fixable rather than
fatal: had only the qualified pilots been run — named sponsor plus
written success criteria — true CAC would be
$170,357, repaid by gross profit in about
16
months — comfortably inside a three-year term. The unqualified
pilots are not underperforming; they are the entire reason the motion as a
whole does not pay for itself, and a qualification gate is the difference
between a broken funnel and a working one.
The funnel
F1 — Conversion by pilot age
Rates are computed on decided pilots only — the 30 still
open are right-censored and excluded, so recent cohorts' rates can move as
they resolve. Conversion collapses past the 9-month mark.
That makes pilot purgatory datable, which is what turns "we should
probably close some of these" into a defensible kill rule with a number
attached. Everything to the right of the dashed line is engineering capacity
being spent at close to zero expected return.
F2 — Conversion by qualification quality
Qualification
Pilots
Decided
Conversion
Cost
ACV won
sponsor + criteria
28
26
54%
$2.38M
$2.54M
sponsor only
20
13
8%
$2.96M
$0.16M
criteria only
35
23
30%
$5.24M
$1.52M
neither
20
11
9%
$2.77M
$0.18M
A named executive sponsor and written success
criteria is the difference between a pilot and an experiment. The
unqualified cohort consumes real cost and returns little — the cheapest
intervention available here is a qualification gate before a robot ever
ships, not a better pilot.
Pilot purgatory — open longer than 9 months
Pilot
Company
Industry
Age
Sunk cost
Hardware on site
P0013
Halcyon Co
cold chain
27 mo
$261,250
$204,000
P0003
Brightpath Co
food bev
29 mo
$235,875
$68,000
P0050
Northwind Co
manufacturing
16 mo
$210,500
$136,000
P0065
Cascade Group
food bev
12 mo
$196,000
$204,000
P0057
Cascade Group
logistics
14 mo
$192,375
$68,000
P0068
Maple Group
cold chain
11 mo
$181,500
$136,000
P0072
Meridian Partners
food bev
10 mo
$167,000
$136,000
P0073
Foxglove Industries
automotive
10 mo
$167,000
$68,000
P0074
Ridgeline Industries
cold chain
10 mo
$145,250
$68,000
9 pilots,
$1,756,750 sunk, and
$1,088,000 of hardware
sitting at prospect sites earning nothing. This is the list to work, and the
finance recommendation is not "try harder" — it is a decision date on each
one, and recovery of the units from the ones that miss it.