◆Agents that turn product signals into revenue.northstar geocities
Simulated case study · Warehouse robots

Halden Robotics

A 180-person company that sells autonomous carts to warehouses: the robots, a yearly service plan, and fleet software. Most new customers start with a paid pilot of a few units.

Simulated case study. The company is invented and every number is a simulation, not a customer result. Changes marked “custom” would be built for that customer and are not in Northstar today. Some standard features are still in beta.
People
180
Confirmed ARR
$7.5M
Fleets live
14
Pilots running
9
The problem
  • Pilots ran for months. Field engineers heard what warehouses needed, but it lived in trip reports nobody in Product read.
  • A hardware change costs far more and takes far longer than a software change, but both were argued about the same way.
  • Revenue came from robots, service plans, and software together, so it was hard to see what one request was really worth.
What we set up

Standard parts are the same for every customer. Custom parts would be built for this company.

standard

CRM connected

Fleet accounts, pilots, and deal stages, with notes from sales and field engineers.

standard

Pre-revenue weighting

Pilots ranked by deal size and stage before any fleet contract exists.

custom

Fleet revenue model

A request is valued by the whole fleet it affects: units × unit price, plus the service plan and software.

custom

Hardware cost and lead time

Build cost includes parts, testing, and supplier lead times, so hardware changes are not mistaken for cheap ones.

custom

Safety and certification factor

Requests that affect safety certification are flagged and carry extra review and cost.

custom

Field reports source

Trip and service reports from field engineers read for product requests, the same way as sales notes.

Rollout

Week 1

CRM connected. Fleet revenue model set up with their finance lead.

Week 3

Hardware cost inputs and the safety factor agreed with engineering.

Week 6

Field reports flowing in. First quarterly roadmap planned from the ranked list.

A week with Northstar

Field engineer

Files the usual trip report. Northstar picks out that two pilot sites need carts to handle narrow aisles.

VP of sales

Sees that narrow-aisle support is holding up 3 pilots worth 60 units. Brings it to the roadmap meeting with the fleet value attached.

Head of hardware

Compares a new sensor (high cost, 5-month lead time, safety review) with a software route-planning fix that solves most of the same problem in 6 weeks.

Year one, simulated
Savings
$490K
Confirmed revenue
$156K
Net after cost
$546K
Back per $1
$6.46
ROI
546%

ROI is the gain after paying for Northstar: $546K on $100K. Counting savings only, with no new revenue, it is still $4.90 back per $1, or 390% ROI.

Savings

Field and sales time back
20 field engineers and sellers × 2 hours a week × 46 weeks × $80 an hour
$147K
Product and engineering lead time back
6 people × 3 hours a week × 46 weeks × $100 an hour
$83K
Hardware change avoided
1 hardware change replaced by a software fix: about $260,000 in parts, tooling, and testing
$260K

Confirmed revenue

Pilots turned into fleets
2 pilots became fleet contracts after the software fix; we count 30% of their $520,000 yearly revenue
$156K

Weighted pipeline (not counted)

Weighted pipeline from the other pilots
Open fleet deals weighted by stage
$1.9M

Year-one cost: $40K setup plus $60K for the year. Assumed for this simulation. This build needs more setup than the others, and custom builds are priced per customer.

Green is savings and confirmed revenue. Gold is weighted pipeline, an estimate by deal stage, and is never added to the return.

Assumptions
  • Hardware savings assume one change avoided in year one. Some years there will be none.
  • Only 30% of new fleet revenue is credited to Northstar, since pilots depend on many things.
  • Weighted pipeline is an estimate by deal stage and is not counted in the return.

Northstar helps decide what to build. It does not control robots or take part in their safety systems, and Agent Control covers software agents only.

More simulated case studies: Larkspur Agent Works (ai agent platform) · TorqueDesk (software for auto repair shops) · All case studies

Make this real for your company

Design partners help us replace these simulations with measured results, and get our lowest prices and a real say in what we build.

Become a design partnerSee pricing