Skip to content
Search

Automation Project Contracts

The contract is a risk allocation document. Every line in the commercial terms is answering one question: who takes the loss when something goes wrong?

Dominant pricing model in North American material handling. The integrator prices a fixed total for defined scope. Cost overruns are the integrator’s problem; cost savings stay with the integrator.

Integrators build contingency into the number — typically 5–15% depending on design completeness at bid. On a contested $20M bid, contingency may be squeezed to 3–5%. Every out-of-scope condition triggers a Change Order Request (COR). This is not opportunism — it is how the model is designed. Vague RFP scope pre-authorizes a series of CORs.

The earning principle: lump-sum pricing is only fair to both parties when the Firm Design phase is complete before the contract is signed.

Owner pays actual costs up to a ceiling. Used when design isn’t finalized but work must start. Common in general construction; appears on large complex automation projects where a lump sum can’t be priced without completing detailed design first.

GMP requires open-book accounting — most integrators are uncomfortable with this — and shifts design-change risk back toward the owner.

Standard for concept studies, assessments, and post-warranty service. Should not be the delivery model for a full system installation. Software change orders are frequently billed T&M even when the base project is lump sum. Always negotiate a budget cap on T&M change orders — uncapped T&M software changes are the silent killer of post-go-live project costs.

AutoStore and THG Fulfil launched pay-per-pick deployments in 2023, converting $3–8M CAPEX AutoStore installations into OPEX subscriptions. The customer pays per bin retrieval or per pick with no upfront hardware investment.

Economics favor the customer when: volume is uncertain or variable, capex is unavailable, or the customer wants to test before committing.

Economics favor the integrator when: volumes are high and predictable — you will pay more per unit on subscription than on a financed CAPEX purchase over five years.

On lump-sum contracts, payment milestones follow this approximate sequence:

MilestoneTypical %
Contract signing10%
Firm design completion10%
Long-lead procurement15%
Fabrication completion / FAT20%
Mechanical installation complete15%
Commissioning complete10%
SAT acceptance10%
Warranty expiration10%

The integrator’s primary leverage is completing commissioning to release the SAT milestone. The customer’s leverage is retainage: 10–15% of contract value withheld until SAT or warranty close on a $20M project = $2–3M.

Throughput: A specific rate — units per hour, picks per hour, cases per minute. Not a range. The Symbotic/Walmart deployment documents 1,350 cases per hour per palletizing cell. The rate must specify measurement conditions: product mix, order profile, staffing at induction and exception stations.

Accuracy: 99.9% order accuracy is standard for sortation; 99.99% is common for AS/RS-based operations. The measurement methodology must be defined: measured at the sorter or at point of shipment verification? Label-read failures: accuracy failure or mechanical failure?

Availability/Uptime: 95–99% is the typical range. The critical dispute: a five-stage sequential system where each stage is individually rated at 99% produces whole-system availability of 99%^5 = 95.1%. The contract must define whether “availability” means any single component or the integrated system end-to-end.

Governed by IEC 62381:2024, conducted at the integrator’s or equipment manufacturer’s facility before shipment. Covers:

  • Documentation verification
  • Physical/dimensional inspection
  • Functional testing of all operational modes and safety interlocks
  • I/O point-by-point verification
  • Software logic testing
  • Communications testing

Post-2023, virtual FAT (remote witness via video and live dashboards) is formally recognized by IEC 62381:2024. Adequate for software and controls verification; inadequate for physical inspection of mechanical components.

IEC 62381:2024 introduced FIT as a fourth stage. Tests multi-vendor subsystems together before shipment — the only opportunity to discover interface bugs before commissioning. For any project with multiple technology vendors, FIT is not optional.

The contractual performance gate per IEC 62381:2024:

  • Minimum 72-hour continuous run at production load (168 hours for complex or regulated systems)
  • Unplanned stops reset the clock
  • Measured throughput must achieve ≥95% of contracted rate at full load
  • Accuracy and uptime measurements captured over the run period

SAT triggers the final major payment milestone, typically 10% of contract value.

Cost-of-defects hierarchy:

  • Found at FAT: 1×
  • Found post-installation (SAT prep): 10×
  • Found post-go-live: 50–100×
CategoryDefinition
ABlocks shipment. Must be resolved before equipment leaves the factory
BMust be resolved before SAT. System can ship but items must be closed during commissioning
CMinor. Documented deficiency with agreed resolution date. Does not block SAT

Tighten the Category B definition in the contract and require a closed-punch-list milestone before retainage release. Projects where 30 Category B items remain open at SAT, get waived, and then languish as Category C for months are common.

Capped at 5% of contract value for integrator-caused delays. More common in Europe than North America; in the US “rarely enforced” — used as negotiating leverage, not direct financial remedy. Their primary function is to focus the integrator on schedule during the bid process.

Triggered by failure to meet throughput or availability guarantees after SAT remediation. Specify explicitly:

  • Performance floor that triggers LDs
  • Measurement period (e.g., 90 days of production)
  • Remediation notice period (e.g., 30 days to propose a fix plan)
  • Remedy execution window (e.g., 60 days to implement and retest)

Industry standard: schedule LDs capped at 5%, performance LDs capped at 10–15% of contract value.

These caps roughly equal the integrator’s profit margin on the project. The economic architecture is intentional: integrators will not bid projects where LD exposure exceeds their margin. The uncomfortable truth: a $3M system at 90% throughput may affect a $500M/year distribution operation far more than the $300–450K LD cap represents. The contract is a floor, not a ceiling, on the integrator’s performance incentive.

Every contract contains explicit exclusions:

  • WMS scope: The integrator’s WCS interfaces with the WMS but does not control it. WMS-caused performance shortfalls — wrong divert commands, orders not releasing, WMS downtime — are excluded from throughput guarantee measurements.
  • Non-conforming product: Throughput guarantee applies only to conforming product within the dimensions, weights, and packaging types defined in the BOD. Higher-than-specified non-conformance rates are the customer’s problem.
  • Building readiness: Floor flatness, column grid accuracy, power service, HVAC, fire suppression. If the floor fails flatness spec and an AGV can’t navigate a zone, that is not the integrator’s problem.
  • Owner-furnished items (OFIs): Network infrastructure, operator workstations, and direct-purchase equipment the customer hands to the integrator to connect.
  • Force majeure: Post-2020, integrators have pushed to include supply chain failures. PLCs/VFDs that went from 8–12 weeks to 40–60+ week lead times in 2021 are the reference case.
  • WMS modifications: If the WMS requires development to provide data the WCS needs, that development is the customer’s cost.

Standard retainage: 10% of contract value, released at SAT acceptance. Stronger customer protection: release in tranches — 50% at SAT, 50% at warranty expiration. On a $20M project, the retainage tranche releasing at warranty expiration is $1M, which directly incentivizes punch list closure through the warranty period.

Warranty bonds: Less common in North America than Europe. On large projects (>$20M) or public projects, customers may require a performance bond from a surety guaranteeing the integrator’s warranty obligations. Bond premiums run 1–3% of project value — a real cost integrators price into their bids.

The structural limitation: contractual mechanisms are necessary but structurally insufficient to compensate actual business loss at scale. A $300K bond on a $3M system does not compensate for six months of throughput shortfall on a DC shipping $500M annually. The only full remedy is a system that works — which requires getting the FDS, ICD, and testing protocol right in the first place.

Source: 2.6-advanced-automation-design

Pro content

Subscribe to read the rest

This article is part of our Pro library — practitioner-level guidance, frameworks, and decision tools written from real projects.

$9/mo Basic · $13/mo Pro · cancel anytime