Your CFO wants payback in 24 months. Your IT director wants a total cost of ownership comparison against the existing Wi-Fi contract. Your operations VP wants guaranteed uptime numbers. And your plant floor team, who will actually use it, wants to know if the AGVs will stop stopping. This article walks through how to build an ROI case for private cellular that answers all four — without inflating the numbers to make the project look better than it is.
Start with costs, not benefits
The mistake most ROI analyses make is leading with projected benefits before establishing a clear cost baseline. Reviewers will challenge the benefits; they will not challenge the costs. Build the cost case first and with precision, then let the benefits speak for themselves.
Capital costs (CapEx). A private 5G deployment has three hardware components: gNodeBs (the RAN access layer), the 5G core (either a vendor-supplied appliance or a cloud-hosted core), and cabling and mounting infrastructure. For an indoor CBRS deployment on a mid-size manufacturing facility (150,000 sq ft):
- gNodeB units: typically 8–12 units for a facility this size, at $4,000–$9,000 per unit depending on vendor and capability tier. Higher-end units support MIMO configurations that matter for dense AGV deployments; don't use entry-level specs if your AGV density exceeds 30 vehicles per cell.
- 5G core: $15,000–$40,000 for an on-prem standalone core appliance, or $0 CapEx for a cloud-hosted core with per-month subscription pricing. On-prem core carries higher upfront cost but eliminates the data sovereignty concern for facilities with OT/IT separation requirements.
- Installation labor, cabling, and commissioning: typically 40–60% of equipment cost. This range is wide because it depends heavily on how much conduit work is needed and whether you're running fiber to every cell or using existing cable infrastructure.
For the example above, expect $80,000–$150,000 in total CapEx before software licensing. Get firm quotes from at least two vendors. The range is wide enough that a generic estimate will be wrong by an amount that materially affects your payback calculation.
Operational costs (OpEx). Annual OpEx includes:
- Network orchestration software: $14,400–$40,800/year at typical orchestration platform pricing per campus (SAS fees bundled or billed separately by your RAN vendor).
- CBRS spectrum access: $0 for GAA; $1,200–$2,400/year for PAL, if applicable to your location and interference situation.
- Hardware maintenance and support: 15–18% of equipment cost per year, depending on vendor support tier.
- IT/OT staff time for network management: 4–8 hours/week for a single-campus deployment in steady state. Budget the higher end for the first 6 months post-commissioning while the team builds operational familiarity.
Over a 5-year period, total cost of ownership for a 150,000 sq ft deployment with on-prem core typically falls in the $280,000–$480,000 range, inclusive of CapEx amortization and ongoing OpEx. Use this range as a sanity check on vendor quotes — if a quote lands well below the lower bound, confirm what's excluded.
The three benefit categories worth quantifying
Private 5G benefits fall into three categories: avoided downtime cost, operational throughput improvement, and deployment efficiency gain. Quantify each separately — they have different evidence requirements and speak to different stakeholder concerns.
Avoided downtime cost
This is typically the largest line item and the most defensible in review because it uses your own operational data. The methodology:
- Pull 12 months of CMMS tickets and maintenance logs. Search for "network," "Wi-Fi," "AGV connectivity," "handover," "timeout," "reconnect," or equivalent terms your team uses. Count the events and record their documented resolution time.
- Use logged resolution time as your duration estimate, not the full production impact time. Reviewers will accept CMMS data; they will question anything that looks extrapolated.
- Calculate production value per minute. For a manufacturing facility, divide your daily output value by the total production minutes in the shift. For a parts assembly plant running two 8-hour shifts at $6M/day throughput, production value is approximately $625/minute. Be specific to your facility, not to an industry average.
- Multiply: events × average duration (minutes) × production value per minute. This is your annual avoided downtime estimate.
An important boundary here: this methodology estimates the cost of time already lost. It does not project future improvement. The projection is implicit — if the failure mechanism (connectivity-related AGV stops) is addressed, you avoid these costs going forward. Make this logic explicit in your presentation; reviewers who work in capital planning will recognize it as the more defensible framing.
Operational throughput improvement
Beyond avoided downtime, AGV fleets with reliable connectivity and enforced QoS can operate at higher planned throughput. Two mechanisms drive this: fewer emergency stop events (each requiring 3–8 minutes of recovery time including manual inspection before restart), and faster handover continuity enabling higher average floor speed in multi-zone travel.
Quantify this as fleet utilization rate improvement. Pull your AGV fleet management system's utilization data — most fleet management systems expose scheduled time, productive time, and idle/pause time as separate metrics. Calculate your current effective utilization rate (productive time / scheduled time). Estimate the delta from eliminating connectivity-related pauses based on your CMMS event data. Convert the utilization delta to trips-per-shift, then multiply by average load value per trip. This requires pulling your own data, not industry benchmarks — the analysis is only as persuasive as its source data.
Deployment efficiency gain
For facilities planning automation expansion — adding AMRs, extending AGV fleet coverage, deploying machine vision stations — predictive coverage planning from an existing private 5G model eliminates a significant rework risk. Average rework cost for a private cellular deployment that discovers coverage gaps during physical installation is 1.2–1.8× the original installation budget for the affected zones, because you're paying labor twice for the same scope. If your 3-year expansion plan includes two new production zones at $40,000–$70,000 per zone in installation budget, the rework avoidance value is $10,000–$50,000 per zone depending on the gap severity. This is option value, not hard cash flow — present it as a risk mitigation item, not a primary ROI driver.
How to structure the payback period
Use a simple payback model for the initial presentation, then build a full NPV analysis for capital committee review. Simple payback:
Payback period = Total CapEx ÷ (Annual avoided downtime + Annual throughput improvement)
For a facility with $120,000 in CapEx, $180,000 in annual avoided downtime, and $55,000 in throughput improvement, the simple payback is approximately 8.5 months. That's a strong number — but check it carefully before presenting it. If your payback calculation comes in under 12 months, review the avoided downtime methodology. Sub-12-month payback periods are legitimate for high-throughput manufacturing facilities with documented connectivity-related downtime, but they invite scrutiny. A reviewer who doubts the downtime calculation and discounts it by 50% still sees an 18-month payback, which is compelling. A reviewer who discounts it by 70% sees 30 months, which is still defensible. Build your avoided downtime number conservatively so it holds up under the discount.
For the NPV model: use a 5-year window, a discount rate appropriate to your organization's capital cost (8–12% is typical for manufacturing capital projects), and include all CapEx in year 0 and ongoing OpEx in years 1–5. Include the option value of future expansion as a line item with a conservative probability weighting — this is where you make the case that the infrastructure investment compounds over time, not just for the current AGV fleet.
The three CFO objections to prepare for
"Can't Wi-Fi 6E solve this at lower cost?" Address this directly in your analysis — don't leave it for the CFO to raise. We're not saying Wi-Fi 6E is inadequate for industrial use — for tablets, barcode scanners, and general-purpose IoT it is the correct and lower-cost answer. Present the Wi-Fi 6E alternative cost (typically 30–50% lower CapEx for equivalent coverage area), then show specifically which device classes in your fleet require the QoS guarantees that Wi-Fi 6E cannot contractually provide. A hybrid architecture that deploys private 5G only for AGV corridors and keeps Wi-Fi 6E for everything else often has a stronger ROI case than full campus private 5G — lower total cost with the same safety-critical performance improvement.
"The downtime numbers look high." Pre-empt this by attaching the raw CMMS export as an appendix. If your maintenance logs don't capture connectivity events with enough specificity, run a 30-day instrumentation period before building the ROI: deploy logging agents on your AGVs that record every network-related pause with timestamp, duration, and cell ID. Thirty days of instrumented data from your own fleet is more persuasive than any industry benchmark figure.
"We have a 3-year IT refresh cycle and this is a 5-year commitment." Address the refresh risk directly: CBRS operates on a standardized 3GPP air interface, which means new-generation UEs (including your next AGV model) maintain backward compatibility with existing gNodeB infrastructure. The hardware investment is not a sunk cost at year 3 — it's the RF layer that future devices connect to. The management software layer is where functionality evolves and updates, not the antenna infrastructure. Structured correctly, the refresh risk is a software subscription decision, not a hardware rip-and-replace question.
The number that matters to operations
Your operations team doesn't care about NPV. They care about one number: will the AGVs stop stopping? Give them a specific commitment grounded in technical architecture: with enforced QoS at 5QI 82 on the AGV control channel, covered by a monitored 99.9% uptime SLA at the network layer, network-related emergency stops should drop to near zero. Define "near zero" explicitly — fewer than two network-attributed events per quarter per AGV, verified against your fleet management system logs. That is a measurable commitment, not a vendor claim.
That specific commitment is what gets operations to champion the project through internal review. The financial case gets it through the capital committee. You need the operations champion before you need the capital approval — without it, the CFO has no reason to believe the benefit numbers.