How to Evaluate Condition Monitoring Solutions for a Multi-Site Manufacturing Portfolio

A Plant Manager evaluates condition monitoring for one plant. A Plant Director evaluates it for a portfolio of 3 to 15 or more sites with different equipment bases, different team capabilities, and different risk profiles. The evaluation criteria are not the same.

A platform that performs well on a single-site deployment can fail to scale across a portfolio if it requires per-site IT infrastructure, produces non-comparable alerts across sites, or cannot give the Plant Director a consolidated view without manual data reconciliation. The evaluation framework at this level is not about sensor specifications or alert accuracy in isolation: it is about whether the platform can serve as a common performance layer across a portfolio that is inherently diverse.

This guide covers the must-have criteria, the red flags that disqualify vendors regardless of their single-site performance, and the portfolio ROI calculation that turns the evaluation into a capital allocation decision.

Why Single-Site Evaluation Criteria Fail at Portfolio Scale

Site-level evaluation optimizes for different things than portfolio-level evaluation. At the site level, the questions are: does it detect faults on our specific asset types, is the alert interpretable by our team, and does it integrate with our existing work order system? These are necessary questions. They are not sufficient at the portfolio level.

At the portfolio level, the additional questions are: does every site produce data in a format that is directly comparable without transformation, does the platform give the Plant Director a consolidated view without requiring each site to prepare a report, and does the pricing model allow the portfolio to grow without renegotiating terms or rebuilding the deployment model?

A platform that requires a different configuration at every site may produce good data at each site and useless data at the portfolio level. Consistency across a diverse site population is not a feature of single-site deployments: it has to be designed into the platform architecture and evaluated explicitly.

Must-Have Criteria for a Portfolio Deployment

Single platform visibility across all sites without manual consolidation. The Plant Director's view must show all sites in a single interface with a consistent data model. Not a dashboard that aggregates reports submitted by each site. Not a data export that requires a monthly consolidation exercise. A native cross-portfolio view with real-time or near-real-time data from all monitored assets.

No per-site IT infrastructure requirement. Platforms that require on-premise servers, VPN connectivity to corporate IT, or local network integration at each site create a deployment dependency on each site's IT team and timeline. In a portfolio where sites have different IT maturity levels and different IT support availability, any per-site infrastructure requirement will delay some sites indefinitely. Cloud-native deployment with no on-premise requirement is the only model that scales consistently across a heterogeneous site portfolio.

Standardized alert taxonomy across all sites. A severity classification must mean the same thing at every site. When the Plant Director reviews portfolio alerts, a severity-2 fault on a stamping press motor bearing at Site A must be directly comparable to a severity-2 fault at Site B. This requires that the platform applies consistent threshold logic, not site-specific calibrations that are adjusted locally and lose comparability.

Deployment model that does not require a resident reliability engineer at every site. Platforms that produce raw vibration spectra or raw sensor data require an on-site analyst to interpret it. Most discrete manufacturing sites do not have a resident reliability engineer. The platform must deliver interpreted alerts that specify the asset, the failure mode, the severity, and the recommended action, with enough clarity that a maintenance lead can act on them without specialized diagnostic training.

Red Flags That Disqualify a Vendor

Requiring site-by-site commercial negotiation. If the vendor's pricing model requires separate contract negotiation at each site, the terms will diverge. Some sites will be under different SLAs, different support tiers, and different pricing structures. Portfolio standardization becomes structurally harder every time a site renews or expands on different terms.

Per-site data silos. Some platforms produce excellent site-level data that cannot be aggregated across sites because the data schema is site-specific or the data lives in separate cloud instances. The vendor may offer a "portfolio view" as an add-on product or custom integration. If cross-portfolio visibility requires additional engineering work or additional licensing, it is not a portfolio-ready platform: it is a single-site platform with a portfolio marketing claim.

Alert systems requiring local interpretation. If the alert says "high vibration on asset 14" and stops there, the value is limited. The maintenance lead must know which asset, what the failure mode is, what action to take, and within what timeframe. A platform that requires the site team to interpret raw sensor data before they can act generates alerts that will be ignored, deprioritized, or acted on inconsistently across sites.

Pricing that cannot scale. Get the full pricing structure for the portfolio, including what happens at 5 sites, 10 sites, and 15 sites, and what happens when coverage is expanded from 20 assets per site to 50 assets per site. A platform with per-asset or per-site pricing that does not include a volume discount structure will become the most expensive item in the maintenance budget before the portfolio deployment is complete.

The Portfolio ROI Calculation

The evaluation is a capital allocation decision. Build the financial case before presenting it.

Step 1: Aggregate downtime cost across the portfolio.

Calculate unplanned downtime cost at each site using the same three-component formula: production loss (Tier 1 downtime hours times production value per hour at each site), emergency repair premium (documented from work order history, typically two to three times the planned repair equivalent), and OEM penalty exposure for JIT supply sites.

Sum across all sites. This is your portfolio baseline.

Step 2: Apply a conservative prevention estimate.

Predictive maintenance targets condition-based failures where degradation is gradual and detectable. In discrete manufacturing, these represent 50% to 70% of unplanned failures. Year-one prevention estimate: 20% of current unplanned events are detected and repaired before failure. Apply this rate to your portfolio baseline.

Step 3: Calculate total platform cost across all sites.

Get firm quotes: hardware for Tier 1 assets at each site, software subscription across the portfolio, and onboarding cost per site. Sum these for the total portfolio program cost.

Step 4: Portfolio payback calculation.

Portfolio net annual benefit = (Portfolio baseline x 20% prevention rate) minus annual platform cost.

Portfolio payback period = total platform investment divided by monthly net benefit.

A program covering a portfolio where aggregate Tier 1 downtime cost is $4M annually, at a 20% prevention rate and $600K annual platform cost, produces $800K net annual benefit with an 18-month total investment payback. That is a capital allocation conversation, not a maintenance budget conversation.

Step 5: Sensitivity test.

Run the calculation at 10% prevention (floor case). If the floor case is still cash-positive, the investment is defensible even under the most pessimistic reasonable assumption. Most multi-site portfolios with $3M or more in aggregate downtime cost will be cash-positive at 10% prevention with a platform priced for scale.

How to Structure the Pilot

Run a structured pilot at one Tier 2 site before committing to portfolio deployment.

Site selection: A Tier 2 site has enough active reliability problems to generate meaningful alerts (not a site that rarely has failures) but not so many that alert volume overwhelms the maintenance team before they build confidence in the platform. The site should have at least five Tier 1 assets within scope.

Pre-defined success criteria: Before the pilot starts, define:

• Alert confirmation rate: what percentage of generated alerts corresponded to actual developing faults
• Time from alert to repair scheduling: how quickly does the team act on an alert
• Documented prevented failure: at least one event where the alert led to a planned repair that would otherwise have been an unplanned failure

Pilot duration: 90 days minimum. The first 30 days is calibration. The signal-to-noise ratio improves as the platform learns baseline operating conditions. Evaluate the platform on days 60 to 90, not days 1 to 30.

The portfolio-readiness question: At the end of the pilot, the evaluation question is not whether the platform worked at this site. It is whether the platform behaves consistently enough to be deployed across a portfolio where the Plant Director cannot personally supervise every site's alert response. Ask the vendor to demonstrate how a severity-2 alert at this site would look compared to the same fault type at a different site in the portfolio.

The Vendor Conversation: Questions That Reveal Portfolio Readiness

Ask these questions directly. The quality of the answers reveals whether the vendor has actually deployed across multi-site portfolios or is pitching a single-site product with portfolio language.

• "Describe a customer who deployed across five or more sites simultaneously. What was the timeline and what were the deployment blockers?"
• "How does the alert taxonomy stay consistent when Tier 1 asset types differ between sites?"
• "What does the Plant Director-level view look like across a 10-site portfolio? Can you show me a live example?"
• "How does pricing change when we expand from 3 sites to 10 sites and from 20 assets per site to 50?"
• "How does data from our oldest site, which has no existing digital infrastructure, flow into the same portfolio view as our newest site, which runs a modern ERP?"

A vendor who gives consistent, specific answers to these questions has portfolio deployment experience. A vendor who defaults to single-site performance claims when asked portfolio-level questions is not portfolio-ready.

How to Build the Evaluation Scorecard

Score vendors on the following criteria before any final recommendation:

Any vendor that fails a high-weight pass/fail criterion is eliminated regardless of score on other criteria. A platform that cannot be standardized across your portfolio is not a portfolio solution regardless of its single-site capabilities.

CapEx lifecycle and capital deferral evidence: Evaluate whether the platform retains 12–24 months of continuous condition trend data per asset in an exportable format. That historical trend data is the evidence base for condition-based CapEx decisions: proving an asset has remaining service life to defer replacement, or providing the condition-based justification when replacement is genuinely needed. A Plant Director presenting a capital request supported by condition trend data is in a fundamentally stronger position with the board than one requesting capital based on age or reactive failure history.

Secondary and catastrophic damage prevention: Evaluate whether the platform detects faults at early severity stages, weeks before they cascade into secondary damage. A $500 bearing on a stamping press motor that is not caught early does not just fail the bearing; it destroys the shaft, contaminates the gearbox, and in a worst case takes out the motor itself. For a Plant Director answerable to the VP for quarterly production targets, a cascade failure on a critical bottleneck asset is the kind of event that defines the quarter. Early-stage detection is what converts those events from catastrophic failures into planned changeover window repairs.

Headcount force multiplier, Auto Diagnosis™: Evaluate whether the platform provides automated failure mode identification, not just threshold alarms, but specific fault classification (bearing fault type, rotor unbalance, misalignment, looseness), without requiring a trained vibration analyst to interpret the data. Tractian's Auto Diagnosis™ embeds the diagnostic expertise in the platform. A maintenance technician receives the failure mode and the recommended action. For a Plant Director who cannot secure headcount approval for additional reliability specialists, this capability multiplies the effectiveness of the existing team across every monitored critical asset.