Every asset-intensive operation runs the same set of maintenance strategies. Work orders get triaged. Inventory gets buffered. Spare parts get kitted. Surplus gets rationalized. Suppliers get scored.
What changes from one operation to the next isn’t which strategies are running. It’s the maturity level each one is running at.
Some plants triage work orders by gut feel and FIFO. Others have priority codes locked into the CMMS. A smaller number have AI-scored, criticality-weighted ranking that recomputes against live production data. All three are technically running the same strategy. They are very different operations underneath it.
This article walks through eleven of those strategies, sequenced along the operational chain (decide what matters, plan around it, execute it, tighten the loop), and maps each one across three operational maturity levels.
Spreadsheets, tribal knowledge, decisions made on personality and escalation. Reorder points set in a workshop and frozen in Excel. Criticality treated as a one-off alignment exercise. Works at small scale, falls apart at plant-network scale.
CMMS, ERP and EAM modules capture the workflow. Priority codes, min/max levels and asset registers exist. But the logic underneath is static, set once and rarely revisited. The system is a record, not a decision support layer.
Live data flows from ERP, EAM and IIoT sensors. AI agents do the heavy lifting (scoring, parsing, classifying, matching), and human experts override with reason. The override trains the next decision. Criticality is continuous, kitting is WO-driven, reorder math is dynamic.
Criticality is the thread that runs through all eleven. The move from Systematized to Intelligent is almost always the move from static criticality to live criticality.
What follows is each strategy, the maturity arc it sits on, the overarching problem it solves, and how the right software (paired with the functional discipline behind it) moves it up the ladder.
Layer 1: The Decision Layer
The first two strategies decide what matters. Everything downstream borrows from these decisions, so when they’re wrong, the rot travels.
1. Criticality-Based Work Order Prioritization
The workflow. Open work orders enter a queue and the queue gets triaged. Jobs with the highest consequence to safety, production or compliance get labor and parts ahead of routine work.
Where it breaks. Triage runs on gut feel, first-in-first-out queues, or whoever has the loudest voice that morning. A leaking gasket on a critical compressor lines up behind a lighting fix in the canteen because both came in at the same time.
The overarching problem is wasted capacity on the wrong work. Maintenance teams are rarely idle. They are usually busy on the wrong things. Every hour spent on low-consequence work is an hour denied to a job that is genuinely putting production at risk.
The software-assisted fix. A criticality score, computed against safety, production impact, lead time and substitutability, drives an objective priority ranking. The AI agent lays out the case for each rank. The planner approves or overrides with a reason attached, and the override teaches the model. This is where evaluación de la criticidad de los activos stops being a once-a-year workshop and becomes a daily decision support.
2. Asset and Part-Level Criticality Analysis
The workflow. Assets, and the parts inside them, are scored for how badly their failure would hurt the operation.
Where it breaks. Most plants run criticality as a one-time alignment exercise, championed by a reliability lead with inputs from several teams. The output is a static register that ages out within months. Worse, almost every implementation carries a quiet assumption that doesn’t survive contact with reality.
The fatal flaw
Conventional criticality assumes every part on a critical asset is itself critical, and that parts on non-critical assets are not. In practice, both halves of that statement are wrong. A single low-cost gasket can take down a tier-one compressor. A high-value bearing can sit on an asset whose failure has zero production impact because a redundant unit is right next to it.
The software-assisted fix. Multi-model scoring (FMECA, VED, ABC) overlaid with first-party ERP and EAM data, plus industry-specific failure pattern training, produces true part-level criticality. This is the kind of calculation no human can do at scale.
The functional discipline matters as much as the model. A subject matter expert reviews the score, can override it with justification, and that learning rolls out across the plant network. The part-level layer is what makes downstream strategies (kitting, reorder points, predictive triggers) honest.
You can read more on the underlying mechanics in part-level criticality scoring and how it ties into FMEA and FMECA best practices.
Layer 2: The Planning Layer
With criticality scored, planning has something honest to work with. The next three strategies turn that signal into prepared parts, dynamic stock levels and clean data underneath. This is where most of the operational waste is created or avoided.
3. Inventory Kitting
The workflow. Every part a job needs is picked and staged as a kit before the technician picks up the work order.
Where it breaks. The technician arrives at the job and a part is missing, wrong-sized, or already issued to another work order. Wrench time bleeds out on storeroom trips. In high-utilization plants, the gap between scheduled start and actual start can run into hours per job.
The overarching problem is downtime that nobody planned for. The job is technically resourced. The parts are technically in the warehouse. The work order is technically open. And yet production is waiting because three of those technicalities don't line up at the bench.
How kitting fails
Pick lists are generated late, parts are misidentified across duplicate SKUs, kits are built without confirming live stock, and substitutes aren't surfaced when the primary part is short. The technician finds out at the asset, not at the storeroom.
How kitting holds
The system reads the work order, pulls the required parts via accurate BOM linkage, confirms availability against live stock, surfaces approved substitutes, and confirms physical location before the job is scheduled. The kit is built ahead of the technician.
The software-assisted fix. Kitting only works if the inventory record is accurate, the BOM linkage is real, and the system knows where every part physically sits across plants. This is where spare parts inventory optimization earns its keep. The part has to be findable, the BOM has to be parsed, and the live stock has to be honest.
MRO360 reads the work order against the asset BOM, validates parts availability against the inventory module, and pre-positions the kit. The discipline of MRO inventory management software here is less about exotic AI and more about treating the storeroom as part of the work order, not a separate problem.
4. Reorder Point and Safety Stock Setting
The workflow. Every part gets a reorder trigger and a buffer designed to absorb demand variability during the replenishment window.
Where it breaks. Min/max levels were set in a workshop years ago and nobody has revisited them since. Critical spares get the same buffer logic as commodity consumables. The result is the worst of both outcomes, stockouts on parts that matter and overstock on parts that don't.
The overarching problem is that static reorder math punishes you twice. It ties up working capital on the parts you'll never need and exposes you on the parts you can't afford to miss. Both are silent costs until a tier-one asset goes down for a part that was technically "in policy."
The reorder math
Reorder Point = (Average Daily Usage × Lead Time) + Safety Stock. The formula is fine. The inputs are the problem. Usage drifts with production volume. Lead time drifts with supplier performance. Safety stock drifts with risk appetite. When the inputs are stale, the output is wrong in both directions.
The software-assisted fix. Dynamic reorder point calculation means the inputs are pulled live. Real consumption from the inventory module, real lead times from supplier history, and criticality-weighted safety stock that buffers tier-one spares aggressively and commodity spares conservatively.
MRO360 also closes the loop when a part drops below the threshold. The system surfaces the gap, recommends a procurement request, or suggests an inter-plant stock transfer when the part is sitting idle at another location. The math becomes honest because the data feeding it is live, and the action becomes systematic because the system doesn't wait for someone to notice.
5. Spare Parts Master Data and BOM Linkage
The workflow. Every part record is clean, deduplicated, correctly categorized and accurately linked to the assets it belongs to.
Where it breaks. The same bearing exists under four different descriptions, two of them misspelled. Three of those records show zero stock and one shows surplus, but the planner can't tell which one is real. The asset BOM in the EAM is incomplete, so even when the part is found, the planner can't be certain it fits.
The overarching problem is that every strategy above this one silently assumes clean data. Criticality scoring needs reliable part identity. Kitting needs accurate BOM linkage. Reorder math needs deduplicated consumption history. When the foundation is dirty, every downstream strategy inherits the rot, no matter how sophisticated the analytics on top.
The software-assisted fix. This is where the master data management for maintenance layer earns its place. Automated normalization deduplicates legacy records. Agentic enrichment fills missing attributes by parsing OEM catalogs and asset BOM documents. Semantic matching recognizes that “BRG 6205 ZZ” and “Bearing 6205-2Z deep groove” are the same physical part.
The MDM Suite’s Harmonize module cleans the legacy and the Integrity module governs every new record at the point of creation, so the foundation doesn’t decay again. MRO360 then handles the BOM linkage layer, autonomously parsing asset BOM documents to build accurate part-to-equipment relationships.
This is the foundation, not a bolt-on. Frame it as the layer that makes every strategy above and below it credible. There’s more depth on the underlying approach in our pieces on cómo limpiar los datos de las piezas de recambio y MRO data taxonomy.
Spreadsheet catalogs, tribal knowledge, duplicate records nobody owns. BOMs live in PDFs in shared drives.
ERP material master with rule-based dedup. Periodic cleansing projects. BOMs uploaded but rarely linked to live inventory.
Agentic enrichment, semantic matching, autonomous BOM parsing, persistent governance through Integrity workflows.
Walk through your current maintenance data, BOM linkage and inventory posture with our team. We will map your eleven strategies against the maturity model, against your own records, and show where MRO360 and the MDM Suite would move the needle.
Verdantis se enorgullece de ser el socio de confianza de organizaciones líderes en todo el mundo.
Desde empresas de Fortune 500 hasta pioneras del sector, nuestros clientes han confiado en nuestras soluciones MDM
Layer 3: The Execution Layer
Decision and planning hand off to execution. This is where the clean data, the kit, the criticality score and the predictive signal either come together at the asset, or they don't.
6. Work Order Scheduling
The workflow. Jobs are sequenced against three constraints at once: labor availability, parts readiness, and the asset or production window.
Where it breaks. Jobs get scheduled without confirming parts are in hand, or without an asset-availability window from operations. Crews show up to a running line that can't be taken down, or to a part that's still in transit. Planned and unplanned work collide and the planner reshuffles in real time.
The overarching problem is that scheduling is the most visible failure surface in maintenance. When a crew can't proceed, three teams know about it within the hour, and confidence in maintenance planning erodes across the plant.
The software-assisted fix. Constraint-aware scheduling that checks parts availability against the kitting layer, balances planned and unplanned demand, and aligns against live asset status. The discipline behind this is the same as in air-traffic control: the system doesn't release a job unless the three constraints clear simultaneously.
7. Predictive Triggering of Work Orders
The workflow. A condition signal from IIoT or SCADA creates a work order before the asset fails, not after.
Where it breaks. The sensor data is collected, sometimes at great expense, but it's disconnected from the maintenance and inventory workflow. A vibration anomaly raises a flag in the control de estado dashboard but doesn't create a work order, doesn't trigger a part pre-position, and doesn't update reorder math. The predicted failure happens, but no spare is staged.
The overarching problem is that predictive maintenance has a return-on-investment ceiling unless the signal closes the loop into inventory and execution. A correctly predicted failure with no part on the shelf is still a stockout.
Signal stops here
Sensor data flags an anomaly. A reliability engineer sees it on a dashboard. The maintenance system, the inventory system and the supplier portal don't.
Signal goes through
The anomaly creates a work order with a predicted failure mode, identifies the likely failure parts, and pre-positions them via the kitting and reorder layers. Execution is ready before the failure lands.
The software-assisted fix. Sensor data feeds a failure prediction model that names the likely failure parts. Those parts get staged via the kitting workflow and weighted into the reorder math. There's more on what the predictive maintenance data layer actually has to look like to make this practical, and on the mantenimiento preventivo frente a mantenimiento predictivo distinction that often gets blurred in practice.
This is one of the cleanest examples of AI agents and functional expertise working together. The model identifies the likely failure parts. The reliability engineer validates the prediction. The maintenance planner pre-positions the kit. None of them does the others' job.
Layer 4: The Optimization Layer
The first three layers run the operation. The fourth tightens it. These are the strategies that find the working capital trapped in your storeroom, the substitution opportunities buried in your catalog, the supplier patterns hiding in your purchase history, and the surplus sitting fifty miles from a stockout.
8. Dead Stock and Surplus Rationalization
The workflow. Inventory is periodically reviewed to identify parts that haven't moved and probably shouldn't be held.
Where it breaks. The review is occasional, often triggered by a CFO question rather than a process. Parts that haven't moved in years are still in the storeroom because nobody is sure if they're a critical spare, an obsolete model, or simply forgotten.
The 25 percent problem
Industry studies consistently put dead stock at around a quarter of total MRO inventory. That's working capital, storeroom space, insurance cost and audit risk on parts that will never be used. In a plant network running tens of millions of dollars of MRO inventory, the bill is significant.
The software-assisted fix. Continuous movement classification (fast, slow, dormant) flags surplus and obsolete spare parts as they drift, not at year end. The system recommends the right action: return to supplier, liquidate for salvage, or transfer to another plant that's actively consuming the part.
The overarching problem this solves is silent capital drag. Nobody notices a part that hasn't moved. The system has to be the one paying attention, with prompts that surface to the people who can act. The Análisis de gastos MRO layer is where this becomes a board-level conversation.
9. Spares Standardization and Substitution
The workflow. Interchangeable parts are collapsed into a standard record. Viable substitutes are recognized and stocked together so a stockout on one doesn't halt a job that another part can complete.
Where it breaks. The same physical part lives under several SKUs because three different OEMs supply it under three different codes, and a fourth code came from an acquisition. Functionally identical alternatives are sitting on the shelf but the planner can't see them because they're not linked.
The overarching problem is over-ordering driven by invisible inventory. The substitute is in stock. The system says it's not. The plant orders the original anyway, and pays for two parts that do the same job.
The software-assisted fix. Semantic matching and substitutability intelligence map functional equivalents across catalogs, OEMs and acquisition data. The model uses the same kind of clasificación de las piezas de recambio and taxonomy work that powers the MDM layer, but applied to interoperability rather than identity.
This is one of the strategies where the AI does work that a human practically cannot. No planner can hold thirty thousand part records and their cross-OEM equivalents in their head. The agent can.
10. Supplier Reliability and Lead-Time Management
The workflow. Supplier performance and real lead times are tracked, and that data feeds back into the reorder and safety-stock math.
Where it breaks. The reorder formula relies on the lead time printed in the supplier catalog or quoted on the original PO. Actual lead times drift, sometimes seasonally, sometimes structurally. The math runs on a number that hasn't been true for two years.
The overarching problem is buffer logic that's accurate on paper and wrong in practice. Either the safety stock is too thin and you stock out, or it's too fat and you carry inventory you don't need. Both are caused by trusting catalog lead times.
Assumed lead time
The number on the supplier catalog. Used in the reorder formula. Updated rarely. Doesn't reflect seasonal volatility, regional disruptions, or the fact that one supplier has been quietly slipping by ten days for six months.
Actual lead time
Computed live from PO history. Segmented by supplier, region and part category. Fed into the reorder math automatically so safety stock matches reality, not a brochure number.
The software-assisted fix. Live lead-time and reliability metrics are extracted from ERP transaction history and fed into the reorder engine. The buffer math becomes honest. The same data also surfaces in Estrategias de adquisición de MRO and supplier scorecards, so procurement decisions stop being driven by relationship and start being driven by performance.
The functional discipline is also worth naming. Supplier governance through the MDM layer (see supplier master data management) ensures the supplier record itself is clean, deduplicated and accurate before any performance data gets attached to it.
11. Multi-Plant Inventory Pooling and Stock Transfer
The workflow. The plant network is treated as a single warehouse. A part idle at one site covers a need at another, before a fresh purchase order is raised.
Where it breaks. Each plant hoards its own stock with no cross-site visibility. One plant raises a rush procurement request for a critical part that's been sitting unused at a sister plant fifty miles away for nine months. The order goes through, the part is duplicated, and the surplus problem at the sister plant gets worse.
The overarching problem is that the network behaves like a collection of silos when it should behave like a portfolio. Each plant optimizes locally. The network pays for it globally.
The software-assisted fix. Network-wide inventory visibility, with the part-identity layer clean enough (back to strategy 5) that the system actually knows the same part exists at two sites under different SKUs. When a need arises, the system surfaces the transfer opportunity before the procurement request is raised. This is one of the highest-leverage moves available in Gestión de inventarios MRO, and one of the simplest in concept once the master data foundation is in place.
The maturity ladder that runs through all eleven
Look at the eleven strategies together and a single pattern emerges. They aren't failing because they're the wrong strategies. They're failing because they're being executed at a maturity level the operation has outgrown.
The move from Manual to Systematized is the move from people to systems. The work that lived in someone's head moves into the CMMS, the ERP, the asset register. It scales. But it doesn't yet think.
The move from Systematized to Intelligent is the more interesting one, and the harder one. It's the shift from static logic to live decisions. From rules that were right two years ago to reasoning that recomputes against today's data. From annual workshops to continuous adjustment. From planners doing the heavy analytical lifting to planners doing the judgment work that actually needs a human.
That second move is where AI-native software and functional expertise stop being separate conversations. The agents do the work humans can't do at scale (scoring, classifying, parsing, matching across millions of records). The reliability engineers, planners and procurement leads do the work the agents can't do (context, override, alignment with business priority). When that balance holds, the eleven strategies stop being eleven separate maturity gaps and start behaving like one coherent operating system.
The honest starting point
Most operations don't need to advance all eleven strategies to Intelligent at once. The right starting point is usually identifying the two or three strategies still at Manual or stuck in static Systematized, fixing the master data underneath them, and letting the rest follow in sequence. The maturity climb matters more than the maturity headline.
