Every highway authority in the world maintains its road network using some combination of preventive and predictive maintenance — yet few have formally examined what the right balance looks like for their specific network, traffic profile, and budget constraints. Preventive maintenance applies treatments on a scheduled basis regardless of actual asset condition: resurface every 15 years, replace bridge bearings every 20 years, clear drainage every 12 months. Predictive maintenance applies treatments when sensor data, inspection analytics, or condition models indicate that an intervention is needed — not before, not after. The distinction matters enormously for highway operations budgets. A highway authority running a purely calendar-based preventive program is spending money on assets that don't yet need treatment while missing assets that do. A highway authority attempting to run purely predictive maintenance without the data infrastructure to support it is flying blind. The highest-performing highway maintenance programs in the world — those achieving the lowest cost per vehicle-kilometre maintained and the highest network condition scores — deploy preventive and predictive maintenance in a deliberate, evidence-based mix that matches the monitoring capability available for each asset class to the maintenance approach it can support. This guide provides the framework for making that determination. Schedule a free highways maintenance strategy review with our team and find out where your current program sits on the preventive-to-predictive spectrum.
Defining the Spectrum: From Reactive Through to Predictive
Preventive and predictive maintenance are not binary opposites — they sit at specific points on a continuous spectrum of maintenance maturity, each requiring progressively more data, analytical capability, and organisational sophistication to implement effectively.
Preventive Maintenance
Scheduled & Condition-Based
Preventive maintenance encompasses all maintenance activities carried out before failure occurs — on a time or usage schedule (calendar preventive) or when measured condition reaches a defined intervention threshold (condition-based preventive). In highways operations, preventive maintenance includes cyclical surface treatments, planned drainage clearance programmes, scheduled bridge bearing replacements, periodic joint and seal renewals, and road marking refresh programmes.
Best applied when
Asset failure has predictable time-based patterns, continuous monitoring is not available or cost-justified, or the treatment cost is low enough that treating healthy assets is acceptable.
Predictive Maintenance
Sensor and AI-Driven
Predictive maintenance uses continuous sensor data, AI pattern recognition, and deterioration modelling to forecast when and where an intervention will be needed — and triggers the maintenance work before the threshold is breached, not on a fixed schedule. In highways operations, predictive maintenance includes IoT-based pavement fatigue monitoring, AI-driven pothole precursor detection from vehicle probe data, bridge structural health monitoring with failure prediction, and slope movement early warning systems.
Best applied when
Continuous monitoring infrastructure exists, failure consequence is high enough to justify monitoring cost, and asset deterioration patterns are variable enough that fixed schedules create significant over- or under-treatment.
The Real Cost Difference: What the Numbers Show
The financial case for shifting the balance from preventive to predictive is well-evidenced in highway operations literature, but the magnitude of the difference is frequently underestimated by agencies that have never quantified their own preventive program's inefficiency. The following comparisons are based on documented outcomes from national highway authority deployments.
Maintenance Scenario
Calendar Preventive
Predictive (Sensor + AI)
Saving
Pavement resurfacing per lane-km — correct timing
£65,000–£90,000 (30% treated before optimal)
£48,000–£65,000 (treated at exact optimal point)
18–28%
Emergency pothole repair vs planned patching
£350–£900 per repair (reactive, traffic management, liability)
£45–£120 per repair (planned, no emergency call-out)
5–8×
Bridge inspection programme — fixed vs risk-based
£12,000–£28,000 per bridge biennial (all bridges same interval)
£4,000–£9,000 average (monitored bridges need fewer inspections)
40–65%
Drainage clearance — scheduled vs condition-triggered
100% of gullies cleared annually (many unnecessarily)
35–60% cleared based on sensor blockage data
40–65% reduction
Slope stabilisation — reactive failure vs early intervention
£800K–£5M per failure event (emergency stabilisation + road closure)
£80K–£350K (planned drainage + minor stabilisation)
85–93%
Find Out Where Your Highway Maintenance Budget Is Being Over- or Under-Spent
Oxmaint helps highway authorities quantify the efficiency gap in their current preventive maintenance programme, identify the highest-value predictive monitoring opportunities, and build the transition roadmap that shifts maintenance spend from calendar-driven to condition-driven.
Asset-by-Asset: Which Approach Fits Each Highway Asset
The preventive-to-predictive decision is made at the asset class level, not at the programme level. Some highway assets are best served by well-designed preventive schedules; others have such variable deterioration rates or such high failure consequences that predictive monitoring delivers clear superiority. The following framework maps each major highway asset class to its optimal maintenance approach.
Highway Asset
Optimal Approach
Rationale
Data Needed
Cost Impact
Bridge Structures (Major)
Predictive
Failure consequence very high; structural health varies unpredictably; sensor monitoring now cost-justified
SHM sensors, GNSS, crack gauges, traffic WIM
40–65% inspection cost saving
High-Volume Pavement (Motorway/A-road)
Predictive
High traffic loading creates variable deterioration; continuous monitoring enables optimal resurfacing timing
Embedded strain sensors, WIM, automated road surveys
18–28% per-treatment saving
Embankments and Cut Slopes (High-consequence)
Predictive
Failure is sudden and catastrophic; early warning window makes intervention 85–93% cheaper
Inclinometers, piezometers, GNSS, rainfall sensors
85–93% cost vs failure event
Road Tunnels
Hybrid
Safety systems (air quality, fire) require continuous monitoring; structural maintenance can be condition-based; M&E on calendar
Air quality sensors, DTS lining monitoring, CCTV AI
30–45% M&E cost saving
Medium-Volume Pavement (Secondary roads)
Hybrid
Continuous monitoring may not be cost-justified; condition survey-based scheduling more effective than pure calendar
Annual network survey data (SCANNER/TRACS)
10–20% saving vs pure calendar
Drainage Systems
Hybrid
Smart sensor gullies now enable condition-based clearance on priority routes; rural networks remain calendar-based
Ultrasonic level sensors, CCTV inspection data
40–65% clearance reduction
Road Markings and Safety Features
Preventive
Deterioration broadly predictable by traffic and weather; road marking AI inspection still emerging; calendar efficient
Retroreflectometer surveys (periodic)
Well-optimised at current maturity
Low-Volume Rural Roads
Preventive
Monitoring cost exceeds benefit; traffic loading low and consistent; simple visual inspection cycles cost-effective
Visual inspection records
Optimised for low-cost management
Five Decisions That Determine Your Programme's Balance
Moving from a primarily preventive to a more predictive programme requires deliberate decisions about five interdependent factors. These decisions cannot be made independently — the right answer for each depends on the others.
1
Asset Criticality Classification
Foundation Decision
The failure consequence of each highway asset class determines whether the cost of monitoring is justified. A motorway bridge failure affecting 80,000 vehicles per day with a closure cost of £1M/week justifies significant monitoring investment. A rural road culvert failure affecting 200 vehicles per day with a detour of 3 kilometres does not — the monitoring cost exceeds the avoidable failure cost. Asset criticality classification based on traffic volume, consequence of failure, detour length, and connectivity value is the first decision that shapes every subsequent programme choice.
Risk-ranked asset register
Monitoring investment thresholds
Priority tier classification
2
Deterioration Variability Assessment
Economics Decision
Calendar preventive maintenance delivers its worst value-for-money on assets whose deterioration rate varies significantly across the network — where some sections deteriorate rapidly and others remain in good condition far beyond the scheduled treatment interval. Quantifying this variability from historical condition survey data reveals how much treatment is being applied to assets that don't yet need it (wasted spend) and how many assets are deteriorating past optimal treatment point before the scheduled date arrives (accelerated deterioration cost). Assets with high deterioration variability have the most to gain from predictive approaches.
Deterioration rate distribution by asset class
Over-treatment quantification
Predictive opportunity sizing
3
Data Infrastructure Readiness
Capability Decision
Predictive maintenance is only as good as the data that feeds it. A highway authority with no continuous monitoring infrastructure, no standardised condition survey program, and no CMMS capable of ingesting sensor data cannot run a predictive programme regardless of how compelling the financial case is. Data infrastructure readiness assessment — covering sensor networks, connectivity, data storage, analytics capability, and CMMS integration — determines which predictive approaches are immediately deployable and which require infrastructure investment before they can deliver value.
Current data capability assessment
Infrastructure investment requirements
Phased deployment plan
4
Budget Structure and Planning Horizon
Financial Decision
Predictive maintenance requires upfront capital investment in monitoring infrastructure that is recovered through reduced maintenance costs over a 3–7 year period. Government highway agencies with annual budget cycles and no multi-year capital commitment face a structural barrier to predictive programmes — the investment and the savings appear in different financial years under different budget heads. The business case for predictive maintenance must be structured to match the agency's planning horizon and budget classification requirements, demonstrating the lifecycle cost reduction across the relevant multi-year programme period.
Lifecycle cost comparison
Payback period calculation
Multi-year budget case
5
Organisational Capability and Change Management
People Decision
A predictive maintenance programme that generates AI-derived maintenance recommendations but is operated by a team that does not trust or understand the model outputs will revert to calendar scheduling within 18 months. Transitioning from preventive to predictive requires deliberate change management — training maintenance planners on condition-based decision frameworks, establishing governance for how predictive alerts override scheduled maintenance, and demonstrating early successes that build confidence in data-driven approaches. The organisational capability gap is usually the binding constraint, not the technology.
Capability gap assessment
Training programme
Change management plan
Oxmaint Supports Every Point on the Preventive-to-Predictive Spectrum
Whether your programme is primarily calendar-based and moving toward condition-based, or already deploying IoT sensors and needing the management platform to make predictive outputs operational, Oxmaint provides the maintenance management infrastructure that makes every approach more effective.
Transition Roadmap: Shifting the Balance in Your Highway Programme
Transitioning from a primarily preventive to a more predictive maintenance programme does not happen by replacing one programme with another — it happens by adding predictive capability to the highest-consequence asset classes while maintaining well-optimised preventive schedules for assets where prediction delivers no meaningful advantage.
Immediate (Months 1–6)
Baseline and Opportunity Quantification
Audit current preventive schedule — what percentage of treatments are being applied to assets that don't need them (over-treatment ratio)
Quantify reactive maintenance cost baseline — emergency repairs, unplanned closures, compensation claims
Rank asset classes by criticality and deterioration variability — identify the highest-return predictive monitoring opportunities
Assess existing data and monitoring infrastructure — what predictive capability can be built without new sensor investment
Short Term (Months 6–18)
Pilot Predictive Monitoring on Priority Assets
Deploy IoT monitoring on the 3–5 highest-consequence, highest-variability asset locations identified in the baseline phase
Integrate existing condition survey data into CMMS to enable condition-based scheduling — reducing over-treatment on assets with measurable condition data
Establish the governance framework for condition-driven decisions — how sensor alerts and survey data override calendar schedules
Measure and document first-year savings against baseline to build the business case for programme expansion
Medium Term (Year 2–3)
Network Expansion and Programme Integration
Expand IoT monitoring to the broader priority network — all high-consequence bridges, strategic route embankments, and motorway pavement sections
Replace calendar-based PM schedules with condition-based intervals for all asset classes with condition survey coverage
Integrate predictive analytics outputs with capital programme planning — 5-year condition forecasts replacing age-based renewal assumptions
Retain well-optimised calendar PM for asset classes where data confirms prediction adds no meaningful advantage
Programme Maturity (Year 3+)
Optimised Hybrid Programme
Majority of maintenance spend on high-consequence assets driven by condition data and predictive models — reactive maintenance below 15% of total
Calendar PM retained only for low-consequence, low-variability assets where it remains the most cost-effective approach
Network condition improving year-on-year despite stable or declining maintenance budget — demonstrable to treasury and public accounts committees
Continuous model improvement as accumulated data increases predictive accuracy — self-improving programme that compounds savings over time
Programme Performance KPIs
A highway maintenance programme that is transitioning from preventive to predictive must track KPIs that reflect both approaches — measuring the effectiveness of remaining preventive schedules while demonstrating the growing contribution of predictive capability to overall programme performance.
Preventive Maintenance KPIs
Treatment Over-Rate
Target: < 10%
Percentage of scheduled treatments applied to assets measured in better than "needs treatment" condition — the direct measure of calendar PM inefficiency
Schedule Adherence
Target: > 95%
Preventive maintenance completed on schedule — a necessary condition for PM to deliver its intended deterioration prevention effect
Assets Reaching Critical Before Scheduled Treatment
Target: < 2%
Network segments that deteriorate past intervention threshold before scheduled PM date — indicates PM intervals are too long for current deterioration rates
Predictive Maintenance KPIs
Predictive Alert Lead Time
Target: > 4 weeks average
Average advance warning between predictive alert and the point at which failure or emergency action would have been required — the planning window created by prediction
Emergency Events on Monitored Assets
Target: < 5% of events
Reactive emergency maintenance events on assets with active predictive monitoring — events above this threshold indicate model performance or alert response gaps
Predictive-to-Total Maintenance Ratio
Target: Growing year-on-year
Share of total maintenance work triggered by predictive analytics versus calendar or reactive — the primary programme maturity indicator for reporting to government oversight bodies
Shared Programme Metrics
Network Condition Index
Improving trend
Overall network condition trajectory — the headline public performance metric
Reactive Maintenance %
< 15%
Share of total budget spent on reactive work — falling ratio demonstrates programme improvement
Cost per Lane-km Maintained
Declining trend
Maintenance cost normalised for network size and traffic loading — the value-for-money metric
Emergency Events Reduction
35–50% Year 1
Reduction versus baseline — the primary short-term financial return metric
A Single Platform for Every Maintenance Approach on Your Network
Oxmaint manages preventive schedules, condition-based work order generation, IoT-triggered predictive alerts, and reactive response — all in one maintenance management platform that gives highway authorities a complete, auditable picture of every intervention across every asset class and every maintenance approach in use.
Frequently Asked Questions
01
How much of a typical highway maintenance budget is wasted on preventive treatments applied to assets that don't need them?
Studies of national highway maintenance programmes consistently find that 20–35% of calendar-based preventive treatments are applied to assets that, if condition-surveyed at the point of treatment, would not yet meet the intervention threshold. This "over-treatment" rate represents direct budget inefficiency — not waste in the sense that the treatment does no good, but waste in the sense that the same money applied to assets that actually need treatment would deliver more network condition improvement per pound spent. The over-treatment rate is higher for assets with consistent condition surveys that reveal the disparity, and it is usually highest for drainage clearance (where a significant proportion of gullies are clear when cleared) and lowest for pavement resurfacing on high-traffic routes where deterioration is more predictable. For a highway authority spending £50M annually on preventive maintenance, a 25% over-treatment rate represents approximately £12.5M of maintenance spend that could be redirected to higher-priority needs through a condition-based programme.
02
Is predictive maintenance always superior to preventive for highway operations?
No — and any approach that claims it is should be viewed with scepticism. Predictive maintenance is superior when: the monitoring cost is less than the over-treatment cost it eliminates; the asset's deterioration rate varies enough across the network that fixed schedules create significant inefficiency; and the consequence of failure is high enough that early warning delivers meaningful cost avoidance. For assets that fail at predictable rates, where monitoring infrastructure would cost more than the over-treatment it prevents, or where the failure consequence is low enough that reactive response is actually the most cost-effective approach, preventive or reactive maintenance remains the right choice. The goal of a sophisticated highway maintenance programme is not to maximise the proportion of predictive maintenance — it is to apply the most cost-effective approach to each asset class. For most highway networks, the optimal programme has predictive monitoring for the top 10–15% of assets by consequence, condition-based preventive scheduling for another 40–50%, and calendar-based preventive programmes for the remainder.
03
What is the typical payback period on IoT monitoring investment for highway infrastructure?
Payback periods vary significantly by asset type and the specific failure modes being monitored. For bridge structural health monitoring, where a single prevented emergency closure or load restriction event can cost more than the entire sensor system, payback periods of 6–18 months are achievable — particularly on high-traffic structures where closure costs are highest. For pavement monitoring that enables better resurfacing timing, payback periods of 2–4 years are typical, with the savings accumulating through reduced over-treatment and slightly fewer emergency pothole events. For slope monitoring, the payback calculation is dominated by the catastrophic failure scenario — a single prevented slope failure on a strategic route that would have cost £2M–£5M in emergency stabilisation and road closure pays back a monitoring system that costs £100K–£300K to install. Government highway authorities making the investment case to treasury should present the payback calculation for each monitored asset class separately, as the range is wide and the strongest cases are often on the assets with the highest consequence of failure rather than the highest volume of routine maintenance spend.
04
How does a CMMS support both preventive and predictive maintenance in a highway authority?
A maintenance management system supports the hybrid preventive-predictive programme that best-practice highway authorities operate by handling both approaches within a single unified platform. For preventive maintenance, it manages the schedule library — all recurring treatments with their intervals, trigger conditions, resource requirements, and contractor assignments — and generates work orders automatically when treatment dates or condition thresholds are reached, tracking completion and updating the next scheduled date. For predictive maintenance, it receives alert outputs from IoT sensor networks and analytics platforms, converts them into structured work orders with supporting sensor evidence, and manages the engineer review and approval workflow before maintenance teams are dispatched. Critically, it provides the unified asset view that makes the two approaches interact correctly — a predictive alert for a bridge bearing replacement should automatically suppress or reschedule the next calendar PM inspection for that bearing, avoiding duplicate work orders for the same intervention. And it generates the programme-level reporting that highway authorities need for public accountability: the reactive-to-planned ratio, the over-treatment rate, the predictive-to-total maintenance ratio, and the network condition trends that demonstrate the combined effect of both approaches on network performance.
