Sending the wrong technician to a job doesn't look like a failed dispatch decision. It looks like a second visit. A return callout. A parts run that takes the morning. An SLA breach logged quietly in a report that nobody reviews until the contract renewal.
The cost of a misassignment rarely appears as a single line item. It distributes itself across a callback, a penalty clause, an unhappy site contact, and a technician who spent two hours travelling to a fault he couldn't fix. For a company running 30 technicians across 500 maintenance contracts, that kind of friction compounds daily.
Dispatch optimisation is the process of getting the assignment decision right the first time, every time. This article covers why that decision is harder than it looks, what happens when you get it wrong, and how field service companies are using AI-assisted dispatch to handle multiple variables simultaneously without adding headcount to the control room.
Why dispatch is harder than finding the nearest technician
The instinctive solution to dispatch is proximity: find the closest available engineer and send them. Proximity-first dispatch works when you have one type of asset, one certification requirement, and unlimited time. Most field service operations have none of those.
Consider what actually needs to line up before a technician can successfully complete a job:
The technician has the relevant certification. An F-Gas Category I engineer can handle fluorinated refrigerants on a chiller; an uncertified technician cannot legally touch the refrigerant circuit. An IPAF-certificated engineer can work at height on a lift shaft; someone without it cannot. In the UK, DBS-checked engineers are required on school and hospital contracts. Certifications expire. The nearest available technician and the nearest certified technician are often different people.
The technician knows the specific asset type. Theoretical qualifications and operational familiarity are not the same thing. A technician who has maintained 200 Schindler 3300 elevators will diagnose a door operator fault in 20 minutes. A technician who last touched a Schindler system three years ago may take two hours to reach the same conclusion. First-visit-fix rates vary by 30–50% between technicians based on asset familiarity, not just formal certification.
The van has the right parts. Elevator door operators, fire alarm sounder bases, refrigerant recovery cylinders, pressure transducers: the parts list for any given fault type is predictable enough to pre-stock. The question is whether the technician heading to the job has them on board. A van stocked for fire alarm work is not stocked for elevator work. Sending an elevator tech whose van ran out of door drive belts last Tuesday means a holding visit, a parts run, and a second appointment.
The technician has the time. Available does not mean available for four hours. A technician with two hours of remaining capacity cannot take on a compressor replacement that will run until 7pm. Dispatch that doesn't model current workload creates schedule collisions that cascade through the afternoon.
The SLA makes the decision for you, sometimes. A trapped passenger emergency with a 90-minute SLA window doesn't allow time for the optimal technician to drive 45 minutes. The fastest available qualified engineer gets the job, even if someone better is across town. An overdue-contract PPM visit, on the other hand, can tolerate a short scheduling delay to match the right technician.
A dispatcher doing this manually works through a mental checklist: available, certified, stocked, close enough, has capacity. On a good day with a familiar team, this takes 90 seconds. On a Monday morning with 14 open jobs, two sick calls, and an emergency in the queue, it takes longer, and mistakes happen.
The real cost of sending the wrong technician
The visible cost is the return visit. A technician travels to site, diagnoses the fault, determines he can't fix it with what's on the van, and books a follow-up. Direct cost: two hours of travel and wasted labour, one maintenance visit charge waived as goodwill, one SLA clock still running.
The less visible costs accumulate in three directions.
SLA exposure. If the initial visit doesn't resolve the fault, the resolution SLA keeps running. The original technician is already committed to his next job. The resolution now depends on getting a second technician to site before the clock expires. Depending on the fault priority and contract tier, this triggers either a penalty clause or a documented breach that counts against your annual compliance rate. Some contracts allow clients to terminate without notice if the resolution SLA breach rate exceeds a defined threshold in any 90-day period.
Customer confidence. Sites with high-stakes assets, a data centre, a pharmaceutical cold store, a high-rise elevator in a residential block, have very low tolerance for return visits. A building manager who has reported a fault twice already will start asking questions about the competence of the maintenance company. In multi-site FM contracts, those questions reach the contract owner and appear in annual review meetings.
Technician productivity. A technician dispatched to a job he can't complete wastes his time and, because he's now behind schedule, pushes back every other job in his diary. If he was the available backup for a second emergency that afternoon, his capacity is gone. Mis-dispatch doesn't only affect the immediate job; it affects everything downstream.
Manual dispatch: what it can and can't handle
Experienced dispatchers are genuinely good at this. A control room team that has worked together for five years develops shared mental models of the technician pool: who's good on Otis equipment, who needs longer on complicated faults, whose van is always stocked for refrigeration work. This institutional knowledge is real and valuable.
It doesn't scale. And it doesn't work at 3am.
When the experienced dispatcher is off sick, when a new hire is covering the shift, when volume spikes to 20 simultaneous open jobs on a Monday after a long weekend, the mental model breaks down. The new dispatcher doesn't know that Technician 7 is the specialist for Kone MRL elevators. He assigns Technician 3 because Technician 3 is geographically closest and picks up his phone. The result is a 3-hour diagnosis instead of a 45-minute fix.
Manual dispatch also can't track van inventory in real time. A dispatcher who assigns a refrigeration callout based on which technician is nearest doesn't automatically know that the nearest technician used his last R-410A recovery cylinder this morning. That information exists somewhere, on a van stock form, in a WhatsApp message, in the technician's head. It isn't in front of the dispatcher when the assignment decision is made.
The result is a gap between the information the dispatcher has and the information the decision requires. In straightforward situations, experienced dispatchers bridge that gap from memory. In complex, high-pressure, or out-of-hours situations, they can't.
What AI-assisted dispatch actually considers
Smart dispatch builds a decision model from the inputs that matter, simultaneously, without the dispatcher having to check them one by one.
Location and travel time. Real-time GPS position of every technician, estimated arrival time based on current traffic (not straight-line distance), factored against the SLA deadline on the incoming job.
Certification matrix. Every technician's active certifications: F-Gas category, IPAF, PASMA, EN 81-28 competency, DBS check, first aid, any contract-specific requirements. Certificates with upcoming expiry flagged so they don't factor into assignments after expiry.
Van inventory. Parts likely needed for the incoming fault type cross-referenced against the stock in each technician's van. A chiller callout needs a refrigerant recovery kit, pressure gauges, and common seal/gasket sizes. A fire panel fault needs panel-specific replacement devices. The system checks which vans carry the right stock before ranking technicians.
Current schedule and remaining capacity. Estimated remaining time on the current job, jobs already booked for later in the day, contracted end-of-shift time. A technician with 45 minutes of remaining capacity and three jobs in the afternoon doesn't get a 3-hour job dropped on them unless there is no alternative.
Historical first-visit-fix rate. Which technicians have the highest success rate on this asset type or fault category. A technician with a 91% first-visit-fix rate on Otis elevator door operators will fix the fault and leave. A technician with a 55% rate on the same fault type will, statistically, require a second visit or parts run around half the time. This is schedulable intelligence.
SLA urgency and job priority. Emergency jobs override normal scheduling logic. A trapped passenger callout with a 90-minute SLA doesn't wait for the optimal technician if the nearest qualified engineer can make the window. The urgency tier of the job determines whether the system optimises for best match or fastest response.
The output is a ranked shortlist of technicians for the dispatcher to review, not an automatic assignment that bypasses human judgment. The dispatcher sees why each technician was ranked where they were: certification, proximity, current workload, relevant parts on board, historical performance. If they know something the system doesn't, a technician who rang in with a slow puncture, a site access note that isn't logged, they override it. The override is recorded.
Two dispatch scenarios compared
Refrigeration emergency: hospital cold store, 4-hour SLA
A pharmaceutical storage facility at a hospital calls in at 6:45am. The chiller on Ward 4's blood bank supply circuit is alarming. The SLA is 4 hours from first contact to technician on site.
Manual dispatch: the night dispatcher checks a printed on-call rota. The primary on-call engineer is logged as available. She's checked against the rota, but not against whether her van is stocked for refrigeration work after she used most of her refrigerant recovery equipment on a callout yesterday. She's dispatched. She arrives in 90 minutes, assesses the fault, and determines she needs a pressure transducer and recovery equipment that isn't on her van. A parts run to the depot adds 2.5 hours. Resolution time: 5.5 hours. SLA breach: 1.5 hours past the 4-hour window. Penalty clause invoked.
AI-assisted dispatch: the system identifies the incoming call as a chiller asset at the hospital site. It flags the 4-hour SLA, checks van inventory against the typical parts profile for chiller callouts, and identifies that the primary on-call engineer's van is low on key equipment. It surfaces her as a second-ranked option and places a technician with a fully stocked refrigeration van and a 45-minute travel time first. The dispatcher confirms the recommendation. The technician arrives in 47 minutes, carries out the repair with on-board parts, and clears the site by 10:20am. SLA met.
Elevator fault: multi-storey residential block, no trapped passenger
A building manager reports that a Schindler 5500 MRL elevator is showing an error code and not responding to calls. No passenger is trapped. The fault priority is medium; the SLA is 8 hours response.
Manual dispatch: the dispatcher assigns the nearest available technician. He arrives at the building in 25 minutes, opens the motor room, and realises the fault code is specific to the Schindler 5500 ACVF controller, which uses a proprietary diagnostic interface he doesn't have on his tablet. He calls the office, gets the diagnostic manually relayed, spends 90 minutes troubleshooting by phone, orders a controller board on next-day delivery, and closes the visit as "parts pending." Total on-site time: 2 hours. Resolution: next day after parts delivery.
AI-assisted dispatch: the system identifies the asset as a Schindler 5500 MRL from the asset record and surfaces technicians with documented Schindler MRL experience. The top-ranked technician is 18 minutes further away but has completed 22 jobs on Schindler 5500 systems in the past 18 months and carries a Schindler diagnostic tablet. He arrives in 43 minutes, runs the diagnostic in 12 minutes, identifies a door drive motor bearing fault, replaces the motor from van stock (a common Schindler 5500 failure he carries parts for), and clears the fault in 90 minutes total. Elevator returned to service, no return visit.
Emergency priority and automatic escalation
Emergency jobs, trapped passengers, fire suppression system faults, critical cold chain failures, require a different dispatch logic. The ranking of the best match gives way to the fastest available qualified response.
When an emergency lands, the system calculates the SLA deadline, identifies every technician who meets the minimum qualification threshold for the job, ranks by projected arrival time against the SLA window, and surfaces the top candidates. If the nearest qualified technician is 75 minutes away and the SLA window is 90 minutes, the dispatcher can see in real time that the assignment is tight.
If no technician can meet the SLA window, the system escalates automatically. The operations manager receives a notification before the breach, not after, with the current status: nearest qualified technician, projected arrival time, SLA deadline, and the gap. That gives the manager a window to decide: call in an out-of-area engineer, trigger a mutual aid arrangement with a partner company, call the client with a proactive update. None of those options are available if the first notification comes after the SLA has expired.
The escalation chain runs without anyone watching the clock. A dispatcher who has eight other jobs open at the same time cannot be expected to manually track the 90-minute window on the emergency while managing the rest of the board.
Dispatch optimisation and the first-visit-fix rate
First-visit-fix rate is the headline metric for dispatch quality. Industry benchmarks vary by vertical: elevator maintenance typically sees rates of 65–80% for non-parts-dependent faults; HVAC emergency repair runs 55–75%; fire alarm fault resolution is higher, typically 80–90%, because panel faults are more consistently diagnosable.
The gap between the bottom and top performers in each vertical is almost entirely explained by two factors: technician-to-asset-type matching and parts availability. Both are dispatch decisions.
A company that tracks first-visit-fix rate by technician and by asset type has actionable intelligence. If Technician 12 has a 45% first-visit-fix rate on access control faults while Technician 4 has an 88% rate on the same fault type, that's a dispatch routing rule waiting to be configured. It's also a training signal.
Without tracking, the pattern is invisible. Dispatch decisions that consistently route the wrong engineer to the wrong fault type generate return visits that look like normal operations. The cost is absorbed as overhead rather than identified as a fixable dispatch problem.
What to look for in a dispatch optimisation system
Not every FSM platform that uses the word "smart dispatch" is actually doing multi-variable optimisation. The distinctions that matter operationally:
Asset-level job matching, not just site-level. A system that knows a Schindler 5500 MRL elevator is at the site is more useful than a system that knows the client's address. The asset type determines the qualification requirement, the likely parts list, and the relevant first-fix history.
Live van inventory, not just technician location. Proximity without parts availability is incomplete. A dispatcher who can see both makes a better decision than one who can only see a dot on a map.
Certification expiry tracking, built in. Certifications that have expired or are about to expire should automatically drop technicians out of eligible pools for jobs requiring those certifications. This shouldn't require a manager to manually maintain a separate register.
Explainable rankings. When the system suggests Technician A over Technician B, the reasoning should be visible: certification, travel time, van stock, SLA urgency, historical performance. Recommendations that can't be interrogated can't be trusted, and can't be overridden intelligently.
Escalation that runs without a dispatcher watching. Pre-breach alerts, automatic escalation to the operations manager, and projected SLA breach warnings should fire automatically. Escalation that requires a human to initiate it fails when the human is busy.
How RemoteOps handles dispatch
RemoteOps identifies each incoming job at asset level. For inbound calls from elevator intercom units, fire panel dialers, HVAC monitoring modules, or access control systems, the platform maps the CLI to a specific asset record before the dispatcher picks up. The asset type, service contract, SLA tier, and maintenance history are on screen before the conversation starts.
Dispatch suggestions are generated from the technician pool based on live location, certification validity, van inventory levels, current workload, and historical performance data for the fault type. The suggestions are ranked and explained. The dispatcher makes the assignment and can override with a recorded reason.
Escalation chains run automatically. If no technician acknowledges an emergency within 10 minutes, the next contact in the escalation chain receives a notification. If the SLA window is under 25% remaining and no technician is confirmed en route, the operations manager is alerted before the breach.
FAQ
What is dispatch optimisation in field service management?
Dispatch optimisation is the process of matching incoming jobs to technicians based on multiple criteria simultaneously: skills and certifications, current location and travel time, van inventory, available capacity, SLA urgency, and historical performance on the relevant asset type. The goal is to assign a technician who can complete the job on the first visit, within the contracted time.
What is the main cause of failed first visits in field service?
Parts and skills. A technician dispatched without the right certification cannot legally complete certain regulated tasks. A technician dispatched without the right parts on board must either attempt a workaround or book a return visit. Both situations are primarily dispatch failures, not technician failures, because the information needed to make the correct assignment was available before the job was allocated.
How does AI-assisted dispatch differ from geographic dispatch?
Geographic dispatch assigns the nearest available technician. AI-assisted dispatch assigns the technician with the best probability of completing the job successfully on the first visit, within the SLA window. Proximity is one input among several. For jobs where the wrong technician would generate a return visit, the optimal assignment is often not the geographically nearest one.
What happens when no qualified technician can meet the SLA window?
In a properly configured system, this triggers an automatic escalation before the breach occurs, not after. The operations manager is notified with the full picture: the job, the SLA deadline, the nearest available qualified technician, and the projected arrival gap. That gives the company a window to make an active decision, whether to deploy from further away, invoke a mutual aid arrangement, or proactively notify the client, rather than discovering the breach after the fact.
Does dispatch optimisation require a large technician team to be worthwhile?
No. The value is proportional to the complexity of the jobs, not the headcount. A company with 5 technicians covering 3 different asset types (elevator, fire, HVAC) still benefits from certification tracking, van inventory checks, and SLA-priority sorting. The manual overhead of tracking those variables correctly grows with the number of variables, not just the number of technicians.
Internal links
- SLA Management for Maintenance Companies: how SLA clocks work, escalation chains, and penalty clause mechanics
- How AI Is Changing Field Service Management: the full AI dispatch model and phone-to-asset resolution
- Field Service Management Software Buyer's Guide: evaluating FSM platforms for critical infrastructure
- Field Service Inventory Management: van stock, parts availability, and how inventory connects to dispatch quality