This post in 30 seconds.
- The idea: triage on what the reporter shows, not on what they typed. The occupant or site contact records the fault at the moment of reporting, from a link in the ticket or a QR code on site, and the helpdesk classifies it as urgent, planned, contractor or not ours from that footage.
- The number: a failed first visit adds two more visits and 14 days to resolution, according to Aquant's 2025 field service benchmark. In facilities, a good share of them begin with a ticket that says "not working".
- The honest limit: video replaces the first look, not the survey for a large job. It makes the visits that remain better informed and gives the contractor something to quote from.
- Written for: facilities managers and FM helpdesk leads running reactive maintenance across several sites with a mix of in-house trades and contractors.
Three sites, one helpdesk, Monday morning. A ticket from the north building says "heating not working, 2nd floor". One from the depot says "water on the floor by the loading bay". One from the office says "door won't lock". None of them has a photo. All three will get a visit, and at least one of those visits will end with someone turning a thermostat back on.
That is what facilities maintenance request triage is trying to prevent, and it usually cannot, because it runs on words. The desk reads a sentence, guesses a trade, guesses a priority, and sends someone to go and look.
This post walks through the alternative: the person who reported the fault shows it on their own phone, guided step by step, and the desk decides from what it can see. It covers the four triage outcomes and the evidence each one needs, how the capture reaches the reporter, how the case routes to an in-house team or a contractor, how the contractor quotes and closes on the same file, and how one flow serves every site.
If you run the helpdesk for more than one site, you already know the cost of a "go and look" visit that turns into a five-minute reset, and the contractor quote that arrives a week after the survey. We build Venta Capture, a product of VentaVid, so that the occupant can show you the fault before anyone is sent. Start for free and send one fault-report link to your own phone before you read the rest.
In this post:
What triage is deciding, and from what
Every guide to helpdesk triage lists the same three decisions: what kind of fault this is, how urgent it is, and who should get it. Facilities adds a fourth that the guides tend to skip. Whose job is it? The landlord's, the tenant's, the contractor under warranty, or yours.
All four decisions are only as good as the inputs, and today the inputs are a sentence typed by someone who is not a tradesperson. A recent piece in Facilities Management Advisor describes AI triage as classification, priority assignment and routing performed on the work order request, and notes that 60 to 75% of requests fall into a small number of repeating categories. That is useful. It also means the ceiling on automated triage is the reporter's vocabulary, because the model is reading the same words your desk reads.
The pressure to get this right is not easing. IFMA's FM Market Pulse report for October to December 2025, drawn from more than 1,400 responses across 80 countries, opened its Facility Management Workload Index at +43, meaning far more facility professionals expect workload to rise than to fall. Only 24% of respondents expected a net increase in headcount. More requests, the same desk.
So the question for a helpdesk lead is what the desk is triaging from. Remote triage from a description is a guess with a priority code on it. Remote triage from footage of the fault is a decision.
Why a typed description sends someone to look
Take "heating not working, 2nd floor". It could be a thermostat set to off, a radiator valve stuck, a boiler in lockout, or a building management schedule that nobody updated after the clocks changed. Two of those are a phone call. One is an in-house visit. One is a contractor. The sentence cannot tell you which.
A typed description fails triage in three ways, and each one costs a visit. The usual three:
- Wrong priority. "Water on the floor" is a dripping tap or a burst pipe. The desk either sends an emergency call-out for a mop job or books a routine slot for something that is soaking a ceiling below.
- Wrong trade. "Door won't lock" goes to the carpenter when it is a failed access-control reader, or to the security contractor when the strike plate has dropped. Either way the first person on site cannot fix it.
- Wrong responsibility. The fault sits with the tenant's fit-out contractor, or under a manufacturer's warranty, and nobody finds that out until a technician is standing in front of it.
The field service numbers on this are blunt. Aquant's 2025 Field Service Benchmark Report, based on around 160 service organisations and more than 600,000 technician records, found that a failed first visit adds two more visits on average and stretches resolution by 14 days, and that 14% of truck rolls were unnecessary altogether. That is equipment service data rather than facilities data, but a truck roll to a plant room costs what it costs whoever sends it.
One of the better guides on this topic makes a point worth keeping: do not make the requester diagnose the fault. Record the symptom. The trouble is that a text field is a poor symptom record, because the occupant has to translate what they see into words, and the desk has to translate it back. A 20-second video with the occupant narrating what they are pointing at skips both translations. It is the reason we wrote about reducing truck rolls with guided capture and the pre-visit assessment as separate mechanics; this post is about applying them to the FM helpdesk.
The four buckets and the evidence each one needs
Triage has four outcomes, and each needs a different minimum set of evidence, which is why "attach a photo" is not enough. The occupant does not know which bucket they are in. The flow has to ask.
A few notes on the table.
- Urgent is decided by what the video shows around the fault, not by the word "urgent" in the ticket. Running water, exposed conductors, a fire door that will not close, a lift with someone in it. The reactive maintenance clock starts now and the case should say who was told, and when.
- Planned is the bucket that saves the most visits, because a working-but-faulty asset with a clear video can wait for the next planned maintenance round instead of generating its own call-out.
- Contractor is where the money leaks, and it gets its own section below.
- Not ours still costs a visit today, because nobody can tell from a sentence that the leak is coming from the tenant's coffee machine plumbed in by their own fitter.
There is also a legal reason to fix the evidence at the point of report. Under Awaab's Law, which applies to social landlords in England from 27 October 2025, the government's guidance on repair timeframes is explicit that obligations begin when the landlord becomes aware of a potential hazard: emergency hazards must be investigated and made safe within 24 hours of that moment, and significant damp and mould hazards investigated within 10 working days. Day zero is the report. A timestamped capture from the tenant is the first item of evidence that the clock was met, which we covered in detail for the damp and mould reporting process. Commercial FM has service level agreements instead of statute, and the logic is the same: the minimum evidence set has to exist from minute one.
Capturing the fault at the moment of reporting
The reporter is already standing at the fault. The only question is how the request to show it reaches them.
- From the ticket. The helpdesk raises the job as usual, then sends a secure, personal capture link by email, SMS or WhatsApp to the person who reported it. For a portal, the link sits on the confirmation screen.
- From the site. A QR code where faults happen: the plant room door, the WC, the meeting room panel, the loading bay, the kitchen. Scanning it opens the fault-report flow for that location with the site already filled in. The code works on screen too, in a tenant portal or on a service display.
Either way the occupant opens a page in their phone's browser. No app, no account, in their own language. If someone opens it on a desktop they see a QR code to continue on their phone, because the recording has to happen on the device in their hand.
The flow is short, and the desk wrote it, which is what turns a video into a triage input. A fault-report flow for a multi-site estate typically runs:
- Where. Site and area, pre-filled from the QR code or picked from a list.
- What. Scan the asset label if there is one (QR or barcode, typed as fallback), so the case lands against the right boiler, lift or door.
- Show it. A wide shot of the area, then a short video close up with an on-screen instruction: point at the fault and say out loud what you see and when it started.
- Ask. Yes or no questions that steer the next step. Is water still coming out? Is there power to the unit? Is anyone unable to use the space? Have you put a temporary control in place?
- Who. Contact details last, so nobody abandons the flow on a form before they have shown the fault.
What arrives at the desk is a case: the footage, the answers as fields, the spoken description as searchable text via transcription, the timestamps, and your own ticket number in the reference field. The desk can copy the case straight into the CAFM ticket, or share a view link with someone who has no login. The mechanics of guided video capture and requesting video by link are the same ones used for guided photo capture in other sectors; the facilities-specific part is the questions.
If you want to see what one of these looks like on your own phone, start for free, open the fault-report template and send the link to yourself. It takes less time than the next "go and look" visit.
Routing to an in-house team or a contractor
Once the case is in the shared inbox, the desk does the four decisions in minutes, from the footage, and records them as fields: a status, an owner, and a route.
- In-house. Assign to the trade or the site engineer. They watch the 20 seconds before they leave, bring the right part, and skip the diagnostic walk.
- Contractor. Send the contractor a view link. They see the fault, the surroundings, the access route and the model plate without a login and without a survey visit for the standard job.
- Escalate. Anything the desk cannot place goes to a supervisor with a reason attached, through the same escalation path the team already uses.
- Not ours. Hand the case to the tenant, landlord or warranty holder with the footage, so the handover is a record rather than an argument.
Routing rules can be set per form, with an SLA in hours, so a case tagged urgent from the north building notifies the on-call engineer immediately while a planned fault at the depot batches into the morning digest. That is the video workflow automation layer; the dispatch decision stays with a person.
The contractor quote is where this pays for itself first. Today the sequence is: ticket, contractor survey visit, quote a few days later, approval, second visit to do the work. With the case in front of them, most contractors will quote the standard jobs from the video and the model plate, and the quote references what was shown. When the invoice arrives, the desk is comparing it with the same file. When a recall is disputed, both sides are looking at the same footage. That is the mechanism we described for remote diagnostics in field service, applied to a contractor rather than an employee.
The limit is worth stating plainly. A roof, a chiller replacement or anything structural still gets a survey. The approach is remote-first, and the survey stays where it earns its cost. What changes is that the survey visit is planned from evidence instead of being the evidence.
If your contractor management is the painful part, book a demo and bring the messiest quote thread you have. We will build the flow that would have replaced it.
Closing the loop with completion evidence
A status field that says "Completed" tells you somebody clicked "Completed". For an in-house team that is a trust question. For a contractor it is an invoice question.
The completion capture uses the same mechanics in the other direction. The contractor or the technician gets a completion link (or, for staff, starts the capture on site without one), and the flow asks for what the desk needs to close the job:
- Before and after. The fault as found and the repair as left, on the same asset label, so the two sit side by side.
- The spoken summary. What was done, what was replaced, anything found that was not on the ticket.
- The controls. Make-safe removed, guards refitted, area handed back, signed declaration at the end.
The completion case is linked to the original report, so one job has one file: the occupant's footage, the desk's decisions, the contractor's quote, and the work as shown. If the after shot does not match, the desk requests a retake, and the reason is stored with it. That is the proof of work pattern from the field service version of this post, and the review and retake step is the same.
On arrival every submission is sealed: a server receipt time, a SHA-256 fingerprint per file and a signed seal with a downloadable manifest, so a year later anyone can check that the file in the dispute is the file that arrived. Be precise about what that proves. It proves when the case arrived and that it has not been altered since. It does not prove when the leak started or who caused it. A sealed submission with a complete audit trail is a strong record in an invoice dispute or a recall argument precisely because it does not claim more than it can show.
One flow, every site
Multi-site is where a video-first helpdesk stops being a nice idea and becomes the operating model. The estate is a field on the case, not a separate process per building.
- One flow per fault type, not per site. A leak flow, a heating flow, a door and access flow, a lift flow. Every site uses the same one, so the desk sees the same evidence in the same order whether the report came from the depot or the head office.
- A QR code per location. The code carries the site and area, so the occupant never types an address and the desk never asks "which floor?".
- Central review, local action. The desk classifies and routes from the footage; the site engineer or the regional contractor acts on it. Nobody needs to be in the building to decide.
- Comparable cases. Because the flow is fixed, case 40 looks like case 400. That is what makes reporting on recurring faults per site possible, and it is the property of a well-run case management record rather than a chat thread.
The scale of the estates this matters for is easy to underestimate. The NHS estate in England alone reported a maintenance backlog of £15.9 billion in its 2024/25 Estates Returns Information Collection, up 15.7% on the previous year, according to BCIS. Every one of those sites has a helpdesk, and every helpdesk is deciding from sentences.
For the desk, the honest measure is how many reports never become a visit. In customer support and field service settings, teams using Venta Capture handle and close 4 out of 5 issues the same day, because the review takes minutes and the decision no longer waits for someone to arrive. Your figure will depend on your fault mix, and the way to find out is to run one fault type through it for a month.
Standard flows come ready-made and your team adjusts the steps, slots and questions in the builder. Branching flows, routing rules tied to your CAFM and integrations with your ticketing system are set up by VentaVid around your process, scoped on a call. The field service and property pages show the two ends of the FM spectrum, and the photo evidence standards post covers what a record needs if it may ever be audited. If you are comparing this with a remote inspection tool, the difference is that the person capturing here is the occupant, not an inspector.
Trade-specific versions of the same triage: HVAC, plumbing, roofing and fire and security panels. The triage decision itself is in reducing unnecessary service visits.
Frequently asked questions
What is maintenance request triage?
It is the set of decisions a helpdesk makes on each incoming fault before any work starts: what kind of fault it is, how urgent it is, who should handle it, and whose responsibility it is. Good triage decides all four from evidence; poor triage guesses from a one-line description and finds out on site.
What are the usual triage categories for facilities requests?
Most desks use a priority scale (emergency, urgent, routine, planned) crossed with a trade. The four outcomes that matter for routing are urgent, planned, contractor and not ours, because each one goes to a different person and needs different evidence to close.
Do occupants or tenants need an app to report a fault with video?
No. The link or QR code opens a page in the phone's browser, with no app and no account, in the occupant's language. The desk decides which steps and questions appear, so the occupant is guided rather than left to choose what to film.
Can a contractor quote from a video instead of a survey visit?
For standard jobs, yes: the contractor gets a view link to the case, sees the fault, the surroundings, the access and the model plate, and quotes from that. Large or structural work still gets a survey, but the survey is planned from the footage rather than being the first look.
How does this fit our CAFM or CMMS helpdesk?
The case carries your ticket number in its reference field and can be copied into the ticket in one click, shared by view link, or pushed into your system by webhook or API as it arrives. The helpdesk keeps its system of record; the capture sits in front of it and feeds it.
What if the occupant records the wrong thing?
The desk requests a retake in one click, with a note on what is missing, and the occupant gets a new link. The retake is stored as a case linked to the original, with the reason recorded, so the file shows what was asked for and when.
Does a video prove when the fault started?
No. The seal proves when the submission arrived and that the files have not changed since; it does not prove when the event happened. A timestamped report is strong evidence that you were made aware at a given moment, which is what a response clock or an SLA measures.
Start with the fault type that costs the most visits
Pull last quarter's tickets and find the fault type with the most "no fault found" or "reset on site" outcomes. Build the fault-report flow for that one, put the QR code where it happens, and route the next month of reports through it. The visit count is the measure, and it will move before the month is out.
What it takes to start.
- Free plan, no credit card. Build the flow, send it to your own phone, then to the next occupant who reports that fault.
- Live in 10 minutes for a first standard flow. Stuck? Book a free setup call and we build it together.
- Everything goes out in your own branding, in the occupant's language, with no app to install on either side.
The VentaVid team builds Venta Capture and will set up your first fault-report flow with you on a free setup call.
Start for free and put a QR code on the plant room door this week. If you would rather see it against your own tickets first, book a demo and bring the site that generates the most call-outs. The Venta Capture product page has the full feature list.