What the report has to prove
A reactive SLA is not one number. It is a set of clocks that start when a job is logged and stop at different points: when someone acknowledges the call, when an engineer arrives, when the hazard is made safe, and when the repair is finished. A client reading "94% within SLA" has no way to tell which of those clocks failed, or whether the figure counts jobs that are not finished yet.
A useful response and attendance report does two things. It shows performance by priority band, because a missed emergency and a late routine repair are not the same failure. And it lists every breach with the timestamps that caused it, so the conversation with the client is about a specific job on a specific night rather than a percentage. Operations directors use the same list internally: the timestamp that went wrong points at the part of the process that needs fixing.
Priority bands and their clocks
Contracts define bands differently, and the units change from band to band. Two published examples show the range:
- A City of London review of repairs response times (October 2018) proposed Priority 0 "attend, complete and make safe" within 4 hours and Priority 1 within 24 hours, both qualified as within normal working hours, then 3, 5 and 20 working days for priorities 2 to 4, plus a separate out-of-hours emergency category at 4 hours.
- An Exeter City Council service statement for responsive repairs (a draft covering 2007 to 2010) used four categories: emergency the same working day, urgent the next working day, and routine within 5 or 20 working days.
So one schedule mixes hours, working days and "same working day". Your report has to store, for each band, which milestones are measured, the target for each, and which clock it runs on. This template uses four example bands:
| Band | Clock | Acknowledge | Attend | Make safe | Complete |
|---|---|---|---|---|---|
| P1 Emergency | 24/7 | 15 min | 2 h | 4 h | not measured (follow-on raised as P3) |
| P2 Urgent | 24/7 | 1 h | 8 h | not measured | 24 h |
| P3 Routine | Working hours | 4 h | 30 h (3 days) | not measured | 50 h (5 days) |
| P4 Planned | Working hours | 10 h (1 day) | not measured | not measured | 200 h (20 days) |
Working hours here are 08:00 to 18:00, Monday to Friday, so one working day is 10 clock hours. These targets are illustrative. Use the ones in your contract.
Clock rules to settle before you count anything
Most SLA disputes are about the clock, not the work. Write these rules down and apply them the same way every month.
When the clock starts. Use the time the job was logged or received, not when it was assigned to an engineer. Assignment delay is part of the response the client is paying for.
Which hours count. On a 24/7 clock every minute counts. On a working-hours clock, a job logged at 18:00 on a Wednesday starts counting at 08:00 on Thursday, and a weekend contributes nothing. Bank holidays come out of working time too; the England and Wales dates are published on GOV.UK, and Scotland and Northern Ireland have their own.
Pauses. Many contracts let the clock stop for causes the contractor does not control: no access, waiting for the client to approve a quote, a client-agreed parts lead time. In this template a pause stops the completion clock only. Acknowledge, attend and make-safe clocks never pause, because nothing the client does should delay you answering the phone or turning up. Every pause needs a start, an end and a reason, and the client will expect evidence.
Jobs that are not finished. At the moment the report runs, an open job is either still inside its target ("not due") or already past it ("open breach"). Not-due jobs are excluded from the percentage. Counting them as passes flatters the month; counting them as fails punishes you for work you still have time to do.
Clock changes. In 2026 the clocks went back at 2am on 25 October (GOV.UK). If your system exports UTC and someone pastes local times alongside, a 24/7 clock for a job spanning that night is an hour out. Store one time basis and convert once.
The data model
Fields
| Record | Fields | Usual source |
|---|---|---|
| Job | job ref, site, priority, logged at | Job system or helpdesk |
| Milestones | acknowledged at, attended at, made safe at, completed at | Job system, engineer app |
| Pause | job ref, pause start, pause end, reason, evidence link | Job system notes or a hold status |
| Band | priority, clock type, target hours per milestone | Contract schedule, usually a spreadsheet |
| Calendar | working day start and end, bank holidays | Contract terms |
Some systems hold these commitments directly. PTC's ServiceMax Core work order data dictionary lists customer commitment fields such as SVMXC__Initial_Response_Customer_By__c, SVMXC__Onsite_Response_Customer_By__c and SVMXC__Restoration_Customer_By__c, actuals such as SVMXC__Actual_Onsite_Response__c, and pause fields including SVMXC__SLA_Clock_Paused__c, SVMXC__SLA_Clock_Pause_Time__c and SVMXC__SLA_Clock_Pause_Restart_Time__c. If your system calculates its own deadlines, recalculate them from the raw timestamps anyway and compare: a wrong calendar or band set-up in the system will otherwise go into the client report unnoticed. More on that system on the ServiceMax integration page.
Joins
| From | To | Key | Note |
|---|---|---|---|
| Job | Band | priority | Gives clock type and targets; a priority changed mid-job needs a rule (usually the band at logging) |
| Job | Pause | job ref | Zero, one or several pauses per job; sum them |
| Job | Helpdesk call | ticket or call ref | Only if the clock starts at the call, before the job exists |
| Band | Calendar | clock type | Working-hours bands read the holiday list |
Formulas
| Measure | Formula |
|---|---|
| Elapsed, 24/7 | (milestone − logged) × 24 |
| Elapsed, working hours | ((NETWORKDAYS(logged, milestone, Hols) − 1) × (Close − Open) + IF(NETWORKDAYS(milestone, milestone, Hols), MEDIAN(MOD(milestone, 1), Close, Open), Close) − MEDIAN(NETWORKDAYS(logged, logged, Hols) × MOD(logged, 1), Close, Open)) × 24 |
| Completion, net | elapsed to completion − pause hours (pause measured on the same clock) |
| End point if open | report run time instead of the missing milestone |
| Result | Met if recorded and ≤ target; Breach if recorded and > target; Not due if open and ≤ target; Open breach if open and > target |
| % met per band and milestone | Met ÷ (Met + Breach + Open breach) |
The working-hours line is the formula used in the download. It counts the part days at each end and the full working days between, and skips weekends and the holiday list.
Run Mon 2 Nov 08:00
Reactive SLA report · Brackenmoor Estates · October 2026
11 of 103 jobs logged
missed an SLA target.
Measured from the time each job was logged, on the clock its priority band uses. Jobs still inside their target are shown as not yet due, not as passes.
- Jobs logged
- 103
- 20 still inside target
- Attended on time
- 95.0%
- 76 of 80 measured
- P1 made safe in 4h
- 10/11
- Emergency, 24/7 clock
- Open breaches
- 2
- Past target, not yet done
By priority band · met ÷ measured
| Band | Clock | Acknowledge | Attend | Make safe | Complete | Jobs |
|---|---|---|---|---|---|---|
| P1 Emergency | 24/7 | 10/11≤ 15m | 10/11≤ 2h | 10/11≤ 4h | n/a | 11 |
| P2 Urgent | 24/7 | 24/24≤ 1h | 23/24≤ 8h | n/a | 22/24≤ 24h | 24 |
| P3 Routine | Working hours | 45/46≤ 4h | 43/45≤ 30h · 1 not due | n/a | 37/41≤ 50h · 5 not due | 46 |
| P4 Planned | Working hours | 21/21≤ 10h · 1 not due | n/a | n/a | 7/7≤ 200h · 15 not due | 22 |
Working hours: Monday to Friday, 08:00–18:00, bank holidays excluded. A client-caused pause stops the completion clock only; 21 jobs had one.
Breach list · the timestamps behind each miss
2 still open
| Job | Logged | Acknowledged | Attended | Made safe | Completed | Breached |
|---|---|---|---|---|---|---|
| RJ-40116P1 · Sedge Row Depot | Sun 4 Oct 09:25 | Sun 4 Oct 09:36 | Sun 4 Oct 11:41 (breached) | Sun 4 Oct 11:56 | n/a | Attend 2h 16m vs 2h |
| RJ-40192P2 · Sedge Row Depot | Sun 4 Oct 22:41 | Sun 4 Oct 23:06 | Mon 5 Oct 10:17 (breached) | n/a | Mon 5 Oct 19:13 | Attend 11h 36m vs 8h |
| RJ-40108P1 · Ropewalk Studios | Thu 8 Oct 15:59 | Thu 8 Oct 16:06 | Thu 8 Oct 16:52 | Thu 8 Oct 22:05 (breached) | n/a | Make safe 6h 6m vs 4h |
| RJ-40102P1 · Fernhill Annexe | Fri 9 Oct 06:44 | Fri 9 Oct 07:07 (breached) | Fri 9 Oct 08:10 | Fri 9 Oct 09:23 | n/a | Acknowledge 23m vs 15m |
| RJ-40332P3 · Linnet Court | Wed 14 Oct 16:28 | Thu 15 Oct 08:22 | Wed 21 Oct 12:44 (breached) | n/a | Thu 22 Oct 17:06 (breached) | Attend 46h 16m vs 30hComplete 60h 38m vs 50h |
| RJ-40245P3 · Linnet Court | Sun 18 Oct 10:00 | Mon 19 Oct 10:54 | Wed 21 Oct 08:21 | n/a | Tue 27 Oct 12:51 (breached) | Complete 64h 51m vs 50h |
| RJ-40170P2 · Fernhill Annexe | Tue 20 Oct 06:46 | Tue 20 Oct 07:14 | Tue 20 Oct 13:17 | n/a | Wed 21 Oct 14:00 (breached) | Complete 31h 14m vs 24h |
| RJ-40196P2 · Linnet Court | Tue 20 Oct 17:56 | Tue 20 Oct 18:43 | Tue 20 Oct 22:36 | n/a | Thu 22 Oct 04:37 (breached) | Complete 34h 41m vs 24h |
| RJ-40304P3 · Fernhill Annexe | Wed 21 Oct 13:47 | Wed 21 Oct 15:29 | Tue 27 Oct 10:45 (breached) | n/a | open (breached) | Attend 36h 58m vs 30hComplete 74h 13m vs 50h at run time |
| RJ-40210P3 · Brackenmoor House | Wed 21 Oct 18:00 | Thu 22 Oct 10:42 | Mon 26 Oct 12:48 | n/a | open (breached) | Complete 70h vs 50h at run time |
| RJ-40293P3 · Fernhill Annexe | Fri 23 Oct 15:38 | Mon 26 Oct 10:41 (breached) | Tue 27 Oct 16:08 | n/a | Thu 29 Oct 14:36 | Acknowledge 5h 3m vs 4hPaused 13h: Site closed by client |
Elapsed times are on each band’s clock, net of pauses. A P2 logged on Sunday night is on a 24/7 clock, so its overnight hours count.
Worked example (illustrative, fictional)
Quillon Building Services, a fictional contractor, runs the report for its fictional client Brackenmoor Estates at 08:00 on Monday 2 November 2026, covering the 103 reactive jobs logged in October. Three jobs from the breach list and the workings behind them:
RJ-40192, P2, 24/7 clock. Logged Sunday 4 October at 22:41, attended Monday 5 October at 10:17. Elapsed: 11 hours 36 minutes against an 8-hour target, a breach by 3 hours 36 minutes. Had P2 been on a working-hours clock, the same visit would have taken 2 hours 17 minutes (08:00 to 10:17 on Monday) and passed. The clock type decides the result, which is why it belongs in the band table and not in someone's head.
RJ-40212, P3, working hours, with a pause. Logged Thursday 22 October at 17:39.
- Acknowledged Friday 23 at 10:56: 21 minutes on Thursday plus 2 hours 56 minutes on Friday = 3 hours 17 minutes. Target 4 hours: met.
- Attended Monday 26 at 09:08: 21 minutes + 10 hours (Friday) + 1 hour 8 minutes = 11 hours 29 minutes. Target 30 hours: met.
- Completed Friday 30 at 09:45. Gross: 21 minutes + five full working days (Friday 23, Monday 26 to Thursday 29) at 10 hours + 1 hour 45 minutes = 52 hours 6 minutes.
- Paused for client approval of a quote from Monday 26 at 10:03 to Wednesday 28 at 13:03: 7 hours 57 minutes + 10 hours + 5 hours 3 minutes = 23 hours.
- Net completion: 52 hours 6 minutes − 23 hours = 29 hours 6 minutes. Target 50 hours: met. Without the pause it would have been a breach by 2 hours 6 minutes, so the approval request and reply need to be on file.
RJ-40210, P3, open breach. Logged Wednesday 21 October at 18:00, the moment working hours ended, so the clock started at 08:00 on Thursday. Attended Monday 26 at 12:48: 10 + 10 + 4 hours 48 minutes = 24 hours 48 minutes, met. Still not completed at run time. Its 50 working hours ran out at 18:00 on Wednesday 28 October; by the run time it stood at 70 hours, an open breach 20 hours over.
The band roll-up for the month:
| Band | Jobs | Acknowledge | Attend | Make safe | Complete |
|---|---|---|---|---|---|
| P1 | 11 | 10/11 (90.9%) | 10/11 (90.9%) | 10/11 (90.9%) | n/a |
| P2 | 24 | 24/24 (100%) | 23/24 (95.8%) | n/a | 22/24 (91.7%) |
| P3 | 46 | 45/46 (97.8%) | 43/45 (95.6%), 1 not due | n/a | 37/41 (90.2%), 5 not due |
| P4 | 22 | 21/21 (100%), 1 not due | n/a | n/a | 7/7 (100%), 15 not due |
Attendance across the bands that measure it: 76 met of 80 measured, 95.0%. Eleven of the 103 jobs breached at least one milestone, two of them still open. P4 completion reads 100%, but only 7 of 22 jobs have reached their deadline; the report should say so rather than let the figure stand alone.
Reading the breach list
The milestone that failed tells you where to look:
| Breached milestone | Where to look first | Usually owned by |
|---|---|---|
| Acknowledge | Out-of-hours call handling, inbox or portal monitoring | Helpdesk |
| Attend | Dispatch decisions, travel, engineer availability | Planning |
| Make safe | Kit on the van, skills match, access on arrival | Engineer and supervisor |
| Complete | Parts, second visits, quotes awaiting approval, unrecorded pauses | Operations |
Two checks before the list goes to a client. First, every pause claimed should have its evidence attached, because a pause is the item a client is most likely to challenge. Second, any job whose priority changed after logging should show both bands and which one was measured.
For the monthly pack with weighted KPIs and service credits, see FM contract KPI reporting. Planned maintenance has its own report in the PPM compliance report template, and the engineer utilisation template covers whether you have the hours to hit these targets at all. All templates are on the field service reporting hub.
How to automate it with IVP.ai
The IVP.ai Data Layer pulls jobs, milestone timestamps and pause records from your job system, reads the band table and calendar from the contract spreadsheet, and applies one set of clock rules to every job. The report above then runs from governed data rather than a monthly export. A national fire safety and security company used the same approach to bring HubSpot, Uptick and finance spreadsheets into one governed view, saving more than 10 hours a week on reporting (case study).
A Deskless by IVP.ai agent can watch the same clocks during the day: flagging a P1 with no attendance logged as it nears 2 hours, or a P3 that will run out of working hours tomorrow, to whoever owns it. By default it flags and waits for a person; it changes a job or tells a client only if you configure it to.
To see your own response times measured this way, start with the free 48-hour pilot. We connect up to three sources, typically your job system and the contract schedule, clean the data and set one report or agent running within 48 hours of read-only access. You then have 14 days of access to choose a plan or walk away.