Integrations

BigChange data for engineer utilisation reporting

BigChange already slices each engineer's day into job, driving, absence and unallocated time; the work is choosing the right records and dividing by the right hours. This guide maps the BigChange data behind a per-engineer utilisation report, with the joins, formulas and checks to run before anyone acts on the numbers.

By Simon DuttonLast reviewed

BigChange records a great deal about how an engineer's day is spent: when they pressed "start driving", when the job started and ended on the device, what they put on their timesheet, and when they were booked off. The difficulty with BigChange reporting on utilisation is not a lack of data. It is that the same day appears in four or five places, each answering a slightly different question, and a figure built from the wrong one can be tens of points out, as the worked example below shows. This page maps the BigChange records that matter, shows how to turn them into one utilisation figure per engineer, and lists the checks worth running before anyone acts on it.

For the general argument about what belongs above and below the line (billable versus productive, whether training counts as available, what to do with overtime), see the engineer utilisation template. This page takes those choices as made and deals with where BigChange keeps the inputs.

Two ways to get the data out

BigChange publishes two routes on its developer portal, and they suit different jobs.

  • The REST API is for integrations and real-time questions. Its timesheets endpoint, for example, retrieves a paged list of timesheets for a resource and needs the resources:read scope.
  • Data as a Service (DaaS) is a set of read-only reporting tables shared into a Snowflake account. BigChange's DaaS FAQ describes them as flat, wide "marts" with most joins already de-normalised and common calculations built in, aimed at retrospective analysis rather than "where is my vehicle now?".

For a utilisation report, DaaS is usually the better starting point, because BigChange has already done the hard part: slicing each engineer's day into job, driving, absence and unallocated time. Two practical points from the DaaS reference: the marts refresh on a published schedule through the working day (UTC), so a figure read at 09:00 reflects data entered up to 07:30, and data entered after 15:30 arrives at 04:30 the next business day. Soft-deleted records stay in the marts with deleted_at filled in, so filter on it.

The BigChange records behind a utilisation figure

Every table and column name below is taken from BigChange's DaaS data dictionary, linked in the first column. Check the mart version on your own account before building on it: BigChange versions its marts (_v1 here) so a new version can change columns without breaking the old one.

Sources and fields

DaaS mart What one row is Columns a utilisation report uses
resource_day_summaries_v1 One engineer, one day shift_duration_in_hours, job_duration_within_wh_in_hours, driving_duration_within_wh_in_hours, absence_duration_in_hours, unallocated_duration_in_hours, availability_duration_in_hours, job_duration_outside_wh_in_hours, plus _as_resource and _as_assistant splits
resource_activities_v1 One activity slice per engineer per day type (Job, Driving, Absence, Unallocated), duration_in_hours, is_within_working_hours, is_assistant, job_id, job_type, job_category
timesheet_activities_v1 One logged timesheet activity name, start_at, end_at, status (Unsigned, Signed, Queried, Rejected, Approved, Paid), _resource_id
resource_absences_v1 One absence request type, status (Requested, Queried, Rejected, Approved), is_paid, start_at, end_at
resource_working_hours_v1 One shift pattern per weekday day_of_week, start_time, finish_time
jobs_v1 One job status, actual_duration_in_hours, actual_driving_duration_in_hours, type_name, category_name, _resource_id
resource_journeys_v1 One tracked journey start_at, stop_at, duration_in_seconds, distance_mi, idling_time_in_seconds

A few definitions from the dictionary are worth knowing before you build. Working hours are the shift pattern set on the resource, so "within working hours" means inside that pattern. Unallocated time is shift time not spent on driving, a job or an absence. Job driving time on jobs_v1 runs from "Start driving" to "Start job" on the device, while resource_journeys_v1 comes from the tracker or tablet and covers the whole journey, so the two will not match and are not meant to.

Joins

From To Key Watch for
resource_day_summaries_v1 timesheet_activities_v1 resource_id = _resource_id, and the day of start_at = date Timesheet times are stored with an offset; convert to the engineer's time_zone before taking the date
resource_activities_v1 jobs_v1 job_id = id Lets you split job time by job type or category, for example reactive against planned
resource_day_summaries_v1 resource_absences_v1 resource_id = _resource_id Only Approved absences should reduce availability in a plan; the summary already reflects actual absence
jobs_v1 resource _resource_id One assigned resource per job; assistants are held in a separate job_assistants_v1 mart, so job-level sums undercount them

Formulas

Measure Formula, using resource_day_summaries_v1
Available hours sum of availability_duration_in_hours (shift less absence)
Job utilisation sum of job_duration_within_wh_in_hours ÷ available hours
Field utilisation sum of driving_and_job_duration_within_wh_in_hours ÷ available hours
Unallocated share sum of unallocated_duration_in_hours ÷ available hours
Out-of-hours job time sum of job_duration_outside_wh_in_hours, reported on its own line
Timesheet gap (approved or paid timesheet hours − device job, driving and out-of-hours time) ÷ device time

Sum the hours first and divide once. A team or region figure is total job hours over total available hours, never the average of each engineer's percentage.

IVP.ai / Data Layer

5 October 2026 to 30 October 2026 · 20 working days

Engineer utilisation · Quillmere Building Services (fictional)

60.8% of available hours
were spent on jobs.

8 engineers. Available hours are shift hours less absence. Job and driving time are device-recorded and counted within working hours only.

Job utilisation
60.8%
729.2 of 1200 h
Field utilisation
79.1%
Job plus driving
Unallocated
250.4 h
20.9% of available
Out-of-hours jobs
32.3 h
Reported separately

Available hours by engineer, split by activity

  • Job
  • Driving
  • Unallocated
  • Aled P. · North62.4% job · 81.3% field
    Aled P.: 160 available hours99.9 hours on jobs, 30.1 hours driving, 30 hours unallocated.
  • Bryony K. · North · 40 h absent67.3% job · 82% field
    Bryony K.: 120 available hours80.8 hours on jobs, 17.6 hours driving, 21.6 hours unallocated.
  • Dev S. · North51.6% job · 82.8% field
    Dev S.: 160 available hours82.5 hours on jobs, 49.9 hours driving, 27.6 hours unallocated.
  • Harriet O. · South69.8% job · 82.8% field
    Harriet O.: 160 available hours111.7 hours on jobs, 20.7 hours driving, 27.6 hours unallocated.
  • Joel M. · South · 8 h absent46% job · 63.2% field
    Joel M.: 152 available hours69.9 hours on jobs, 26.1 hours driving, 56 hours unallocated.
  • Kasia W. · South65.9% job · 85.1% field
    Kasia W.: 160 available hours105.5 hours on jobs, 30.6 hours driving, 23.9 hours unallocated.
  • Liam T. · South60.1% job · 75.5% field
    Liam T.: 160 available hours96.2 hours on jobs, 24.6 hours driving, 39.2 hours unallocated.
  • Noor R. · North · 16 days64.6% job · 80.9% field
    Noor R.: 128 available hours82.7 hours on jobs, 20.8 hours driving, 24.5 hours unallocated.

Bar length is available hours (longest 160 h). A shorter bar means absence or a shorter working week, not lower utilisation.

The numbers behind the chart

Hours and utilisation per engineer for the period, with team totals
EngineerAvailable hJob hDriving hUnallocated hOut of hours hJob %Field %
Aled P.16099.930.1306.362.4%81.3%
Bryony K.12080.817.621.6267.3%82%
Dev S.16082.549.927.6251.6%82.8%
Harriet O.160111.720.727.612.669.8%82.8%
Joel M.15269.926.156046%63.2%
Kasia W.160105.530.623.98.265.9%85.1%
Liam T.16096.224.639.2060.1%75.5%
Noor R.12882.720.824.51.264.6%80.9%
Team1200729.2220.4250.432.360.8%79.1%

Team figures are total hours over total hours, not an average of the eight percentages. Out-of-hours job time is shown but left out of both sides of the fraction; counting it gives 61.8%.

Week by week

  1. w/c 5 Oct · 304 h62.2% job · 80.2% field
  2. w/c 12 Oct · 312 h61.9% job · 80.4% field
  3. w/c 19 Oct · 272 h58.9% job · 78% field
  4. w/c 26 Oct · 312 h59.8% job · 77.9% field

Checks before the figures are shared

resource_day_summaries_v1 + timesheet_activities_v1

  1. Harriet O. · Timesheet gapFor review118.9 h on approved timesheets vs 145 h recorded on the device (-18%)
  2. Joel M. · High unallocated timeFor review56 h of 152 h available not on a job or driving (36.8%)
  3. Dev S. · Driving shareNoted49.9 h driving, 31.2% of available hours
  4. Liam T. · Mostly assistant timeNoted67.2 of 96.2 job hours as assistant: missing if utilisation is built from jobs_v1 alone

Timesheet check flags gaps over 10% between approved timesheet hours and device-recorded hours

Illustrative data

Illustrative, fictional data. Company, engineers and hours are invented; column names follow BigChange's published data dictionary. Not customer results or a live product screenshot.

Worked example (illustrative, fictional)

Quillmere Building Services is an invented company with eight engineers on BigChange, split between a North and a South team. The figures are for the four working weeks from Monday 5 to Friday 30 October 2026, and they match the report above.

Available hours. Seven engineers work five 8-hour days and one works four, so shift time is 7 × 160 + 128 = 1,248 hours. Absence is 48 hours: a week's leave for Bryony (40) and one day for Joel (8). Available hours are 1,248 − 48 = 1,200.

Team utilisation. Device-recorded job time within working hours totals 729.2 hours, so job utilisation is 729.2 ÷ 1,200 = 60.8%. Adding 220.4 hours of driving gives 949.6 ÷ 1,200 = 79.1% field utilisation. The remaining 250.4 hours, 20.9%, is unallocated. Those three add back to 1,200. Averaging the eight engineers' job percentages would give 61.0%, slightly off because Bryony, Joel and Noor had fewer available hours but count as much as everyone else.

Out-of-hours work. Engineers logged 32.3 job hours outside their shift patterns. Leaving them out keeps the figure honest about scheduled capacity. If you include them, add them to both sides: (729.2 + 32.3) ÷ (1,200 + 32.3) = 761.5 ÷ 1,232.3 = 61.8%.

Absence handled properly. Bryony did 80.8 job hours in a 160-hour shift month with 40 hours of leave. Against her available 120 hours that is 67.3%. Against shift hours it would be 50.5%, which would make the second-best performer look like one of the weakest.

Driving. Dev covers a rural patch. His job utilisation is 82.5 ÷ 160 = 51.6%, the second lowest, but his field utilisation is (82.5 + 49.9) ÷ 160 = 82.8%. Driving takes 49.9 hours, 31.2% of his available time. That is a routing and territory question, not an effort question.

Assistants. Liam is an apprentice who spends most of his time assisting. Of his 96.2 job hours, 67.2 were as assistant. Built from jobs_v1, which carries only the main resource, he would show 96.2 − 67.2 = 29.0 hours and 29.0 ÷ 160 = 18.1% utilisation, instead of the 60.1% the day summaries give.

Timesheets. Harriet's device shows 145.0 hours of job, driving and out-of-hours time. Her approved timesheet activities total 118.9 hours, a gap of 26.1 hours or −18.0%. Either time on site is not reaching the timesheet, which matters if timesheets drive pay or labour charges, or the device was left running. The report flags it for a person rather than picking one.

Joel is the other flag: 56.0 of his 152 available hours (36.8%) were unallocated, against a team share of 20.9%. Before treating that as a performance issue, check whether he was doing stores runs, training or admin that was never set up as an activity.

Where BigChange utilisation figures go wrong

  • Using job records alone. jobs_v1 gives one main resource per job. Assistant time sits in the day summaries and activities marts with an _as_assistant split or an is_assistant flag.
  • Shift hours as the denominator. Use availability_duration_in_hours, which already removes absence, or the leave weeks will drag people down.
  • Mixing device and tracker driving. Job-card driving and tracked journeys measure different spans. Pick one per measure and label it.
  • Unsigned timesheets. Filter timesheet_activities_v1 on status. Unsigned, queried and rejected entries should not feed pay or charging figures.
  • Running the report at the wrong time. DaaS refreshes in batches on UTC times. Anything entered after 15:30 lands at 04:30 the next business day, so a Monday pack should run after that refresh, and a mid-morning rerun will still be missing the last two hours.

Automating it with IVP.ai

The IVP.ai Data Layer reads BigChange alongside your other systems, keeps the joins and formulas above as tested definitions, and runs the utilisation report on a schedule, so the per-engineer chart is ready each Monday without anyone exporting anything. Because it sits in one governed warehouse, the same figures can sit next to job margins from finance or a CRM pipeline when the question moves from "are we busy?" to "are we busy on the right work?". For other systems feeding the same kind of report, see the Uptick, Simpro and ServiceMax pages, and the wider field service reporting overview.

Deskless by IVP.ai can take the checks list further: it can send Harriet's line manager the timesheet gap or ask the planner about Joel's unallocated hours, and wait for a person to decide. It does not change timesheets or schedules unless you configure it to.

A national fire safety and security company used the IVP.ai Data Layer to bring HubSpot, Uptick and finance spreadsheets into one governed view and now saves 10+ hours a week on reporting (case study). More on how it works is on the platform page.

To see your own engineers' figures, start a free 48-hour pilot. We connect up to three sources, clean the data and have one report or agent running on your real BigChange data within 48 hours of read-only access. You then have 14 days of access to decide whether to choose a plan or walk away.

Questions

Frequently asked questions

Can I report on engineer utilisation inside BigChange without a data warehouse?

For a quick look at one team, the screens inside BigChange may be enough. A repeatable utilisation report that combines device time, timesheets and absence, and keeps one agreed definition, is easier to build from BigChange's Data as a Service marts or its REST API, because you can store the formula once and rerun it every week.

Should utilisation in BigChange use timesheets or device-recorded job times?

Use device-recorded time from the day summaries for utilisation, because it is captured automatically as the engineer starts and finishes jobs. Use approved timesheet activities as a check. A large gap between the two for one engineer is worth a conversation: either time on site is not reaching the timesheet, or the device status is not being updated.

Why is an engineer's utilisation so low when they were busy all month?

Three common causes in BigChange data: the engineer mostly assists on jobs and the report was built from job records that only hold the main resource; leave was left in the denominator by using shift hours instead of available hours; or long driving times on a rural patch. Check field utilisation, which adds driving, alongside job utilisation.

How fresh is BigChange DaaS data for a Monday report?

BigChange publishes a refresh schedule in UTC. Data entered in the evening arrives at 04:30 the next business day, and daytime data arrives in roughly two-hourly batches. A weekly report scheduled after the early-morning refresh will include everything up to the previous evening. Check the current schedule on BigChange's DaaS reference before relying on it.

About the author

Simon Dutton

Simon has more than 10 years of experience as a data scientist, working with start-ups, SMEs, Google and the CMA.

Last reviewed 6 November 2026.

Your first report or agent in 48 hours.

Tell us which systems feed it. We connect them, clean the data and set the report running automatically. If it isn’t useful, walk away.