Crew time and work quality

Every shift from ClockTimes is matched to that person's booked entries for the same day and split into five parts. Nothing is inferred from the ticket record alone — the shift clock defines the paid day, which matters because the two disagree.

• Booked on customer property — time attributed to a customer's work ticket.
• Booked at the yard — time attributed to a ticket whose property is the branch itself.
• Gaps between booked entries — clocked on, between two pieces of booked work.
• Before the first ticket — clocked on at the start of the day with nothing booked yet.
• After the last ticket — last ticket ended, still clocked on.

What does not reconcile is reported as residual rather than hidden: it arises where booked entries overlap each other or fall outside the clock span, so it is a data-quality measure as much as a time one.

Three things this shows

Utilisation spans 80.6 – 92.3 points. On 3,148,099.4 clocked hours that range is worth a great deal in principle. In practice most of the gap needs validating before it can be banked — see the caution below.

The idle time is at the front of the day, not the end. Time clocked on before the first ticket runs 3.4 – 7.9% of the shift and is the largest unbooked block in almost every tenant. That is a morning-start question — loading, briefing, dispatch — not an end-of-day one.

Only one tenant books yard time at all. Signature attributes 5.9% of the shift to yard tickets; every other tenant records zero. Their shop and load time is therefore either inside customer tickets or unbooked, which is the clearest single proof that these columns measure recording practice as much as behaviour.

Aspire holds no defect or complaint record, so quality has to be inferred. Three fields carry real signal, in descending order of directness: warranty labour is work redone at no charge, which is the closest thing to a defect; review shows whether anyone checked the work before it was billed; cancellation shows work planned and dropped.

Warranty labour is real and material. It is concentrated in very few tickets — typically a few hundred out of tens of thousands — averaging tens of hours each, so it is large-job rework rather than routine touch-ups. Those tickets are individually identifiable and worth reading.

Review discipline is the sharpest divide in the whole analysis. It ranges from 4.2 – 61.1%. Approval, by contrast, sits at 99.8% almost everywhere — so approval is a formality and review is the only real control on work leaving the branch. Where review is in single digits, nothing checks the work before it is billed.

Revisits as a rework signal. Counting properties completed again within seven days returned 72–87%, which measures weekly service cadence, not rework. Tightening the window, 28–35% of consecutive completions are same-day — several services on one visit — and only 7–14% are next-day. Even that cannot separate rework from a legitimately separate service. Doing it properly needs the service identity on both visits, so that like is compared with like.

Completed work with no invoice reference. Between 54% and 76% of completed tickets carry no InvoiceID, representing tens of millions in Price. This looks like unbilled revenue and almost certainly is not: Aspire invoices fixed-price maintenance at the opportunity level through InvoiceOpportunityService, not per ticket, so an absent ticket-level reference is expected for contract work. Establishing whether anything is genuinely unbilled requires pulling that join and testing time-and-materials tickets specifically. The figure is deliberately not quoted here.

• Read the warranty tickets. A few hundred tickets carry thousands of hours of no-charge labour, and they are individually identifiable. That is a morning's work and it either finds a pattern or rules the area out.
• Ask one branch what happens before the first ticket. The largest unbooked block is the morning, and whether it is loading, briefing or dispatch delay is not in the data.
• Establish whether review matters. Test whether tickets that were reviewed differ in warranty rate or estimate variance from those that were not. If review changes nothing, the divide between 4% and 61% is not worth closing.
• Settle the invoicing question properly by pulling InvoiceOpportunityService, so the uninvoiced figure can be either quantified or dismissed.
• Normalise the recording differences — yard booking and warranty practice — before any cross-opco target is set on either measure.