At 17:20, the daily report says: “Good progress. Some delays. Materials outstanding. Continue tomorrow.”
It sounds familiar. It is also almost impossible to manage.
Which work progressed? How much was completed? Which crew waited, for how long, and what could unblock the next shift? A manager reading that report still has to phone the supervisor to discover what happened.
A useful daily site report should let the next person make a decision without reconstructing the day from messages, memory and photographs. It is the link between the work you released, the resources you used, the output you can support and the conditions you hand over.
That requires a few precise records. It does not require a longer diary.
Give the report three jobs
Build your daily site report around three questions:
- What happened against the released plan? Record the work package, planned output, actual output and relevant resource hours.
- What changed the result? Record material events with times, evidence and the action taken.
- What does the next shift need? Transfer open constraints, decisions, changed conditions and accountable owners.
Use the 48-hour look-ahead to establish the ready work entering the plan. The report closes that loop by showing what the released work produced and what must change before tomorrow.
Keep detailed payroll, specialist inspection records, incident investigations and commercial correspondence in their appropriate systems. Reference them where needed. Duplicating every record inside a general daily report creates conflicting versions and unnecessary access to personal information.
Freeze the reference before the shift changes it
A plan-versus-actual report needs a plan that survives contact with reality. Record a plan version and release time before execution. For each work package, capture the location, scope, unit, quantity, resource allowance and planned start or finish.
When the client changes the priority at 10:00, add a change record. Preserve the original commitment and identify the authorised revised instruction. Do not quietly replace the morning target with the afternoon result.
Otherwise a difficult day becomes a successful plan simply because somebody edited the denominator.
One site can have many work-package rows. One work package can continue across several days. Give each a stable ID so the daily report, time record, delay event and quantity evidence describe the same work.
The minimum field register
Use a short summary with linked detail. The following fields are enough to build the first useful version.
| Record | Minimum fields | Why it exists |
|---|---|---|
| Report identity | Site, date, shift, report ID, cutoff time, author, revision | Distinguishes the period and the current record. |
| Released plan | Plan version, work-package ID, location, scope, planned unit and quantity, planned hours | Preserves the original commitment. |
| Output | Completed quantity, accepted quantity, pending quantity, acceptance basis and evidence reference | Separates physical progress from verified acceptance. |
| Resource hours | Crew or resource reference, work package, category, hours, source and reconciliation status | Shows how time was used without reproducing payroll detail. |
| Delay event | Event ID, start/end, affected resources, primary code, impact and evidence | Connects a disruption to a bounded exposure. |
| Change or instruction | Reference, time received, issuer role, changed scope and authorisation status | Makes deviations from the released plan visible. |
| Quality or control exception | Work affected, hold/status, responsible role and controlled record reference | Prevents incomplete or unreleased work being presented as finished. |
| Next-shift action | Action, owner, required-by time, status and closure evidence | Transfers work that still needs a decision. |
| Review | Reviewer, review time, unresolved queries and correction reference | Separates submission from verification. |
A field deserves space when it supports a decision, reconciliation or evidence trail. If nobody uses it and no defined reporting requirement needs it, remove it.
Record output in units someone can verify
“Nearly finished” is a conversation. “120 m² accepted in Zone B against a released target of 160 m²” is an operating record.
Define what each quantity means for the activity. Installed, inspected, accepted and handed over are different states. A report can show all four where useful, but it must label them and avoid counting the same output twice.
For example, 120 m² accepted plus 16 m² installed and awaiting inspection does not equal 136 m² accepted. The pending quantity remains visible until the acceptance condition is met. State who verifies the quantity and which drawing, measurement sheet or inspection reference supports it.
When yesterday’s pending work is accepted today, preserve its production date and record the acceptance date. Otherwise today inherits output made with yesterday’s labour and the daily productivity calculation becomes misleading. Use a consistent cumulative work-package view when acceptance timing crosses shifts.
Do not combine metres, square metres and item counts into one “total production” figure. Do not treat an unmeasured percentage estimate as the same quality of evidence as an agreed quantity.
Separate elapsed time from resource hours
A two-hour disruption is not automatically two labour-hours. If six people were unable to work for the full two hours, it represents 12 person-hours. If four moved to ready work after 30 minutes, the affected hours are different.
Record the affected crew and the relevant interval. Split the event when exposure changes. Time reassigned to productive work belongs against that work package, not simultaneously against delay on the original package.
Choose a time basis and write it down. Are the report’s hours paid hours, attendance hours or direct working hours? Are paid breaks and travel included? Do planned and actual hours use the same rule? A crew of eight people on an eight-hour paid basis represents 64 person-hours. That is not an eight-hour total.
Use mutually exclusive time categories within each resource interval: first-pass work, rework, recorded delay, necessary support, work travel, and any other explicitly defined category. Site-wide supervision can sit in a separate support cost pool if it cannot be allocated credibly. Allocate it once if you distribute it.
The control is simple: a person cannot contribute the same half-hour to two work packages or two time categories. Detailed names can remain in an access-controlled time system; the general report normally needs crew references and totals.
Describe delays so tomorrow’s owner can act
“Material delay” leaves too much missing. A better entry reads:
Event D-024: six-person crew unable to continue B-17 from 08:00 to 10:00. Specified components unavailable at the workface. Primary code: material availability. Exposure: 12 person-hours. Delivery receipt and supervisor check referenced. Work resumed at 10:00. Yard coordinator to verify tomorrow’s complete pick by 14:00.
That records an observation and an action. It does not automatically establish which organisation caused the delay, whether a cost is recoverable or whether every minute affected the completion date.
Use the shared site delay codes so similar events can be compared. Keep the observed condition, the suspected cause and any verified cause distinct. “Supplier failure” should not be selected merely because the material was missing; the original requirement may have changed after picking.
Where several constraints overlap, assign time once and record additional conditions as linked contributors. Do not add an hour of missing materials to the same hour of unavailable access and call it two hours lost.
A worked daily report
Synthetic example: the following figures illustrate the method. They are not a project benchmark.
Work package B-17 is a defined installation scope measured in square metres. The frozen plan calls for 160 m² at an allowance of 0.40 person-hours per m²: 64 planned person-hours. The shift closes with 120 m² completed and accepted, no additional completed quantity awaiting acceptance, and 40 m² remaining.
| Record | Reported result | Verification or follow-up |
|---|---|---|
| Released output | 160 m² | Frozen plan P-031, B-17. |
| Accepted output | 120 m² | Measurement record Q-017; acceptance confirmed. |
| Actual labour | 68 person-hours | 64 ordinary paid hours plus four overtime person-hours. |
| Time allocation | 44 first-pass work + 4 rework + 12 delay + 6 support + 2 work travel = 68 | Totals reconcile to the same paid-hours basis; unpaid breaks excluded. |
| Material event | D-024: six people × two hours = 12 person-hours | Observed shortage and restart time recorded; deeper cause still under review. |
| Next shift | 40 m² remains on B-17 | Supervisor to confirm readiness and revised commitment before release. |
The headline is not simply “four hours over budget”. The operation used 68 hours against a 64-hour plan while accepting only 120 of the planned 160 m². At the agreed allowance, accepted output represents 48 earned hours.
The planned versus actual labour-hours method explains why the comparable-output gap is 20 hours and why that still requires investigation. The report supplies the evidence; it should not jump directly from a variance to a judgement about the crew.
Nor should somebody multiply the 12 delay hours by a crew-wide rate. They are already person-hours. If management needs financial exposure, pass the event and resource records into a site downtime cost calculation with a clear rate basis.
What supervisors should stop recording
- Pages of routine activity with no result. “Attended meeting, checked site, spoke to stores” does not explain whether B-17 is ready or complete. Record the decision or change that matters.
- Repeated copy-and-paste constraints. An unresolved item needs an owner, age and required-by time. Repeating “awaiting materials” every day is not escalation.
- Unsupported blame. Record who supplied an instruction where relevant, but separate verified facts from assumptions about motive or fault.
- Photographs without context. Reference the work package, location, capture time and what the image demonstrates. A folder of pictures is not a measured quantity.
- Personal incident narratives in a general distribution report. Use a restricted incident reference and the operational condition or action needed. Keep medical details, identity documents and unrelated personal information out of the shared report.
- False precision. Mark estimates as estimates. An invented minute-by-minute history written from memory is less useful than a clearly bounded uncertainty.
- Decorative weather entries. Record weather conditions when they affect work, access or a specified reporting requirement, with the relevant interval and evidence. “Sunny” does not explain an output shortfall.
Build the report during the day
Before release: the supervisor confirms the plan reference, work-package rows, output units and open holds. The planner owns the approved plan baseline; the supervisor should not have to reconstruct it.
At the event: the supervisor or nominated recorder captures a material change, delay start, reassignment or restart. Record it near the time it happens, using a short event entry. Attach evidence once and reference it elsewhere.
Before the reporting cutoff: the supervisor reconciles output and resource hours. The quantity verifier confirms the accepted amount or marks it pending. The time-record owner resolves missing or overlapping allocations. Unresolved items remain visibly provisional.
At handover: the incoming responsible person acknowledges open holds, changed conditions and required actions. Acknowledgement means the information was received, not that every quantity or commercial position was agreed.
At the next review: the site manager works through exceptions: missing evidence, major deviations, overdue actions and tomorrow’s readiness. Do not spend the meeting reading every completed row aloud.
Correct the record without rewriting the day
Reports will contain errors. The answer is a correction trail: original value, corrected value, reason, author and time. Mark the current revision and preserve the earlier submission.
If a pending quantity becomes accepted, update the acceptance state and reference. If an uncertain delay cause is resolved, retain the original observation and add the verified finding. If hours are moved between packages, the site total should remain reconciled.
Set distribution by role. A client-facing summary, internal cost record and restricted incident file can reference the same event without exposing identical detail to every reader. Record locations precisely enough to act while keeping unnecessary personal information out of routine circulation.
Test whether the report changes a decision
Start with one supervisor, one shift and a small number of work packages. Check whether the report can answer these questions without follow-up calls:
- Which released outputs were accepted, and on what evidence?
- Do the recorded hours reconcile to the chosen time source?
- Which events explain material changes, and where is the cause still uncertain?
- What must happen next, who owns it, and by when?
Review late submissions, unresolved quantity queries, unallocated hours and overdue actions. Their purpose is to improve the record and response, not to reward the longest report.
After a week, remove fields that nobody uses and strengthen those that keep producing unanswered questions. Keep any required project reporting fields. The report should become easier to complete because the operating definitions are clearer.
The OperatorOS Principle
Record the day so the next decision can be made.
A strong daily site report preserves the commitment, shows the verified result and transfers what remains. It makes uncertainty visible before uncertainty becomes an argument.
The supervisor’s job is not to write a novel about being busy. It is to leave a record that another responsible person can understand, check and use.