Your supervisor texts you at 6:40 a.m. The client called. The third-floor restrooms were not touched last night. You open the timesheet and two cleaners clocked in at 6:02 p.m. and clocked out at 10:47 p.m. Everything looks normal.
That is the entire problem in one paragraph. A timestamp proves someone was on the clock. It says nothing about whether trash got pulled on floor three, whether the entry glass got wiped, or whether anyone opened the restroom door.
Tracking shift completion means building a record that answers a harder question: was the work that was due tonight actually performed, and can you show it to a client 36 hours later? Below is the evidence structure, the math, the alert thresholds and the review rhythm that make that record hold up.
Four layers of evidence, and what each one actually proves
Most operators track one signal, usually attendance, then act surprised when it fails to predict complaints. Attendance and completion are different measurements. You need both, plus verification.
Here is what each type of evidence buys you and where it stops.
| Evidence | What it proves | What it does NOT prove | Effort to collect |
|---|---|---|---|
| Scheduled shift record | Who was supposed to be where, and for how long | That anyone showed up | One-time setup |
| Geofenced clock-in and clock-out | A phone was inside the site boundary at two specific moments | What happened between those two moments | Seconds per shift |
| Task checklist marked by the cleaner | The cleaner claims the task was done, and when | That the task was done to standard | 1 to 3 minutes per shift |
| Photo evidence on selected tasks | The physical state of a specific area at a specific time | Anything outside the frame | 15 to 30 seconds per photo |
| Supervisor quality inspection | Sampled quality against a written standard | Quality on the nights you did not inspect | 20 to 45 minutes per site |
| Client confirmation or portal view | The client saw the record and did not dispute it | That the client read it carefully | Automatic once set up |
Before you track anything: five prerequisites
You cannot measure completion against a scope that lives in a PDF nobody has opened since the contract was signed. Fix these first or your completion rate will be a number with no meaning behind it.
Setup checklist
- A task-level scope of work per site: not "clean restrooms" but the actual line items, each tagged nightly, weekly, monthly or quarterly.
- One system of record for the schedule: if the schedule exists in a spreadsheet, a group text and someone's memory, completion data will never reconcile.
- Confirmed site addresses and entry points: including which door the crew actually uses, because that determines where a geofence has to reach.
- A labor budget in hours per shift: derived from production rates, not from what you have always billed.
- A named owner per shift: one person who is accountable for that building that night, even on a two-person crew.
The step-by-step system
Step 1: Turn the scope into a nightly task list
Take the contract scope and split it by frequency. A 62,000 sq ft medical office building might have 22 nightly line items, 9 weekly, 5 monthly and 3 quarterly.
The crew should only ever see what is due tonight. A 39-item list every night trains people to check boxes without reading them, which destroys the value of the data you are collecting.
Group the nightly items by area, not by task type. Cleaners move through a building room by room, so a list ordered restroom by restroom gets used. A list ordered by "all trash, then all dusting" gets ignored.
Step 2: Set a time budget for every shift
Completion tracking without a time budget produces a familiar and useless result: everything marked complete, every night, in half the expected hours.
Build the budget from production rates. ISSA publishes task-level cleaning time standards that give you defensible starting numbers, and APPA's five levels of clean let you tie the rate to an agreed service level rather than to opinion.
Example: your own historical data says a crew covers about 3,500 sq ft per labor hour in open office space at APPA Level 2. At 62,000 sq ft you budget roughly 18 labor hours. Split across a four-person crew, that is about 4.5 hours each.
Step 3: Define the site boundary and the grace windows
A geofence is a radius around the site coordinates. For a standalone building with its own lot, 100 to 200 meters usually covers the parking area and the service entrance without reaching the coffee shop across the street.
Tighten it for dense downtown blocks where two client buildings sit within a block of each other. Widen it for campuses, hospitals and industrial parks where the crew parks a quarter mile from the door.
Then set two windows in writing: how late a clock-in is still considered on time, commonly 10 to 15 minutes, and how far under the budgeted hours a shift can run before it is flagged. Without those numbers, every deviation becomes a judgment call at 7 a.m.
Step 4: Capture completion at the task level, not the shift level
A single "shift complete" button gives you no diagnostic value. When the client complains, you still cannot tell which item was skipped.
Task-level check-off tells you exactly where the shift broke down. Over a month it also shows you patterns: the same three items get skipped on Fridays, or the loading dock never gets done when a specific person covers.
Require photos on the three to five items that generate the most complaints at that site. In most office buildings that means restroom floors, the break room, entrance glass and trash staging. Photo requirements on all 22 items will not survive week two.
Step 5: Build an exception report, not a completion report
If you have 40 shifts a night, nobody is reading 40 records. You will read the five that broke.
Your exception report should surface only these: no clock-in by the grace deadline, clock-in outside the geofence, shift hours more than 20 percent under budget, any incomplete required task, and any shift with no task activity at all.
Everything else is silence, and silence should mean the shift ran clean. That is the only way an operations manager can supervise 40 sites without working nights.
Step 6: Verify a sample with real inspections
Self-reported completion drifts upward over time. Inspections are what keep the number honest, because the crew knows the claim can be checked.
A workable sampling rule: every site inspected at least monthly; sites in their first 90 days inspected weekly; any site with an open client complaint inspected weekly until it posts two consecutive passing scores.
Score against the written standard with a numeric result, and attach photos to anything below passing. An inspection that produces the word "fine" is not data.
Step 7: Put the record in front of the client before they ask
The operators who never lose accounts to a completion dispute are the ones whose clients can already see the record. A facility manager who can pull up last Tuesday's shift and inspection score stops treating you as a suspect.
Give the client access to schedules, completion history and inspection scores, or send a monthly summary. Either works. Guessing does not.
The five numbers that tell you whether it is working
Track these monthly, per site and per supervisor. The targets below are reasonable internal starting points, not industry benchmarks. Set your own once you have 90 days of your own data.
| Metric | Formula | What it catches | Starting internal target |
|---|---|---|---|
| On-time start rate | Shifts clocked in within the grace window ÷ scheduled shifts | Chronic late starts before the client notices | 95% or better |
| Coverage rate | Shifts with the required headcount on site ÷ scheduled shifts | Silent short-staffing on hard-to-fill routes | 98% or better |
| Task completion rate | Tasks marked complete ÷ tasks due | Which specific line items get skipped | 99% on nightly items |
| Hours variance | (Actual labor hours − budgeted hours) ÷ budgeted hours | Shifts finishing impossibly fast, or bleeding overtime | Within ±10% |
| Verified completion rate | Inspected shifts that passed ÷ shifts inspected | The gap between what is claimed and what is true | 90% or better |
The gap between task completion rate and verified completion rate is the single most useful number in this whole article. A site reporting 100 percent task completion and 62 percent inspection pass rate has a checkbox problem, not a cleaning problem.
Exception thresholds worth acting on the same day
| Signal | Trigger | Same-day response |
|---|---|---|
| No clock-in | 15 minutes past scheduled start | Call the cleaner, then the backup, then dispatch a supervisor |
| Clock-in outside boundary | Any occurrence | Verify address and geofence radius before assuming misconduct |
| Short shift | More than 20% under budgeted hours | Compare against the task list and pull photos for that night |
| Missing required task | Any nightly item unchecked | Schedule catch-up on the next shift and notify the account manager |
| No task activity | Clocked full hours, zero items checked | Treat as unverified and inspect within 48 hours |
| Repeat exception | Same site or person, 3 times in 30 days | Retrain, re-time the route, or reassign |
Common mistakes
- Treating clock-out as completion: it is an attendance record. Payroll needs it. Your client does not care about it.
- Letting the checklist grow until nobody reads it: if the nightly list takes more than two minutes to work through, it will be batch-checked in the parking lot.
- Collecting data nobody reviews: a completion log that is only opened after a complaint is a legal record, not a management tool. Read the exception report every morning.
- Setting geofences too tight: a 25 meter radius on a hospital campus generates dozens of false failures per week and teaches supervisors to ignore alerts.
- Skipping inspections once completion rates look good: that is exactly when self-reported numbers start drifting.
- Rolling it out without telling the crew why: frame it as proof of work that protects them from unfounded complaints, because that is genuinely what it does. Explain what is captured and what is not.
- Sloppy time records: the Fair Labor Standards Act requires accurate records of daily and weekly hours worked. Retrofitting timestamps after the fact is a payroll exposure, not a shortcut.
How often to review what
| Frequency | What you review | Who owns it |
|---|---|---|
| Daily, before 9 a.m. | Last night's exception report only | Operations manager or dispatcher |
| Weekly | Repeat exceptions, open complaints, hours variance by site | Operations manager with supervisors |
| Monthly | The five metrics per site, inspection scores, client-facing summary | Owner and account managers |
| Quarterly | Time budgets vs. actual hours, scope creep, checklist pruning | Owner |
| At contract renewal | Full completion and inspection history for that account | Owner and account manager |
The quarterly review matters more than operators expect. Scopes creep, buildings change occupancy, and a task list built 18 months ago is quietly measuring the wrong work.
Prune it. Any nightly item with a 100 percent completion rate and zero related complaints for two quarters is either genuinely automatic or nobody is checking it honestly. Find out which.
How CleanTrack360 supports this
CleanTrack360 puts the layers described above in one place. Crews clock in and out through the phone browser inside a geofence, with a default 150 m radius you can adjust per location, so your campus sites and your downtown sites can have different boundaries. Location is captured at clock-in and clock-out, not continuously between them. Quality inspections run on custom checklists with photo evidence and automatic scoring, which gives you the verified completion number that keeps self-reported data honest. Reports export to CSV when you want to build your own variance math.
Clients get a browser-based dashboard showing schedules, inspection reports and service requests, which is usually enough to end "were you here Tuesday" conversations for good. Plans are Starter at $99/month for up to 5 team members, Pro at $199/month for up to 20, and Business at $249/month for up to 50, priced per plan rather than per user. There is a 14-day free trial with no credit card. The native mobile app is still in development, so crews work in the phone browser today.