Build the summary from unique person-session records. Keep someone who was not scheduled separate from a scheduled participant whose attendance is unknown.
Define the counted population first
Use a person ID, session ID, scheduled flag, attendance status, record date, and source reference. Names are unnecessary in a sanitized planning example. The pair of person ID and session ID identifies the record; the same person can legitimately appear in different sessions.
Choose whether Late counts as attended for this summary, and write that choice beside the total. This example counts Present and Late as observed attendance. It does not impose an organizational policy or change the underlying status.
An eight-slot volunteer workshop
Fictional demonstration data: session V1, October 13, 2026. Eight participants were scheduled. A ninth person was not scheduled and stays outside the denominator.
| Person-session ID | Scheduled? | Recorded status |
|---|---|---|
| P01-V1 | Yes | Present |
| P02-V1 | Yes | Present |
| P03-V1 | Yes | Present |
| P04-V1 | Yes | Present |
| P05-V1 | Yes | Present |
| P06-V1 | Yes | Late |
| P07-V1 | Yes | Absent |
| P08-V1 | Yes | Unknown |
| P09-V1 | No | Not scheduled |
The scheduled status counts reconcile: 5 present + 1 late + 1 absent + 1 unknown = 8 scheduled slots. Six of eight scheduled slots have recorded attendance, or 75%, under the stated late-counts-as-attended rule. Label that recorded attendance so far, not a final attendance rate, because P08-V1 is unresolved.
The one unknown is 12.5% of eight scheduled slots. P09-V1 is not part of either calculation. Display it separately so a reader can verify the exclusion rather than wonder whether a row disappeared.
If you also show a known-record-only measure, its denominator is seven: six attended out of seven known scheduled records, about 85.7%. That answers a different question. For this review, keep the scheduled denominator of eight prominent and avoid substituting the higher percentage for it.
Drill from a count to its source rows
Clicking or selecting “Unknown: 1” in a reading dashboard should identify P08-V1 and the missing status. The action is to check the approved attendance record or ask its recorder; an empty cell does not establish absence.
Before summarizing another session, check for duplicate person-session IDs. If P06-V1 appears twice, stop that session's total for reconciliation instead of counting two late arrivals. Do not silently choose the latest row unless the record owner has provided that rule.
When switching sessions, change the heading, date, denominator, totals, and corresponding roster together. A tally for V2 above rows from V1 would be an incorrect result even if both tables looked plausible.
Draft a summary from your approved roster
Create an attendance dashboard planning summary in US English.
Sessions and dates: [enter]
Approved sanitized records: [paste person ID, session ID, scheduled flag,
recorded status, record date, and source reference]
Rule for Late in the attended count: [enter]
Authorized correction process: [enter]
Identify duplicate person-session keys before calculating totals. Preserve
Present, Late, Absent, Unknown, and Not scheduled as distinct states. Use
scheduled unique person-session slots as the primary denominator. Show the
late-count rule and reconcile subtotals to that population. Unknown is not
Absent; Not scheduled is excluded and displayed separately. If unknowns
remain, label attendance a recorded-so-far measure, not a final rate.
Do not invent statuses, reasons for absence, names, or corrected records.
Return the totals, corresponding source rows, and correction questions.
If requested, make an HTML reading copy with session selection and status
details; changing session must update all matching totals and rows. Do not
collect sign-ins, publish private personnel data, or make HR/payroll decisions.
Finish by sending the record owner the exact unresolved key and any duplicate keys, using the approved channel. Recalculate only after corrections are recorded. The coordinator should leave with a reconciled summary or a clearly identified gap, not a percentage that conceals missing attendance.
Optional: read a sanitized attendance summary in NOS
NOS can save and display an HTML summary for a workshop-planning review. Use it to read the scheduled count beside its source rows. Attendance entries and corrections remain in the approved roster or spreadsheet.