> ## Content Index > Fetch the complete content index at: https://beathantechnologies.com/blog/llms.txt > Use this file to discover other available public pages before exploring further. # Your client says the guard was not there. Can you prove otherwise? - URL: https://beathantechnologies.com/blog/proving-guard-attendance/ - Published: 2026-09-24T08:00:16.000Z - Updated: 2026-10-05T12:29:00.000Z - Description: If your answer to a disputed shift is a timesheet your own supervisor typed up, you have already lost the argument. Here is what a record needs to contain instead. - Author: Liam van Huyssteen - Tags: Sentry, Security operations, #kw: security guard tracking Your client says your guard was not on site on Tuesday night. You believe they were. What does your security guard tracking let you send them? If the answer is a timesheet your own supervisor typed up, you have already lost the argument, because it is your word against theirs and you are the one being paid. A defensible record answers four questions without anyone having to trust you. ## Who Not "which employee number was entered". Who. An employee number proves somebody knew a number. An NFC tag proves somebody held a tag. Neither proves identity, which is why the method matters. [Face recognition](https://beathantechnologies.com/sentry/face-recognition) puts a person against the record. For the other methods, a selfie taken at the moment of clocking does the same job with less machinery: the photograph sits in the timesheet row and anybody can look at it. This is also the honest answer to buddy clocking. Face recognition reduces it. It does not eliminate it, and anyone telling you their system makes it impossible is selling you something. ## Where, and how confident This is where most systems are weaker than they appear. A GPS coordinate on its own is not evidence. Every phone will give you a coordinate. The question is how good that coordinate was, and most systems throw the answer away. [Sentry](https://beathantechnologies.com/sentry) stores the accuracy of the fix alongside the fix itself. A clock-in recorded at fifteen metres is a different fact from one recorded at three hundred, and when the two are stored identically you cannot tell them apart six weeks later. It matters most at the edges. When a guard is just outside the [geofence](https://beathantechnologies.com/sentry/roll-call), the app does not silently refuse and leave them arguing with a screen. It tells them which site is nearest, how far away they are, and what the site's radius is. If the miss is inside the margin of their own GPS fix, it says so, and points them at the site's NFC tag instead, which does not care about satellites. That turns an argument about trust into a conversation about a number. ## When Covered properly in the [previous post](https://beathantechnologies.com/blog/clock-in-no-signal/), so briefly: the recorded time has to be the time the guard acted, not the time the record reached the server. If your system stamps the upload, a shift that started in a dead zone is recorded hours late and your own records will contradict you. ## What happened during the shift Attendance on its own is a thin story. A client disputing a night shift is usually disputing whether anything was done, not merely whether someone was present. The [occurrence book](https://beathantechnologies.com/sentry/occurrence-book) is the answer, if it is digital and timestamped. Entries made during the shift, photographs attached, each one carrying its own time and position. A patrol logged at 02:14 with a photo is harder to argue with than a signature in a book that could have been written at any point that week. ## Putting security guard tracking together What you send the client is not a screenshot. It is a record with: - the identity, with a photograph or a face match - the position, with its accuracy - the time the guard acted, and separately the time it reached you - the occurrence entries made during that shift - an audit log showing nobody edited any of it afterwards That last one is the part people forget. A record that your own admin can quietly change is not evidence. If your system does not log who changed what and when, the client is entitled to ask why they should believe the rest. ## The uncomfortable part Sometimes you will pull the record and the guard was not there. That is the point. A system that always produces a flattering answer is a system nobody should believe, including you. The value of a defensible record is that it settles the question in both directions, which is also what makes it worth sending to a client unprompted after an incident rather than only when challenged. The operators who get the most out of this are the ones who send the record before the client asks. --- *Next:* [*what PSIRA and POPIA actually require you to keep*](https://beathantechnologies.com/blog/psira-popia-records/)*, and for how long.*