In an office, attendance takes care of itself: people are in or they are not. Remote, the question splits in two. Is this person working today, and roughly when? And is the time they record as work actually spent working? The first is attendance; the second is where most teams overreach, buy a monitoring tool, and discover that measuring everything makes people trust the company less and game the numbers more.
This guide is about doing the first well and the second proportionately. It is written for small teams — software houses, agencies, startups — where the owner wants to know the day is being worked without becoming the person who watches screens.
Decide what you actually need to know
Before choosing a tool, write down the question you are trying to answer. Three answers come up most often, and they need different amounts of tracking:
| What you need | What answers it | How intrusive |
|---|---|---|
| Who is working today, and when they started and stopped | Clock-in and clock-out with breaks | Low |
| How many hours went to each client or project | Time recorded against a project or task | Low |
| Whether recorded time was active | An idle policy, and screenshots only where justified | Medium to high |
Most teams need the first two rows and only occasionally the third. The difference between attendance, time tracking and timesheets is covered in more depth in attendance vs time tracking vs timesheets.
Make clocking in the default, not the exception
Remote attendance starts with a clock-in that takes one click from wherever people already work. A browser clock-in, with break types your organisation defines — lunch, prayer, a school run — gives a truthful record of the day without anyone having to explain it later. Ask people to pick the project or task they are on when they start, and the same record answers the "which client" question as well.
Two rules make the record reliable. First, corrections go through an approval rather than an edit, so a forgotten clock-out becomes a request a manager approves and both versions are kept. Second, leave and holidays live in the same system, so a day off is never mistaken for an absence.
Use an idle policy instead of watching
The honest version of "is the time active?" is an idle policy: if there is no activity for a set number of minutes, the clock pauses. It is fairer than watching because it is the same rule for everyone, written down in advance, and it removes the need for a manager to make a judgement call about any one person.
The details matter. A good idle policy pauses the timer from the moment inactivity began, not when it was noticed. It lets the person undo a pause within a window if they were working away from the keyboard — reading a document, in a call, thinking on paper. And the decision to pause is taken by the server from the evidence every open device reports, not by one browser tab that happens to be in the background. In AMS, that is how it works: the organisation sets the threshold, people can say "I was still working" within the undo window, and once that window has closed an approver can restore the time.
Our idle time tracking page sets out exactly what the policy measures and what it never records — it reports activity timings, not what anyone typed.
If you use screenshots, make them proportionate
Periodic screenshots are the most intrusive option and the one most likely to damage trust, so use them where there is a reason — client billing that must be evidenced, a contract that requires it — and not as a default for everyone.
- Tell people before you switch it on, what it captures, how often, and who can see the images.
- Exempt the roles that do not need it. Per-person exceptions mean the policy covers the people it is for, and nobody else is asked to share a screen.
- Never capture breaks. Capture should stop when the clock is paused or on a break.
- Limit who can view team screenshots, and give each person access to their own.
- Watch capture health, not just captures. A list of people who are clocked in but not being captured — declined the prompt, stopped sharing, capture failed — tells you about a broken setup before you misread a quiet day as an idle one.
The time tracking with screenshots page shows how AMS handles each of these, including what it deliberately does not do.
Report on outcomes, review the exceptions
The value of attendance data is in the pattern, not the minute. Look weekly at hours per project against what was planned, at late starts that repeat, and at days with long unexplained gaps — then have the conversation with the person, not with the spreadsheet. A dashboard that shows everyone everything invites exactly the micromanagement the exercise was meant to avoid.
And close the loop: when attendance feeds payroll directly, nobody re-keys hours at month end, and the record people see is the one they are paid from. That is the strongest reason for people to keep it accurate.
A short policy you can adapt
- Clock in when you start work and out when you finish; record breaks with the right break type.
- Record time against the project or task you are working on.
- The clock pauses after the agreed minutes without activity; undo a pause if you were working away from the keyboard.
- Corrections are requests, approved by your manager; both versions are kept.
- Screenshots, where used, apply only to the roles named in the policy, never during breaks, and are visible only to the people named in it.
Built for your industry
