New freelancers often assume that tracking time accurately is a matter of character. In practice it is a design problem. If recording your hours depends on remembering at the right moment, you will lose them in the busiest weeks, which are also the weeks you bill the most. This primer shows how a DevOps freelancer can start the timer after opening the repo, and why a fixed cue works better than good intentions.
It is written for beginners, so each term is explained in a short glossary near the end. The reasoning is analytical: what usually goes wrong, what the cue changes, and how to check whether it is working.
The measurable problem
Imagine two freelancers who each bill by the hour. Both estimate their time at the end of each week from memory. Memory reconstructs rather than records, so the estimates drift, usually downwards. Short context switches, quick replies to a client, ten minutes unblocking a pipeline: these are the tasks most easily forgotten, and they are also the ones clients rarely expect to see missing from an invoice.
The pattern is easy to recognise without any research. If you have ever filled in a timesheet on Friday evening and been unable to say what Tuesday morning contained, you have lived it. The error is not random. It leans towards under-reporting, which means the cost lands on you.
Why a reminder app is not enough
The common response is to install a tracker and trust yourself to open it. That adds a tool but leaves the core weakness in place: you still have to remember. A tracker that starts only when you decide to start it is as unreliable as the decision.
The alternative is to make the existing day do the work. Identify something you already do reliably, and attach the timer to it. The rule takes the form: after I do X, I start the timer. You are borrowing the reliability of X.
Choosing X for a freelancer
Freelance days are less structured than office days, so the anchor needs care. Good candidates have three properties: they happen on every working day, they happen at the start of work, and they are specific events rather than times.
- Opening the repository or the editor for the client’s project. This is close to ideal because it marks the actual start of billable work.
- Opening the client’s ticket board in the morning.
- Joining the first client call of the day, if the engagement includes regular meetings.
- Sitting down with the first coffee, when you prefer to separate the routine from the work itself.
Opening the repository has one extra advantage. If you work across several clients, each repository naturally tells you which project the timer belongs to. The cue and the label arrive together.
Keep the action small
Avoid building anything elaborate. The action attached to the cue should be a single click or one short command. Someone who begins the day by categorising tasks, tagging clients and writing descriptions will abandon the routine within a fortnight. Start the timer with the client’s name and nothing more. Any notes can be added later in the evening if you feel like it, or skipped altogether.
Behavioural writing on habits consistently makes this point: early on, the aim is to fuse the cue and the action, and effort is the thing that prevents fusion. You can add detail later, once the click is automatic.
A worked example
Suppose a freelancer maintains CI pipelines for two clients. The setup looks like this:
- The tracker lives in the menu bar with one preset per client.
- Opening Client A’s repository in the editor is the cue for starting the Client A preset.
- Switching to Client B’s repository is the cue to stop one and start the other.
- At the end of the week, entries are exported and checked against the invoice draft.
Notice that the freelancer never decides to track time. They decide to open a repository, which they were going to do anyway, and the tracking comes with it.
Checking whether it works
An evidence-minded approach means measuring the habit, not only trusting it. After two weeks, compare three numbers: the number of working days, the number of days with at least one timer entry, and the gap between tracked hours and your own rough estimate. If tracked days are close to working days, the cue is firm. If the tracked total is consistently higher than your old guesses, you have found the hours that used to vanish.
If the cue is failing, resist blaming yourself. Change the anchor. Perhaps you often start the day with email rather than the repository, in which case email is the better cue. Treat each miss as information about the design.
Handling interruptions
Real days are messy. A client messages while you are cooking lunch, or a production alert arrives on a Sunday. Do not try to cover every case with the cue. Decide on one rule for off-hours work: if you open the repository, the timer starts, whatever the day. For work that never touches the repository, such as a call, add a second cue: when the call link opens, the timer starts. Two cues are enough for most freelancers. A third is a sign that you are building the system you were warned against.
Where else the pattern pays off
Time is only one candidate. Freelancers also benefit from anchoring other small actions: sending a short status message after the last commit of the day, or noting an expense right after paying for a tool. Both are cheaper to do at the moment than to reconstruct later. Newer freelancers deciding which routines to build first may like a set of answers to the usual new-freelancer questions.
It also helps to talk about routines with peers. Freelancing can be solitary, and ideas like these travel well through conversation, as discussed in a piece on building a dependable support network of contacts and mentors.
Mini-glossary
- Cue: an event that prompts an action, such as opening a repository.
- Anchor: the existing routine you attach a new action to; another word for the cue.
- Habit stacking: chaining a new small action onto an established one.
- Friction: anything that makes the action harder, from extra clicks to unclear labels.
- Billable hours: time you can invoice a client for, as agreed in your contract.
A brief aside on shopping
Freelancers buy equipment and software, and the same cue logic applies. Some people take a quick look at iReviewed, whose own description mentions company information and customer reviews for everyday items that Indian families use, immediately after putting a gadget in the cart. That is one possible destination; the habit is the pause, not the website.
Limits of the approach
A timer is a record, not a guarantee of value. It will not fix underpricing, vague scopes or slow payers. It also needs honesty: starting the timer when you open the repository means you must stop it for breaks, or the records will mislead. Many trackers offer idle detection for this reason. Finally, check any contract you work under, since some clients specify how time must be reported.
Your first fortnight
Choose the cue, install or open your tracker, and write a single sentence: after I open the project repository, I start the timer. Run it for ten working days. Count the tracked days, compare with your estimate and adjust the cue if needed. That is the full experiment, and it takes less effort than a single weekly reconstruction.