Screen capture vs task logs: Tracking AI work without NDA breaches
Continuous screen recording endangers client NDAs, while event-driven logging provides clean proof of work.
Use session records, task counts and estimated earnings to set a realistic pace and decide when a work block is no longer paying off.
A daily target is useful only if it fits the work actually available. Task complexity, platform rates and batch availability can change from one session to the next. Instead of setting one task-count goal for every platform, build targets from your own recent sessions and compare like with like.
RaterSidekick tracks work sessions, completed tasks and estimated earnings across AI contractor platforms, including micro1 and DataAnnotation. That gives you a record to review instead of relying on memory. It does not tell you which platform will have work tomorrow, and a tracked estimate is not the same as a confirmed payout.
For several workdays, record sessions on the sites you choose to track. Keep project profiles and platforms distinct where the work differs. A batch of short classification tasks is not a fair benchmark for a project with longer written responses, even if both show the same task count.
RaterSidekick detects task completion using a keyboard shortcut you select. Use the shortcut you normally use to submit work, and designate only the websites you want it to track. The extension does not capture screenshots, keystrokes or task content. For a fuller discussion of why event-based work records can be preferable to screen capture, see the case for tracking task events without recording the screen.
After a few comparable sessions, note the time spent and tasks completed in each. If you have platform-reported pay for the same period, keep that alongside the estimates. Separate active work from breaks or time spent waiting for a new batch; otherwise, a slow supply of tasks can look like a slow work pace.
Pick a session length that suits your schedule, then use it as a trial block rather than a universal rule. For example, start with a focused hour on one project. Record the tasks completed and estimated earnings, then compare that block with other sessions on the same kind of work.
Build two targets:
Use a range, not a hard quota. If your recent sessions vary, a target near the middle of that range is more realistic than using your best session as the daily expectation. Revisit the target when the task type, project or platform changes. Do not treat a low count as a personal performance problem when the batch itself is slower or more complex.
Task rate is simple to calculate: divide completed tasks by focused hours. If you complete 18 comparable tasks in 90 minutes, that is 12 tasks per hour. The figure is useful for comparing similar sessions, not for ranking unlike projects.
To calculate an effective hourly rate, divide earnings for a period by the focused work hours in that same period. Label it as an estimate until platform reporting or payment confirms the amount. Use platform reports for actual pay, and make sure the hours and earnings cover the same dates.
RaterSidekick’s Pay Checker compares your records with platform reports for time, tasks and pay, period by period. Use it to spot a mismatch worth investigating, not as a substitute for checking the platform’s own records or pay rules.
One disappointing block is not enough evidence to abandon a project. Compare at least a few sessions with similar task types and time conditions. If task pace and estimated earnings stay below your own acceptable range, consider allocating the next available block elsewhere, provided you remain eligible and follow each platform’s requirements.
Keep the comparison practical. A platform with a higher estimated hourly rate may have fewer available tasks; another may offer steadier work. You can compare both the rate and the time you actually had work to do. Avoid counting an idle wait as productive session time, but do not assume that switching will produce a better batch.
At the end of the day, compare each project’s sessions, task counts and estimated earnings. Note which blocks met their target and which did not, along with any change in task type or availability. Use that pattern to set tomorrow’s session plan. A target should guide your allocation of time, not pressure you to rush tasks or ignore a platform’s quality requirements.
Because RaterSidekick keeps tracked work separate from platform reports, use the platform’s records to verify actual pay. The useful outcome is a clearer comparison: how long similar work took, how many tasks you completed and what the period appears to have earned. Adjust targets from that evidence, one comparable session at a time.
Continuous screen recording endangers client NDAs, while event-driven logging provides clean proof of work.
A practical guide to Chrome extension permissions, site isolation, and keeping work records safe across AI annotation portals.
Consolidate legacy spreadsheet records and Clockify exports into a unified work history hub without losing your historical earnings data.