Case Studies

Hourly contractors paid for work, not just logged time

AI training data · Hourly contractor workforce

An AI training-data company pays a large team of hourly contractors to label and review data, most of them now remote. Its operations leaders needed to know that every paid hour bought real work. Time Doctor checks the hours against the work and flags activity faking within the weekly pay cycle, with the ground rules explained to contractors upfront.

Industry
AI training data
Workforce
Global workforce of remote and hybrid hourly contractors
Team
Operations, with data labelling and review teams
Focus
Paying only for verified work
10%+productivity increase in the first team measured after go-live
1 weekfrom go-live to the first activity faking flagged and reviewed, inside the pay cycle
$10K/monthgoing to time nobody could verify, by the Director of Operations' estimate
The challenge

What they couldn’t see

The company produces training data for AI. Its labelling and review work is done by hourly contractors, a team that grows and shrinks with project volume. With most of the team moving to remote or hybrid work, operations leaders were paying for hours they could no longer see.

Their own clock-in tool said some contractors were idle for half a shift while they were still submitting tasks, so nobody trusted the number either way. Output varied widely between people on the same shift, and nothing could spot auto-clickers or other tools that fake activity. The Director of Operations estimated that, if the clock-in data was right, as much as $10,000 a month could be going to unverified time.

Why Time Doctor

Why they chose Time Doctor

  • A check on every paid hour. Time Doctor runs alongside the company’s own clock-in system, which stays the pay record, and shows what the paid time was spent on.
  • Activity faking flagged inside the pay cycle. Unusual Activity reports flag patterns such as automated clicking and keys held down in 30-minute windows, so leads review them before the weekly pay cycle closes. The team lead also reviews flagged patterns in Claude through Time Doctor’s AI connection.
  • Rolled out with contractors, not at them. Screen recordings stay off, contractors start and stop their own timer, and personal time stays personal.

If we pay someone with $10, we would want to get $10 worth of work back.”

Director of Operations
How they use it

Putting it to work

  • Paid for productive work only

    The rule is simple: press play when working, stop for anything personal. Team leads walked every contractor through it before go-live.

  • Flags reviewed with the people involved

    Leads check each flag and talk to the contributor. Legitimate patterns, such as holding a key while correcting labels, are cleared; deliberate faking is not.

  • Quotas tested against real hours

    Seeing where idle time builds up has the team asking whether its task quotas fit the hours it pays for.

  • Insights into the reports Finance already uses

    Time Doctor data sits next to the company’s own task-output figures, so the question becomes whether the productivity gain pays for itself.

Results

What changed

  • The first team measured improved its productivity by at least 10% after go-live, measured on the company’s own task data.
  • In the first week, flags surfaced automated clicking and keys held down for up to 30 minutes, and leads confirmed them with the contributors involved. Flags are now reviewed within every weekly pay cycle.
  • Team leads are re-checking task quotas against the hours actually worked.
  • Every paid hour is now checked against the work behind it, covering the $10K/month the Director of Operations estimated was going to unverified time.

“It makes us recheck… if our set quotas are good enough. Like, given if there was high idle time, are our quotas enough for the hours that we had set?”

Team Lead, Computer Vision
What’s next

Where they go from here

As the contractor team grows, the company is looking at the same approach for its other teams.

Pay for the work, not just the hours.

Workforce visibility for distributed teams, trusted by 10,000+ distributed teams.

Book a demo →