Encouraging workers to clock in on-time
Context
About Instawork
Instawork is an on-demand staffing platform for blue-collared workers (catering & warehouse work), where they can choose when and how much they want to work.
Project background
When an Instawork Professional arrives at the location of the customer (typically a catering venue or a warehouse), they start the shift by clocking in on the app, and once the shift ends they clock out. Their earnings are calculated based on the hours worked and paid out through Instawork.
Problem statement
One of the most common types of support tickets we used to get was around the clock-in/clock-out timings. Such tickets used to come from both businesses and professionals, because it directly impacted how much they got charged/paid.
However, this meant a lot of manual effort checking the logs, communicating back-and-forth with both parties and resolving the dispute. More than that, this caused unnecessary confusion for both the professional and the business and could lead to a lack of trust.
Hypothesis
One very likely reason for all this confusion was the way the users clocked-in in their mobile app. Here’s how it worked:
- The app starts tracking the user’s location a little while before the start-time of the shift
- Once the user is at the shift venue, it clocks them in automatically. But this would only work as long as location data is accurate. Places with bad connectivity could mean wrong clock-in times.
We had preliminary data from earlier user research to validate this. The support tickets confirmed this too.
Problem deep-dive
The automatic clock-ins caused two kinds of problems:
- Early clock-insThe user would be clocked in as soon as they’re in the vicinity of the location, which sometimes meant they’re still on the way, or finding parking. The business would be notified but they wouldn’t be able to find the professional yet.
- Late clock-insThis happened mostly when the automatic clock-in failed due to bad connectivity. As this doesn’t happen in most cases, the professional wouldn’t expect this. By the time they notice they’re not clocked in yet, they can manually clock in, but it’s late already.
Real problem
Yes, the clock-in data was wrong, but more specifically the recorded timings were earlier than the actual clock-in times in most cases. So, the actual instances of late clock-ins were probably more than what the data showed.
Late clock-ins affect both the professionals and the business partners. As business partners rely on the professionals for their events, each professional turning up late affects their ability to start things on time, which in turn affects their credibility. On the other hand, it had cascading effects on the professional’s livelihood — when they receive bad ratings for being late, it creates a lack of trust and hurts their future earning opportunities.
Goal
- Reduce instances of late clock-ins
- Make clock-in and hours calculation transparent
Step 1 — Fixing data
If we were to remove the automatic clock-in and ask professionals to clock in manually every time, it would give us more accurate data, but would mean a change of behavior for those who’re used to the current system. However, in the interest of making things transparent to the users while getting more accurate data to improve on, this is what we implemented at the risk of this short-term impact.
You can only improve what you measure.
Well, only if you measure it correctly.
Step 2 — Research
As an organization, we value basing our product decisions on data and user research.
Also, Instawork has a high bar for quality, the professionals are activated only after different quality checks. So we strongly believe that if a professional clocked in late, it is most probably due to an external factor, which we could help with. But we had to understand these factors first.
Data
We started with a hypothesis of different parameters that might cause late clock-ins — including experience level of the professional, day of the week, hour of the day, location of the event, etc — and started slicing the late clock-in numbers by each parameter. Two of them stood out as strong correlations.
- Time of the dayGigs that start early in the morning had significantly higher tardiness
- VenueThere were specific venues that had a higher rate of late clock-ins. These are mostly large venues like stadiums and golf clubs, which typically have multiple entrances and are tricky to navigate, or where the workplace is far away from parking.
User interviews
We also sent out surveys and interviewed users to understand this deeper. This confirmed the learnings from data but also explained why that was the case.
- Not able to clock in even after reaching the venue, because finding the exact place/person to clock in with is not clear at some big venues.
- When the shift is early in the morning, the reminders we send a couple of hours before the shift were not very useful.
Step 3 — Solution
Quoting an Indian statesman,
If you want people to do good, you need to make it easy to do good, and extremely hard to do bad things.
Once you understand the reasons for the current behavior, you can help change it. Taking the learnings from the research, we started tackling them one by one.
Helpful reminders
Prev evening SMS for early morning gigs
For shifts that start early in the morning, we started sending reminders the previous evening, along with an estimated time to leave home to arrive on time.
Commute time SMS before the gig
For other shifts, we were already sending reminders a couple of hours before, but now started including the time to leave based on their location.
Left — commute time reminder. Right — previous evening reminder for early morning shifts
Addressing complicated venues
When a business posts a shift at a venue that’s known to cause confusion navigating, we suggest them to adjust the map marker to the exact clock-in location and add clear navigation instructions.
When a professional is looking to book such a gig, they’re shown a heads-up too so they can be prepared and arrive a little earlier.
Transparency
When it’s close to the shift start-time, the app shows a real-time view of the time and distance to cover, giving the professional complete transparency of the clock-in time.
Results
As we implemented these changes one by one, the instances of late clock-in started reducing gradually, and eventually by more than half.
