When I used analytics to resolve a dispute

Long back, I used to travel by office cab, and the person who got off the cab last in the evening, happened to be the "route monitor", who would decide the best route the cab takes. Over a period of time, he observed areas where traffic was frequently causing more delays, and suggested variation in the route. Since the new route required a female employee to get off the cab earlier and walk along a long stretch of lonely road, she protested, and the matter went to HR and Admin, as a women's safety issue. Meetings were held, where they reached a stalemate because nobody could prove anything. 

Observing that there was a lack of data, I stepped in and asked for two weeks to objectively analyze the route timings. The first week, we took the new route, and the second week we took the old route. This is the rough picture of the data I gathered:

Route analytics. Click image to view it larger.

At first, I noted the timings only for a few prominent points where colleagues were dropped off (grey circles). As I noticed a fluctuation in time, I added more landmarks. The results were surprising.

When the cab reached the point shown by the blue square, we had the option of taking route 1 (the new route suggested) or route 2 (the old route). Both routes would eventually lead to the point shown by the yellow circle. The route monitor's assumption was that the time delay between the blue and yellow points, was what caused everyone else to reach home late. However, the data revealed an entirely different picture. These were the points I noted:

• Route 1 avoided initial delay: The blue and yellow points are consistently closer for route 1. So there was indeed some time saved by avoiding route2's traffic signal.
• Route 1's initial time saving futile: On the Thursday of Route 1, we reached the final point (my stop) at 7:50, despite there being a very short time delay between the blue and yellow points.
• Route 2's initial time loss insignificant: Even though the Thursday of Route 2 took up some time between the blue and yellow points, I still reached early, a little after 7:30.
• First point mattered: The very first grey circle was the first landmark we reached after leaving office and crossing a frequently jammed stretch of road. The slight fluctuation of the first point appeared to match with the massive fluctuations of the other points on the same horizontal line. This indicated that there were certain days when more people in other companies left their office sooner, causing more jams at various parts of the city, which caused delays at various points.
• Rain irrelevant: The rain didn't cause any significant delay.
  
• Travel after sunset: One of the complaints raised was that the lady would sometimes have to walk on that stretch of road after sunset, which made her uneasy. The diagram above captures the various times of the year sunset happens, and the arrows show a few other colleagues who also had to travel a bit in the dark after getting off the cab to reach home.
• Insufficient data: The patterns showed great fluctuation, which led me to conclude that we needed more data to make a proper conclusion. However, the existing data was sufficient to make a preliminary conclusion.

Resolution

We concluded that we could continue using the old route. However, the initial arguments between both parties and their rallying of other colleagues for support, created a gloomy mood in what was earlier a happy group. During the two weeks, I noticed people attempting to influence the outcome of the dispute. Still, in the end, the data showed everyone that it was better to verify facts than make assumptions. It led to a healthy resolution. The incident also showed that no matter how capable a leader is (like the route monitor), there will come situations when the team goes against the leader. Such situations need to be handled carefully and early, before it snowballs into larger problems that have long-term effects on morale.

When people are willing to go into details, the facts often provide enough reason to not launch into unnecessary turmoil. Data analytics can even provide surprising results for companies. For example, one company assumed that people would purchase raincoats or flashlights during hurricanes. However, the data showed them that people were purchasing strawberry pop-tarts!