Bak Kut Teh for Breakfast, Lunch or Dinner?
When I arrived in Singapore from Malaysia, I noticed people queuing to eat Bak Kut Teh at lunch or dinner, but hardly ever for breakfast. I was surprised. You see, I’d always associated Bak Kut Teh with mornings. Whenever I was home, I’d wake up especially early for it. At 8am, hot bowls of pork and soup would be steaming on the hawker stall stoves. By 2pm the best cuts would be sold out. Come 2:30pm and the roadside stalls would lie quiet and empty, chairs stacked on tables.
Seeing the lines outside Song Fa at noon left me mystified. My friends laughed at me. “Of course you’d have BKT at lunch!,” they said. I was skeptical of these anecdotes. I wanted to confirm what I was seeing with real data. So I did some data analysis.
Google Trends
Google Trends tracks a search terms popularity over time. As a metric, it’s a useful proxy for measuring what food people are interested in at particular times of the day (think about how people Google for restaurant opening times when trying to decide where to go for lunch or dinner). I decided to use Google Trends to measure whether people’s interest in BKT would peak at breakfast, lunch or dinner.
Compared to Trends, a more truthful measure would be to physically count the people making up those queues outside Song Fa in Chinatown. But as a research method, counting queues was impossible. Google Trends can be accessed from a laptop in an air-conditioned room. Counting people in a line would mean standing in the heat with a counter. Give me air-conditioning over the sun any day. Using Trends also had other benefits: it covers the whole of Singapore. Hence, it tracks of how an entire city, rather than pockets of it, feels about an idea. Considering all of this, I decided on Google Trends.
According to Trends, a score of 100 is the highest popularity a term can achieve. A score of 50 means that the term is half as popular. A score of 0 means there was not enough data. The scores represent “search interest relative to the highest point on the chart for the given region and time”
Hypothesis
If BKT is a lunchtime and dinner food, search interest should peak at 12 pm and 7pm daily for a one week period
Results
There is a nice daily seasonality to this graph. Interest builds and wanes with each day, which probably reflects the people’s daily rhythms. Interest rises when people wake up to eat and falls when they are asleep. Looking at specific days, there are usually two peaks everyday. The first occurs at 12pm (lunch, interest score: 100) and 7pm (dinner, interest score: 63).
There is a small peak at around 8am breakfast time with an interest score of 42, but interest is not as high as that during lunch (63) or dinner (100).
Looks like Bak Kut Teh is mainly eaten at dinner! I decided to act on my findings. The next time my friends visited from Kuala Lumpur, I brought them for Bak Kut Teh after work, not before. We sat down to pork ribs and trotters and watched the sunset.
You can explore more here by mousing over the line graph or by shortening the timeline to one day or extending it to 30 days. You can also add in other search terms for comparison. Macdonald’s is particularly interesting, sometimes peaking at 4am or 2am. I guess that’s when people get midnight munchies.
Data analysis is relevant not only to business questions, but to daily life as well. With a question, a metric and some data, life experiences can be analysed in a deeper way, a way that moves beyond anecdotes and speculations. Although they are not perfect truthful, when used corrently, they can be tools that drive action. For me, I’m going to continue pairing evenings with Bak Kut Teh dinners.
