Cooking Trends in the Pandemic
It’s a common refrain that people have formed a new bond with their kitchens during the pandemic. In the early days, as restaurants closed and many of us lost jobs or started working from home, my Twitter feed was abuzz with interest in the kinds of slow kitchen projects that keep you ducking in and out of the kitchen all day. Suddenly there was abundant time at home to simmer a slow-cooking pot of beans or bake sourdough. I think many of us also needed these sorts of projects, to make our homes feel more homey and soothe our worries about a world descending into crisis.
But now, 10+ months into the pandemic, I get the sense that Americans are no longer so enthusiastic about beans and sourdough. Perhaps we’re sick of our own cooking, perhaps we want to support local restaurants, perhaps beans just aren’t as tasty as people thought they would be when they stocked up on them in March. I started to wonder about this ebb and flow of interest in cooking – has the novelty actually worn off, or have people just stopped posting about it on social media? If it has worn off, when did that start to happen? And how does interest in restaurants compare to our interest in cooking?
To dig into this, I ran a few quick analyses using the Google Trends API in R. Google Trends tells you how popular certain search terms were over a particular time period. This should be a reasonable index into how interests change over time – though of course, it’s not perfect (e.g., some people might go straight to their favorite cooking blog for recipes rather than going through Google).
To start to get a sense of how interest in cooking evolved over the course of the pandemic, I retrieved the popularity of a few cooking-related search terms: ‘recipe’, ‘beans’, and ‘bread’. I zeroed in on the period from January 1st - December 31st, 2020. This way, I’d have data from a couple of pre-pandemic months for comparison.
This code retrieves the percent of all searches that included these search terms for every week in 2020.
if (!require(gtrendsR)) {install.packages("gtrendsR"); require(gtrendsR)}## Loading required package: gtrendsRif (!require(ggplot2)) {install.packages("ggplot2"); require(ggplot2)}## Loading required package: ggplot2# setup search parameters
cookingTerms <- c("recipe", "beans", "bread")
timePeriod <- "2020-01-01 2020-12-31"
# get google trends data
cookingTrends = gtrends(cookingTerms, gprop = "web", time = timePeriod)[[1]] # pull out the dataframe
# quickly check out the data:
# head(cookingTrends)
# make sure the 'hits' are all numeric (otherwise the plotting order will be nonsensical):
cookingTrends$hits <- as.numeric(cookingTrends$hits)
As you can see, it seems like there’s a dramatic uptick in searches for “recipe” starting around March, 2020. This surge in recipe searches peaks in April and then slowly tapers off through the summer and fall. It ticks up again around Thanksgiving and the winter holidays, but this holiday up-tick is consistent across the years, and not specific to the pandemic period. “Bread” searches also have a bump at the beginning of the patterns, but interest in bread-baking seems to have pretty much gone back to pre-pandemic levels. Or, who knows, maybe everyone learned how to bake bread in March and April, and now they’ve continued baking without having to Google for recipes! As I mentioned, this metric is imperfect; but, I think that it does at least reflect trends in the information people need at a given time. Beans, on the other hand, really don’t show much of a change in searches over the course of the year. Perhaps bread was just much more instagramable, and therefore generated more interest – as Peggy Olson discovered on Mad Men, it’s very hard to make beans look attractive.
So what happened once people stopped Googling for recipes? Did they start Googling for restaurants instead? To get a sense for this, I added a couple of search terms to my analysis: ‘restaurant’ and ‘takeout’.
# setup search parameters
restaurantTerms <- c("restaurant", "takeout")
# get google trends data for both cooking and restaurant terms
allTrends = gtrends(c(cookingTerms, restaurantTerms), gprop = "web", time = timePeriod)[[1]]
# make sure the 'hits' are all numeric (otherwise the plotting order will be nonsensical):
allTrends$hits <- lapply(allTrends$hits, gsub, pattern = '<1', replacement = '0', fixed = TRUE) # replace any hit value < 1 with 0
allTrends$hits <- as.numeric(allTrends$hits) # convert to numeric
# add in a column to differentiate between cooking- and restaurant-related terms for plotting:
allTrends$keywordType <- ifelse(allTrends$keyword %in% cookingTerms, 'cooking', 'restaurant')
# head(allTrends)
This plot suggests that searches for “restaurant” plummeted at the start of the pandemic, as many restaurants shuttered while their owners figured out how to pivot in a locked-down world. Searches picked back up around May, as restrictions started to roll back, some cities approved outdoor seating options, and many sit-down restaurants re-opened under a take-out model. It seems like searches for restaurants recovered almost to pre-pandemic levels by mid-summer (roughly 80% as high as searches in the summer of 2019), but then dropped again as cases surged in the fall. In contrast, searches for ‘takeout’ didn’t really change at all during this time period, suggesting that either people are less interested in takeout than I expected (this might happen if people mainly search for restaurants in order to eat there in person), or that they just don’t use the term “takeout” when Googling for a restaurant, even if their intention is to get takeout.
So, to summarize, there was a sudden and unprecedented interest in searching for recipes to cook at home at the beginning of the pandemic. This interest tapered off starting in the summer, and has since returned to pre-pandemic levels. At the same time that interest in cooking slowed down, interest in restaurants started to rise. But now we seem to no longer be interested in either – we’re not really searching for recipes or restaurants anymore. Maybe we’ve replaced an interest in food with an interest in other social-distance-friendly things, like movies or politics. Maybe we’ll see another spike in cooking interest, as the winter wears on and we re-focus on cultivating a sense of coziness in our homes. Regardless of our search habits, I hope that we’re still deriving joy from food – it can be a powerful balm against the chaos of the world, and we certainly need such balms right now.
