Rain, rain, go away…unless you are an online retailer.
The methodology seems sound:
I began my exploration by pulling sales data for four major retailers from differing verticals (Home/Furniture, Wholesale, Clothing, and Big Box), concentrating specifically on the greater Seattle area. For weather I used daily temperature and precipitation data from the Seattle Tacoma Airport, and long-term temperature normals from the National Climatic Data Center to identify unusually hot, cold, and rainy days. I isolated noise from the long term sales trend by subtracting daily sales data from a moving average, and cropped the dataset to only look at Saturdays and Sundays-—days where I’d expect people’s behavior to be the most influenced by weather since they’re not subject to the work week.
And the results are in-line with what we’ve hypothesized for a while, based on our clients’ aggregated PPC and e-commerce sales data:
The difference between a clear day and a cloudy day means a difference of 10-12% in orders for our Clothing, Home/Furniture, and Wholesale retailers, and, just as interesting, makes no significant difference for our Big Box retailer. In general, people of Seattle buy less online on sunny weekends, and in particular they buy less home/furniture, wholesale, and clothing goods.
If you are a more visual learner, here are their graphs showing the change in order volume on clear vs. cloudy days.
This research is interesting, but not immediately actionable. First of all, we can’t control the weather. Second, the author points out that the deviations are small enough to leave room for debate.
In our client testing, we have proven that changing our Google AdWords ads based on time of day can have a significant impact on Click-Through Rates and Conversion Rates for offline businesses. Should we also be testing different calls to action triggered by different weather conditions?
It’s an interesting question, and one I’m sure somebody somewhere is working on.