Two weeks ago we put out a prediction regarding US online spending for Valentine’s Day: “You CAN Put a Price on Love” to the tune of $500 million.

Marketers interested in the power and usefulness of predictive analytics will be interested in seeing the actual results vs. our prediction (below). The model ended up being very accurate. Online spending was within about 7% of the day-to-day prediction.


Formulating the model that drove this prediction revealed some interesting behavioral patterns around Valentine’s shopping.

  • Valentine’s shoppers wait until the last minute even for online shopping.
    • Marketers should expect this and not sell off inventory too soon.
  • Shopping patterns are highly dependent on the day of week that Valentine’s falls
    • 2013 lagged behind 2012 in total shopping the week prior to Valentine’s, but made up for it the week of Valentine’s because Valentine’s fell on a Thursday rather than a Tuesday.
  • Shopping starts 3 weeks before Valentine’s and that early shopping sets the tone for YoY growth.
    • Marketers don’t need to wait until the week of Valentine’s to see if their promotions are working.

Does this jive with what you saw last week? Let me know @tyrwhite.