For a holiday that was celebrated yesterday by almost everyone universally in some form or other, once you get past certain parts, some surprisingly important details are a little sketchy for most of us.
For example, when do you celebrate Thanksgiving in the United States?
Did you say the last Thursday in November?
It is OK, most people think that.
But Thanksgiving is actually celebrated on the fourth Thursday of November, which most years is identical to the last Thursday, but if November 1st or 2nd is a Thursday, the month has five and Thanksgiving is early.
If you got that one wrong, don’t feel bad. Until 1939, you would have been right. That was the year FDR — at the urging of retailers who dreaded a shortened holiday shopping season — decided that in the grips of a depression, everyone could use a little more Christmas, and made the change to the fourth Thursday via Presidential proclamation. In 1941, a joint session of Congress codified it into law.
The rest, as they say, is history.
And the above, other than adding the pop quiz you were looking for to jumpstart your Black Friday shopping excursions today, is also illustrative of the funny thing about Thanksgiving. Once you get past the stuff that everyone knows — turkey, stuffing, the Macy’s parade, football, family, fellowship, shopping, pilgrims — it quickly turns into the stuff almost no one knows. Even when it’s the easy stuff, like when the holiday is each year.
Then you get into the hard stuff like predicting what it is consumers are going to do when the official national shopping celebration actually gets off the ground — and things can get really tricky. In some years — 1939, for example, or 2009 for a more recent example — one can guess what the news is going to be (in both of those cases: bad.) In years like 2015, when consumers are spending — though how they’re doing it and where that spend is going seems to be shifting — it gets exponentially harder.
As we’ve covered for the last few weeks, the 2015 environment hasn’t lacked for indicators about the shape of retail to come, but those indicators seem stubbornly devoted to pointing in different directions.
But sometimes data only appears to be too weird, complex and contradictory for interpretation — when in fact all it needs is a computer to read it instead. And holiday data is even getting some special attention in the form of a look from a celebrity computer.
While most people have heard of IBM’s Watson computer, the context most are familiar with Watson is in “Jeopardy.” Watson was, in fact, built to win “Jeopardy,” and he did so quite spectacularly.
But more remarkable than that is how Watson won at “Jeopardy” — or, specifically, how he was programmed to win, which requires the program to do two things.
Lots of algorithms can answer questions, but programming a computer that can correctly guess intention from an “improperly” worded question well be enough to answer it almost instantly is not easy.
And that kind of thinking machine is useful in more applications than simply humiliating humans at “Jeopardy.” And Watson’s next project, Watson Trend, is using a mix of machine learning, sentiment analysis, keyword analytics and natural language analysis to get a better grip on just what consumers will be doing today on Black Friday and beyond.
Some of what Watson knows is not too surprising — and didn’t need a champion computer scientist to figure it out, since anyone with a smartphone and Twitter could have guessed that Star Wars, Apple Watches, LEGOs and smart TVs were going to be trending this week, and this season in general.
But Watson can add some granularity to what it sees.
With its focus on natural language, Watson can know more than if people are talking about something, but also if they are talking about it for the “right” reasons. Lots of consumers might be talking about the Apple Watch or Star Wars, but the tone of the conversation matters a lot especially from a retail prediction point of view.
Watson is also able to stack up the chatter against the behaviors of actual shoppers in the field — and see what patterns are actually showing up — and in a way much more efficient and focused than a casual (or biological) observer might.
Watson isn’t a tool for predicting the future, so much as it is an innovation bent on giving better, sharper context to the present, and better insights into what consumers are actually doing, aspiring toward, mocking and embracing in a way more powerful than counting heads and drawing conclusions.
It is a system still in progress and, its developers note, still making observations that are expected. But as the season wears on, as Watson’s ability to pick out consumer intent in a field full of noise gets a real-time real-world test, those same developers think it might start seeing how we shop in a very different light.
We’re always on the lookout for opportunities to partner with innovators and disruptors.
Learn More