Knowing when the best time to buy something online could save you a lot of money.
It used to be, as the saying goes, that timing was more an art than a science. But sveral new websites are using big data analytics aims to change this by telling you when the best point in time to buy everything from a sofa to a plane ticket.
Nifti, which was launched this month, aims to track the prices of consumer goods like clothing and home goods, alerting users who install the bookmark on their browser when the price of the item they’re coveting drops below a certain price.
But more than alerting users to price falls, it aims to gather pricing data over time to better understand price fluctuations.
As e-commerce platforms have become more sophisticated, and merchants are more experimental with their pricing tactics, prices are changing more than you would expect.
“Maybe two years ago it wouldn’t have been as big of a problem. But now – Amazon, Walmart, Best Buy – they’re some of the retailers changing prices the most. Intra-day it will change several times on the most popular items,” says Shauna Casey, vice-president of research and marketing at Decide.com.
Like Nifti, it focuses on helping consumers to better time purchases on more than three million items using 100 different factors.
It’s only going to get more complicated, as more purchasing takes place on the web. Online commerce is expected to top £800 billion globally in 2014, according to research firm eMarketer.
The good news is this explosion in online retail and dynamic pricing has correlated with a rise of more and better data. This in turn has made it possible for new technologies to employ new learning techniques to pinpoint the perfect time to purchase.
“The machine learning technologies we use have been around a long time,” says Giorgos Zacharia, Kayak.com’s chief scientist, who helped build the travel website’s fare price predictor, which looks at airline and hotel prices.
Airline ticket prices were one of the first arenas that data scientists targeted with these improved technologies- partly because of the easy availability of pricing data.
“Airline ticket prices are one of the most perfect examples of chaos theory in the world,” says Stefan Weitz, Microsoft’s senior director of search. “Some small variable somewhere has kicked off a chain of events that’s kicked off price variation or price variability.”
When to buy airline tickets in the US:
- The lowest average domestic airfares are found 21 days out
- International airfares are lowest 34 days out
- September is the cheapest month for domestic travel; February and March are cheapest for international flights
- Travel to Toronto now – it was the only major city where fares dropped in 2012
- If you’re travelling internationally, leave on a Tuesday and return on a Wednesday – prices are 21% lower than average
Using variables such as historical data, capacity, and what’s happening in different areas, Bing Travel – formerly known as Farecast.com, which was acquired by Microsoft in 2008 – claims to tell you with 86% accuracy whether you should buy a plane ticket now or wait, because the price will drop in the near future.
When it gets a prediction wrong, crucially, it aims to show users why the wrong prediction was made – displaying colourful pricing graphs and data points – to maintain trust.
User feedback can be plugged into the algorithm to improve further future performance.
Beyond the limitations of the algorithms powering price predictions, the reality is that there are certain areas where knowing when to buy is of little use- like concert tickets.
SeatGeek aggregates and tracks prices of tickets to events like sports games and concerts.
However, “the data would indicate that the best time was always to wait a little bit longer because due to the perishability of tickets, the ticket has a finite expiration date,” says Will Flaherty, SeatGeek’s director of communications.
“What you tend to see is the very best deals in the marketplace are going to show up at the last minute” – which isn’t helpful for anyone who wants to make plans more than an hour in advance of the event start time.
The irony of the big data knowledge circle is that the research by buyers is based on sellers’ knowledge of the buyers’ habits. So the more that each other knows about each other the more that purchasing relationship is going to breakdown as each attempts to out bluff each other.