Algorithms Tell Consumers When to Buy Tech Products

It's a classic question when a lovely new gadget comes out on the market: when is the right time to buy? Buy the product early, and you get bragging rights and more time enjoying its features, but you almost certainly pay more than if you'd waited.

Case in point: HP's departure from the tablet business, with its announcement last week that it would stop making its Touchpad. Touchpad owners who had paid the full price of $499 for a device unlikely ever to see new apps, peripherals, or support from its parent company were insulted further when HP discounted the Touchpad to $149 for the 32-gigabyte Wi-Fi edition and $99 for the 16-gigabyte model. In one weekend, an estimated 350,000 people bought Touchpads, at more than $300 off the prices they would have paid just a couple of months before.

"People are constantly feeling burned by all these things," says Oren Etzioni, chief technology officer and cofounder of Decide, a startup dedicated to helping consumers decide whether to buy products or wait. Etzioni says it's possible to automate intelligent predictions about product pricing based on several factors. The company arrives at these conclusions by monitoring price trends, news, rumors, and technical specifications. Decide has not yet launched predictions for the tablet category, but it currently makes recommendations for other consumer products, such as digital cameras.

Prior to starting Decide, Etzioni founded Farecast—acquired by Microsoft in 2008—which predicted the optimal time to buy airline tickets.

But predicting prices for consumer electronics is more complicated. Airline tickets get used once; consumer electronics are meant to be kept and enjoyed. People don't want to wait so long to buy a product that they're stuck with an outdated model, no matter how cheap it is. "We believe there's only a limited time people will wait," says Etzioni.

To formulate its insights, Decide factors in predictions of new models or discontinued service. It also might direct consumers to newer models instead of discounted old ones. In some cases, Etzioni says, new models actually cost less than predecessors due to high supply.

Microsoft, which researches price prediction for its Bing search engine, remains interested in the topic. Yesterday, at the 17th Association for Computing Machinery conference on Knowledge Discovery and Data Mining in San Diego, a group of Microsoft researchers presented work focused on calculating the value a product could offer a customer in relation to its price. "It is crucial to consider loss of use due to waiting," said researcher Samuel Ieong.

Ieong and colleagues studied price and sales data from the market research company NPD from January 2005 to September 2008 to predict the best times to buy products such as camcorders, digital cameras, printers, and television sets. They found that while more utilitarian products such as printers had fairly stable prices, flashier products such as digital cameras could vary wildly, leaving consumers playing tricky guessing games.

Etzioni says the long-term impact of price prediction could be huge. It's not just a question of when to buy a flashy new toy, he says. As companies become better at predicting prices and features for all types of devices, buying at the right time could help consumers own better-quality products across the board. For example, he says, buyers might be able to predict when to purchase a refrigerator, taking into account both price and energy efficiency over time. Small gains in efficiency could become significant for a household product that people often keep for decades, he says. (Technologyreview)

Posted in: Computing