“I believe that when the robots rise up, ATMs will lead the charge,” quipped The Big Bang Theory’s Sheldon Cooper. You don’t have to be a genius physicist (arguably on the autism spectrum) to have a theory about “machine learning”, but the reality is it’s not as scary as screenwriters, novelists, and Hollywood makes it out to be.
In fact, it might just be revolutionizing businesses that have accepted subpar figures—like the travel industry, for example. Did you know just one percent of visitors to a travel site will actually make a booking? In any other industry or other type of website, that would be a red flag that something either needs to be fixed or the business is sinking fast.
However, when economists met machine learning, something magical happened. Say what you will about economists (They’re right about as often as the local weather man, their academic outlook is rampant with tunnel vision, or they’re clueless about the reality of business? They’ve heard all that before). Just like everything else in the digital era, economics is undergoing a much-needed growth spurt. It’s enjoying a revolution. Their new mission? Fix their client’s troubles (like those pesky travel sites) with real solutions, and it’s all starting in Silicon Valley.
hiQ(uality) on the Horizon
Taking the travel site example, it’s easy to see some patterns that show how likely that one percent is (no, not that one percent) to actually book. For example, the faster a person can click, the happier they’ll be—but that doesn’t equate to more sales. With travel, the longer a person stays on the page, the more likely they are to make a purchase. Unless they just left the window open and went to make a sandwich, of course. How many times a person has visited before, where they’re located, what time of day they visit, and other big data gold can also help figure out how likely they are to buy.
By creating an algorithm, you can see how likely someone is to become a buyer. Then, algorithms can figure out gains and risks of revenues and how they relate to advertisements. In fact, it takes less than a second to introduce these algorithms that lead to machine learning and ultimately, a stronger website.
It’s all about pinpointing patterns floating around in an ocean of data, which is challenging according to astrophysicist cum data scientist Genevieve Graves. Her background is in studying galaxy data at Princeton, but today she’s the Chief Data Scientist for hiQ Labs, the latest promising startup in San Francisco. Her approach? Machine learning.
Rise for the Machines
The concept of machine learning was actually coined in 1959 by Arthur Samuel—which basically means creating computers that can teach themselves. The programmer commands a computer to keep repeating the same task over and over, and successes are measured. Kind of like with A/B testing, a slight change is made to the command, and the computer must note whether the change led to a better or worse result. That’s how it “learns”. Machine learning has been used in helicopter flights, spam filtering, and even for recording voices. Why not to optimize a website?
For hiQ Labs, machine learning is a perfect match to change business for the better. For example, what if a firm knew ahead of time which of their employees was most likely to give notice? Graves used data from one company (including job titles, pay scales, etc.) and compared it to other companies in the same industry. It led to pretty solid data on who would quit while giving insights that humans just couldn’t pick up. With this information, thanks to machine learning, managers have advance notice to either speed up the process or try to save the defector.
Challenges with the New Economics
According to Stanford economics PhD Scott Nicholson, figuring out those buried types like buyers, content, etc. is one of the biggest hurdles in the new landscape of economics. He’s an advisor for hiQ, and says newer firms are learning grounds—and also a meat market for sellers and buyers. He’s worked with LinkedIn and large-scale health companies figuring out a “customer’s type” so they’ll be best matched with their links. Just like some people’s “types” are tanned blondes while others prefer lanky, tall dark and handsome, matching customers with their preferred links (and other matchmaking skills) is crucial for a website.
“Linking”, beyond actual links on a website, is the foundation for business success. For example, a retail chain could link all their tills together via computer to see who’s performing best. However, a brand new platform is a tough sell—and the Airbnb head of economics Riley Newman agrees. As you probably know, this company connects “hosts” with “guests” around the country. However, it was tough to match up supply and demand. Facebook ads geo-targeted helped up supply where demand was too big. Google Ads, on the other hand, increased demand. However, with a little data crunching the company figured out why some hosts, regardless of location, were doing so well: They had better photos. Today, Airbnb offers complimentary photography services.
All in all, this is economics at work. How is it working for your business?
from Darlene Milligan http://ift.tt/1NZowHA via London digital marketing company
from Tumblr http://ift.tt/1ydbMv4
