If someone asked you what professional problem you’re addressing at the current time, you’d be able to offer a short overview of the project and why you’ve taken it on. That explanation, and the reasons behind it, are sure to be sound and strong. However, it’s more than likely that it’s not the largest problem at hand for your company. Ask yourself instead a different question: Assuming anything was possible technologically, what problems would you be working on today? Chances are, that’s what you should be worrying about.
Machine-learning technology turns that exploratory intellectual exercise into an actionable reality. After hearing from some Shutterstock customers that it took them longer than they’d like to locate the images they needed, some of our engineers formed a computer-vision team a year ago to address the pitfalls that come with typing words into a search bar. During that time, we developed technology that came to understand the 80 million images inside our collection. We launched it publicly in March.
Machine learning and data science aren’t topics reserved for engineers and technologists, though. To ensure you’re spending your time and resources wisely, you must keep the customer in mind throughout the process. Otherwise, you’ll risk winding up with a fun novelty that doesn’t appeal to anyone other than those who built it. Here are some lessons we discovered from our deep learning:
1. Technological innovation cannot come at the expense of the customer.
Every technological challenge for a company to undertake must begin with a hypothesis related to and relevant to customer research. Sometimes, companies sacrifice the customer’s best interests in the name of technology, hoping that the technology will be so enticing that it pleases the user right away. Odds are you’ll fall short on this hope. You’re better off spending the early stages of your innovation getting to know your customer and to really understand their pain points.
For us, that showed itself in keywords and locating the images they really wanted. So our team dug in from that spot. We wondered what a search experience might look like without a search bar. That’s not only a fascinating concept for us, but we knew it was helpful and useful to the end user.
2. We know that technology can’t solve everything.
Emotion is hard for a computer to fully grasp. People are sometimes just looking for inspiration from our collection, and relying on our algorithm to help them locate what they want. The computer in cases like these can’t do much for them. A traditional search bar is the beginning of their journey, serving as a necessary component. Today, we give people the option to search either with some words or with an image they already have at their disposal. We know that technology works best when it’s geared toward people, not when it replaces them.
3. Computer-vision technology comes to life when people use it.
Concepts like machine-learning might seem new, but it’s actually existed and been debated for decades in an academic setting. What’s changed, though, is its application to businesses over recent years. To understand its power and use, we now know you need to put it in front of users. Since we released our machine-learning technology just a few weeks ago, we’ve already made enhancements based on customer feedback. There’s no such thing as a perfect algorithm.
Now, the conversation continues between customers and marketers, and in turn technologists. They need to hear what’s working and where the computer falls short — some searches can be more challenging for the computer than others. We’re working to fix what we can to better serve the search behavior of our customers. Ultimately, it all comes back to understanding the customer, and responding to their needs. That starts with us and extends to the technology we produce.
Contributed by: Kevin Lester, VP – Engineering, Search and Discovery, Shutterstock. Shutterstock is a leading, global technology company providing high-quality licensed imagery and music to businesses, marketing agencies and media organizations. Shutterstock has created the largest and most vibrant two-sided marketplace for creative professionals to license content – including images, videos and music, as well as innovative tools that power the creative process.
Sign up for the free insideBIGDATA newsletter.