Skip to Content

Google Wants Search to Know What You Know

A revamped search engine might return different results depending on whether you’ve researched a subject before.
August 26, 2011

Google seems to have developed almost an almost preternatural ability to divine what users are really looking for when they enter a search term. Its engine often returns useful results for even the most egregiously misspelled queries. But Google’s user experience team hopes to give search an additional layer of intelligence—the power to grow with users over time, returning different results depending on whether users are just beginning to investigate a subject or have become old hands.

At Google, any change to search—which is probably one of the most financially lucrative products of all time—is done with caution. “As a team, we say all the time, ‘Don’t break search,’ ” says Jon Wiley, lead designer for the user experience team.

Google’s user experience team has two branches: designers and researchers. The design group builds the visual experience, and the researchers test it to see if it works, says Dan Russell, lead researcher for the team. Designers come up with possible changes, which the researchers test by inserting them into live search results for selected groups of everyday users and gathering data about their responses.

For example, Wiley is a new father, and over time, he’s performed a progressive set of searches related to parenthood. He envisions a system that could eventually respond intelligently to that progression, understanding that the questions he has about diapers right after the baby comes home from the hospital are probably different from those he will have in six months.

Wiley also imagines the search engine tracking and supporting research into chronic medical conditions, travel plans, or even years of study about a favored hobby, adapting as the user goes.

Of course, that’s a moon-shot vision. The team will gather information and make small changes toward that goal over a long period of time. Both designers and researchers do extensive field studies and interviews with users, standing over users’ shoulders to observe how they approach search tasks. In the past year, Google ran more than 20,000 experiments and made more than 500 adjustments to search.

Wiley says that adjusting users’ approach to long-term search tasks is a particularly delicate process. “How do you give help at the right moment and in the right way?” he asks. There are dangers to distracting or confusing users, or to pulling them into an experience that feels unfamiliar.

One key may be simply providing people the right guidance on refining their searches. He notes that there are many repetitive patterns in people’s searches that might give Google’s system clues about what stage people have reached in a search project.

However, the team isn’t sure how much of this support should come automatically. “It’s better to underpromise and overdeliver,” says Wiley.

Russell points to tools that can already help savvy users refine a task. For example, he says, users who allow Google to save their search histories can mine that information to review material they’ve seen in the past. Google also offers search tools such as views of related searches, which Russell says can help break blocks for those who are stuck. He runs a site called A Google A Day that uses puzzles to teach people more sophisticated approaches to search.

One way to improve the search experience might be to change the way the results page looks in a way that would help users discover more possible approaches, says Wiley.

Observing users also gives Google’s engineers and designers clues for tricks that could help others. For example, many people go to YouTube to find medical information, Russell says. While that may seem counterintuitive, the team discovered that people wanted to see examples of how conditions progress, how surgical procedures work, or how to execute home treatments and physical therapy exercises properly.

The team sometimes makes slight alterations to search results to emphasize approaches that may not occur to everyone. For example, they recently adjusted results for music searches to make it easier for users to find sites where they can listen to a requested song.

Russell stresses that no matter how intelligent the interface, it will never substitute for a skilled user. “There’s a deep structure to these things, and you have to know what resources are possible,” he says. “The technology is constantly changing, and the user interface is constantly changing. One of the big takeaways is you’ve got to pay attention.” 

Deep Dive


Capitalizing on machine learning with collaborative, structured enterprise tooling teams

Machine learning advances require an evolution of processes, tooling, and operations.

The race to destroy PFAS, the forever chemicals 

Scientists are showing these damaging compounds can be beat.

How scientists are being squeezed to take sides in the conflict between Israel and Palestine

Tensions over the war are flaring on social media—with real-life ramifications.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at with a list of newsletters you’d like to receive.