Keywords and Inferred Intent: How to Think like Your Search Engine
Search engines use keywords and inferred intent to interpret a query. Understanding which clues your wording supplies can help you choose terms that bring back more relevant results.
THE IDEA
Keywords and Inferred Intent: How to Think like Your Search Engine
Browse the latest tech news and you might think artificial intelligence is on the verge of replacing humans. But sometimes artificial intelligence isn't all that intelligent. And until machines can think like humans, humans sometimes have to think like machines. Put differently, in order to get the machine (or search engine) to do what you want, you have to know a bit about how it turns your request into a result.
Let's start with this: machines don't think. They seem to act like humans much of the time, and sometimes they get the same result. But what happens under the hood is quite different.
Imagine, for example, you buy a robot to fetch milk from the center shelf of the fridge. Most days, Mr. Robot works just like a human. It goes to the fridge, opens it, grabs the milk, and pours it on your Lucky Charms. One day you're being lazy when you put away the groceries, and you throw the milk in the side door and put your sriracha on the center shelf. You've reversed the positions. Guess what happens? Mr. Robot goes to the fridge, grabs the sriracha, and pours it on your cereal, causing you to discover, definitively, that not everything is improved by sriracha.
But you've also discovered something else: the robot does not “see” the way you do. When you look for the milk, you're looking at a variety of factors to determine what's milk and what's not. For you, where it's placed in the fridge doesn't have much impact. For the robot, position apparently matters much more. That knowledge is valuable to you. Now that you know how the robot “sees,” you know that to avoid condiment-based disasters, you have to organize your refrigerator in a certain way. Sriracha in the side door. Milk on the center shelf.
Google can seem human-like. We ask it questions. It gives us things we might find useful to read. But like our sriracha-snatching robot, Google “sees” differently than we do. First, Google was built on a technology of keyword searching. Second, in more recent years, Google has devoted a lot of resources into something called “machine learning” to better determine user intent. Let's talk about how understanding what Google “sees” when it goes looking for things can make you a better searcher.
When to use it
When researching a topic, try its specific names and terminology, then revise the query after inspecting the results.
When a voice assistant misinterprets a request, rephrase it to make the intended action and object explicit.
When designing a website, optimize content with clear keywords and structure to align with how search engines index pages.
When troubleshooting a problem, imagine how a search engine would ‘see’ your query to refine your search terms.
FROM THEORY TO EVERYDAY LIFE
In practice
Robot and Sriracha
A robot programmed to fetch milk from the center shelf grabs sriracha after the groceries are rearranged. In this imagined case, the robot relies on position, a clue the human user did not realize mattered.
Search Query Optimization
If you search 'best ways to cook chicken' Google uses keywords and inferred intent to rank results; using specific phrases like 'oven-roasted chicken thighs recipe' yields better matches.
A MOMENT TO REFLECT
Put it to work.
How would you rephrase a failed query to give the search engine clearer clues about both the subject and your intent?
Recall a time you received irrelevant search results; how might the robot analogy explain why?
In what other areas of technology does thinking like the machine help you achieve better outcomes?
Often confused with
CONTINUE THE CONVERSATION
Source texts
- Verified How to Think Straight, Get Duped Less, and Make Better Decisions about What to Believe Online (Mike Caulfield, Sam Wineburg)