Search by Algolia

Sorry, there is no results for this query

Autocomplete (predictive search): A key to online conversion
facebookfacebooklinkedinlinkedintwittertwittermailmail

Have you ever had a friend who would finish your sentences? It’s amazing when someone knows you that well, isn’t it?

The equivalent of this interactive best buddy in the tech world is autocomplete, also known as predictive search, type-ahead, autosuggest, auto-fill, and autofill. It’s online search functionality that seems to be thinking about what you want as you try to enter it in a search box or text field. It’s AI predicting which keystroke, word, or phrase you’re going to enter next and then trying its best to finish your sentence.

Adventures in autocomplete

For example, let’s say you suddenly have a craving for a chocolate sundae and want to find an ice cream store. You start entering “ic” in the search bar on your mobile phone, and before you can even add the “e,” the mouth-watering idea of “ice cream near me” appears. 

Ah, AI knows you well. And in case sprinkles on Rocky Road is not what you want, it also guesses some back-up ideas: iCloud, Ice Cube, and Iceland. You can then quit the annoying process of entering characters, click the correct suggestion, and get to your relevant results.

Humble beginnings

The concept of autocomplete predictions was first talked about in 1967, according to Wikipedia. In its infancy, autocomplete’s first application was to help people with physical disabilities type faster and decrease the number of keystrokes they had to hit while writing.

The appeal of suggested words and phrases ultimately proved to be much, much greater, for instance, among doctors, who were having to write prescriptions for hard-to-spell drugs, and so were thrilled to have long names served up for easy consideration. Plus, even non-doctors of all types could appreciate that kind of text input.

In 2004, Google introduced an experimental version of its autocomplete technology, the pioneering “Google Suggest,” and it was released in 2008. 

The suggested autocompletes that you could initially get from starting a Google search, which would factor in data about what other people had already been searching for, were endlessly wacky and entertaining, even inspiring some people to dub them “accidental poems.”

A complete success

Today, autocomplete is a familiar old friend that makes logical-sounding suggestions to people using search engines, web browsers, database query tools, email, and more. Website visitors expect it to just be there, as they’re used to digital hand-holding thanks to the high standards set by the likes of Google, Amazon, and Netflix. They expect suggestions and the right search results instantaneously; if the technology is lacking, they’ll simply go to a competitor’s site.

In fact, autocomplete makes search so much easier and faster that it’s earned status as a usability best practice. And that’s a good thing, because although people are accustomed to being able to use effective search, their own search skills are lousy: according to one study, only 1 percent know how to tweak their search tactics to find more-effective results. So without access to autocomplete, people are, well, incomplete.

Back-end details

How does autocomplete work its magic? It’s all about algorithms. Based on various factors, algorithms are applied to identify the possible completions a user will see when they start typing a query. Autocomplete algorithms could factor in:

  • What’s currently being talked about and how many people are searching for it
  • How much or often the searcher has entered a term
  • The click-through rates for particular search suggestions
  • “Semantic” considerations, which pinpoint what a searcher specifically means — the context; synonyms can be substituted to improve relevance
  • On an ecommerce site, which queries a prospective buyer previously entered and clicked, and which led to a purchase

Operational options

You may have noticed that autocomplete features you’ve used don’t all work in exactly the same way. Specific techniques include:

  • Instantly providing suggestions without delay when typing starts, which not only provides a great user experience but seems more conversational.
  • Highlighting similarities among — or differences between — search suggestions. This makes it easy to quickly scan for the right item (an especially big help on small mobile screens). For instance, here’s how online retailer Under Armour highlights the similarities — “top” is in bold:

  • Limiting suggestions. It’s critical to not overwhelm people with too many choices, so you can configure autocomplete to present only the best suggestions, as determined by search analytics.
  • Optimizing suggestions, that is, presenting those that show the most relevant results. This expands the chances of success on a searcher’s first try by reducing opportunities for typos and awkward phrasing that could derail the search. Autocomplete suggestions are typically based on business relevance, data from site search analytics, and business goals. An optimized autocomplete search engine is typo tolerant and uses the same relevance factors as your search engine.
  • Teaming autocomplete with federated search functionality, which expands the breadth of selectable search results. For example, on a beauty website, a shopper could be provided with products, related brands, and editorial guides and tips on things like “How to choose a moisturizer.”

A is for autocomplete, B is for benefits

Autocomplete’s search-transforming power is borne out by statistics, these according to Google: 

  • Users complete forms up to 30% faster when using autocomplete
  • It’s especially handy on mobile devices, but efficient for desktop, too; on average, autocomplete reduces typing by about 25% (cumulatively, that amounts to more than 200 years, globally, of typing time per day)

Of course, improvement in search usability can lead to a subsequent boost in, you guessed it, conversion. “The less you ask of users, the more inclined they are to complete a form, and faster form filling increases conversion,” explains Shanmuga Priya Pandiyan of eBay. Autocomplete helps people find relevant results faster, reducing the bounce rate from confusion, distraction, or the inability to find an item because the search engine can’t parse the query. 

How much can autocomplete typically be a game changer? One estimate: it can improve sales and conversions 24%, says search analytics expert Mike Roberts (Spyfu). “Autocomplete makes search queries longer,” he notes, and “longer queries have a higher conversion rate.”

If conversion isn’t your main priority, autocomplete can still be a huge help. How? If your site has a shopping cart, autocomplete can increase the value of an average order. It also improves user engagement, especially for verticals outside of ecommerce. The less time that people must spend searching for information, the more time they have to engage with site content. 

It’s a no-brainer that websites in many industries can benefit from adding (and fine-tuning) autocomplete. Of course ecommerce sites can use it to help shoppers navigate their product catalogs; media companies can use it to help users find relevant news articles and videos. But surprisingly, despite evidence of autocomplete’s awesomeness, according to Baymard Research, as of 2020, 66% of ecommerce websites were not providing it.

A win for businesses

If you’re a company with a website, what exactly do you get by investing in an autocomplete feature? By showing people popular queries that will lead them to results they like, you can improve the chances of their sticking around and checking out more of your content. You can:

  • Educate people about your site. For instance, you can have autocomplete show product categories that expand what people know about your offerings. You can align autocomplete suggestions with your business priorities, and also have it accommodate multiple languages to circumvent confusion. Increasing visitors’ awareness can increase their satisfaction.
  • Decrease the amount of time people must spend searching. Since autocomplete provides fast contextual suggestions, it’s unlikely that someone will need to type in their entire query. They’ll be happy to avoid entering long queries and incurring typos, which an unoptimized search engine could have trouble parsing.
  • Support merchandising initiatives. You can use autocomplete suggestions to curate the content your users interact with. For instance, you can:

    • Show the most relevant items first based on users’ general search behavior (e.g., what’s most popular or trending)
    • Base suggestions on business priorities (e.g., promote items with higher price points or margins)
    • Curate seasonal merchandising using autocomplete rules
    • Set up personalization for returning users
  • Improve the user experience. Getting people immediately to their desired content translates to excellent usability.

Mobile isn’t (always) moving

One particular area where businesses are failing is with mobile apps. 

Let’s say you’re out and about and want to see what time the train leaves. You key in “train schedule” on your iPhone or Android phone and pull up a squished-looking corporate website that makes you jump through hoops just to get to the schedules. It proves to be a giant hassle, and one that could have been avoided if the company had optimized for mobile.

This type of problem happens a lot. Conversion rates for ecommerce mobile sites are 66% lower compared with those of desktop sites (Smart Insights). Mobile apps don’t provide people with the correct auto-fill results, for instance. 

Baymard Institute backs this up. It did mobile-screen autocomplete usability testing (87 companies in all) and found that some predictions ended up being misleading. “Mobile autocomplete is particularly difficult to get right,” it concludes, “as the user interface suffers from an already small screen…combined with the user’s touch keyboard being shown at the same time (taking almost 50% of the screen).”

Automatic advantages

Despite the need for more autocomplete, who could argue that it isn’t the greatest thing ever? When it’s available on a mobile device or anywhere else, it:

  • Makes life easier
  • Provides proven queries for users to click
  • Increases searchers’ chances of successfully finding items
  • Decreases the number of times people get zero results from a search
  • Increases searcher confidence: you can see that you’re on the right track and then just add details
  • Helps support merchandising initiatives
  • Could expand searchers’ awareness of other products (cross-selling)
  • Through this expanded awareness, could reduce abandonment, gaining potential conversion opportunities

“You complete me,” said Tom Cruise to his admin/love interest in the movie Jerry Maguire. Whether this concept of completing is applied to relationships or search terms, it’s clearly a truly powerful thing, and in terms of search technology, one that’s only going to become more well loved.

For more information

Partnering with a search as a service partner can provide the UI, analytics, and other tools to create great search. If you’re interested in delving in to the possibilities of autocomplete, here are a couple of resources:

About the author
Catherine Dee

Search and Discovery writer

linkedin

Recommended Articles

Powered byAlgolia Algolia Recommend

Autocomplete: How search suggestions increase conversions
ux

Elliott Gluck

Senior Product Marketing Manager

What is predictive search? What is autocomplete?
ux

Xabier Ormazabal

VP of Product Marketing

Building a composable front-end search with autocomplete and instant search results
product

John Stewart

VP Corporate Marketing