Over the past year Apple has hinted at plans to extend Apple Pay well beyond mobile apps, to an increasing number of mobile sites.

Their pitch: By making it much faster and easier to purchase products, simply tapping a finger versus entering credit card details on a small device, we will help you convert more mobile shoppers into buyers.

Mobile shopping rates have been skyrocketing, but to date, “shopping” has mostly meant browsing and researching products online. While the gap has been narrowing, mobile devices have continued to lag behind desktops when it comes to actually driving conversions. This is because desktops have traditionally delivered a purchase process that is smoother, easier and faster than mobile.

Bloomberg News

Apple Pay coming to mobile sites represents the latest major advance in making mobile purchases more “frictionless,” thus helping mobile sites meet or even exceed desktop conversion rates in the not-too-distant future. However, shoppers will not buy products they can’t find, no matter how quick and seamless the checkout process may be. Mobile sites that aren’t optimized to deliver highly relevant search just might miss one of the most promising opportunities in years. With this in mind, what are the key site search capabilities mobile sites need to consider?

Search that is intelligent: Increasingly demanding mobile shoppers expect to be able to find products and make purchases in the time it takes to hail a cab, or wait for a “don’t walk” sign to change. This requires search capabilities to be highly intuitive, capable of compensating for human error and language variances while delivering highly relevant results with razorlike precision.

At the most basic level, search must be capable of identifying “fat finger” spelling mistakes that often occur on mobile devices — so a search for “blue sweters” will still return “blue sweaters.” Search capabilities must also be able to identify synonyms, such as “jacket” and “coat,” or “sleepwear” and “nightgown.” Advances in natural language processing (NLP) are enabling these capabilities, ensuring that sites don’t miss sales due to human tendencies.

Search must also be able to identify the primary item being searched for, and the most important descriptors for individual items. For example, if a shopper is searching for “size medium yellow rain boots” the search engine must be able to identify what is the primary item being sought — boots. A result for “size medium yellow rain coat” may contain four of the five search words, but it is not at all relevant and should be deprioritized in the results.

From there, search results must be weighted base on the importance of various descriptors. In the case of boots, size might be the most important descriptor, followed by style and then color. So, “size medium blue rain boots” will be ranked higher than “size large yellow rain boots” or “size medium yellow fashion boots.” This kind of capability leverages machine learning to go far beyond basic keyword-based matching, and can be especially helpful in the mobile realm where too many results often inundate and alienate shoppers.

Search that is personalized: In addition to making it easy for shoppers to find what they are looking for, search that factors personalization into results, or, displays results in accordance with an individual shopper’s past buying trends, can also help drive more sales.

Consider a shopper looking for a casual dress. Personalized search can analyze the individual shopper’s past behaviors (including searches and purchases), ideally across both in-store and online channels, to determine which attributes are most likely to resonate with this shopper — for example, color or style. If the search engine can deliver results that appeal most effectively to the individual customers’ tastes (for example, blue sundresses), the site increases its chances of actually closing the sale tremendously.

Search that is profit-aware: Once a shopper finds what he or she is looking for, mobile sites need to display results in a strategic manner that allows them to “play their strongest hand” in driving their own profitability. For example, if a shopper searches for a “red Nike T-shirt size large” and there are several results that are highly relevant, they should be displayed according to which items are the strongest performers. Which product drives the most sales? Which product drives the most click-throughs? These profit-focused factors should influence the ultimate results display.

We are gradually moving in the direction of a “mobile first” world, with the most recent holiday shopping season providing several telling examples. According to Adobe, Thanksgiving Day 2016 had its best-ever showing in mobile sales. Furthermore, while desktop purchases represented the majority (64 percent of revenue) of U.S. online sales on Black Friday, this was the first day ever to generate more than a billion dollars in online sales from mobile devices. The $1.2 billion spent via mobile devices represented an increase of 33 percent from last year.

As mobile forges ahead as a conversion-driving platform, mobile sites must continue to make the purchase process as convenient and slick as possible. Industry leaders like Apple are addressing this through bold moves, and others like Google are following (just this week, Google announced that Android Pay now works on any mobile site accepting PayPal). In order to capitalize on these shifts, mobile sites in their entirety must be ready, including optimizing those features most vital to the conversion process, like search.