Behind the scenes in search

Voice search brings new challenges, highlights old issues

According to ComScore, which provides metrics on cross-platform consumer and audience behavior, 40% of adults use voice search once a day. ComScore predicts that within 2 years, 50% of all searches will be done by voice. Google says that 20% of searches are voice-related. The increased use of mobile devices is a major factor—typing on such devices is more difficult than on a keyboard so voice is more convenient.

“Voice search is very different from keyword queries, said Khrysti Nazzaro, VP, brand strategy, at MoreVisibility. “Natural speech patterns elongate query strings significantly. In addition, adjustments should be made in the keywords that are being leveraged.” MoreVisibility is an online advertising company that specializes in SEO, pay per click, and website design. The combination of length and natural language in voice queries affects the way advertisers and webmasters should structure information for optimal retrieval. “Content on the website should cover all the main questions,” advised Nazzaro, “so that Google can understand a question even if it is formulated in different ways.”

The structure of the data is key to successful retrieval, particularly for natural language inquires. “The facts are not different based on what service a consumer is using,” observed Erin Jaeger, director of product marketing at Yext, “but the search strategy may be. We work with a lot of customers, asking them to describe their voice search strategy. As consumer services like Google and Alexa evolve, the way the search services parse the natural ?language queries will also evolve, but there will always be a need for structured data.”

Given the context of voice queries, companies that want their websites to be found should develop concise pieces of content. Keywords should reflect the conversational tone of voice queries so the content will be found. And, industry experts advise, in order to be a match for such searches, content carried on a website needs to have a structured data markup that provides equivalent detail. 

Searching the site

If a prospective customer searching for a product or service arrives at a website, either by going directly to a particular site or via organic search, the goal is to get the individual to the information that will help them achieve their objective. Some websites operate entirely via links from one page to another or tabs at the top of the page. Others may offer faceted search, which presents a list of choices based on a taxonomy, and generally offers filters that include a product’s price, color, brand, model, or other descriptor based on the type of customer who would find that attribute of use in a buying decision. Companies typically offer onsite search with a search box that allows the user to enter a query; however, those queries alone sometimes lack meaningful context.

The key to maximizing a positive search outcome is context. “If a user is searching the site of an industrial product manufacturer, ‘mold stripping’ can mean many things,” said Seth Earley, CEO of EIS. “To a plastic manufacturer, it might mean breaking down an injection mold. For that, certain tools would be needed. A maintenance company might need to remove mold from a wall, and a carpenter might be stripping paint off wooden trim, so they might need chemicals. Clues such as prior purchases or information about the searcher’s profession could help the system retrieve the right information.”

In order for this to happen, though, information about the product needs to be structured, stored, and accessible to various enterprise systems. Ideally, onsite search should be integrated with enterprise systems that also help provide as complete a picture of the user as possible. Being able to access other information helps to disambiguate queries; for example, knowing what industry the searcher was from, could help narrow the options. This data provides additional “signals” to enrich the weak signal that keywords alone typically provide.

In some cases, emerging technologies have served as a catalyst to promote a new look at structuring content. “Chatbots are a retrieval mechanism that forces enterprises to think about content and information,” said Earley. “If a bot is retrieving information from a source, it has to be highly curated and have the right structure. The bot has to be able to understand what the user wants, which may require a semantic search, especially for more complex queries.”