This year, AI firmly attached itself onto enterprises’ strategic roadmaps. Deloitte, in its most recent State of AI report, found that 42% of surveyed executives believe that adopting AI will be of “critical strategic importance” within the next two years. For many, it’s not a matter of when to invest in AI, but how and in what form, as companies investigate the potential benefits that autonomic and cognitive technologies can deliver for their businesses.
This article is the first in a series that we’re calling IPsoft’s 2019 AI Trends detailing what we believe will be the dominant developments and movements in the Enterprise AI market next year. Our predictions and viewpoints are influenced by our work with clients, who are already realizing new benefits and innovations with our products, as well as our 20 years of experience in the AI industry. These blogs will be published regularly through the end of the year.
In the coming year, Artificial Intelligence (AI) will dramatically affect the way people search for content online, as well as the way enterprise brands optimize for searchers. AI has already impacted search by helping companies like Google and Microsoft optimize algorithms to display the most relevant content for users’ search phrases. Additionally, voice assistants have provided people will a quick, hands-free way to find a specific product or service.
Unfortunately, both search methods rely too heavily on the searcher using precise keywords, or browsing through link after link of recommendations, to find relevant responses. We expect that within the next year, a new form of search will begin to transform how searches are conducted and how responses are delivered. This overall trend is labelled Conversational Search across the broader market, however, we have a different vision for how this trend will evolve starting in 2019 and beyond.
Conversation + Search ≠ Conversational Search
SearchEngineJournal published a recent article predicting how brands will need to shift search strategies to account for AI. In the piece, the author ponders how voice will disrupt the standard keyword-plus-intent method by which users currently find content online.
“Voice also introduces a new layer of complexity into the equation because search terms are phrased differently and are far more varied,” the author states. “For example, someone searching for my services using traditional text search might use a search phrase like ‘Tampa web design’ but when using voice search, they would likely use a more conversational search phrase like, ‘Which web design company in Tampa designs websites for contractors?’”
Another interesting post on MarTech Today examined the ways in which chatbots will be used to help users find specific products and services on company websites. “Chatbots are also becoming more and more popular,” MarTech Today notes. “Many brands are utilizing chatbots to present information to consumers as quickly as possible. Instead of sifting through content on a website, chatbots will allow the consumer to enter specific questions and get their response immediately. This process would potentially replace the need to search in a traditional manner.”
Unfortunately, both articles fail to anticipate a deviation from the common search procedure. Today, a person utters or types a set of keywords. The search yields a list of results. The user chooses a single result and leaves the search engine (or the user enters a new set of keywords and the search engine presents a new list of links). The search engine is no longer part of the browsing process.
In traditional, conversation-based search, a person speaks a question, a voice assistant provides a response, or a list of responses, and the voice assistant directs the person to the most relevant information. For example:
Searcher: “Can you tell me which used car dealerships are nearby?”
Voice Assistant: “There are two car dealerships within five miles of here. John’s SUVs and Mary’s Luxury Vehicles.”
In this example, traditional voice serves as a call-and-response mechanism that directs users to exactly what they knew they wanted when they started searching: a used car dealership nearby. Once the searcher calls, or clicks on, a dealership link, the search engine ceases to be useful to the searcher. Beginning in 2019, we’ll begin to see search engines, powered by AI, prove valuable from the beginning of a search through the purchase process and beyond, into customer service and resale.
Conversational Search in Action
Let’s continue with a car-buying example to detail how Conversational Search can be a game-changer for consumers and enterprise brands. We envision that this trend will be enabled with a cognitive AI interface layered on top of a search engine and integrated with specific brand websites.
Here’s how it will work: Let’s pretend a customer named Mary has never purchased a car, and doesn’t know which models and features make sense for her. Her first question might be something along the lines of: “Where can I buy a used car for less than $10,000?”
A Conversational Search Engine, powered by AI, will respond by asking very open-ended questions such as:
- Would you like me to help you find a car?
- How concerned are you about gas mileage?
- Are you interested in a hybrid vehicle?
- What is your favorite color?
Based on Mary’s answers and preferences, a Conversational Search engine narrows down the list of potential cars and dealerships. Mary can also interject with specific questions and statements to ensure that her exact needs and requirements are being considered, such as:
- I like cars that don’t have keys. Can you show me those?
- I would like Wi-Fi. Do any of these models have it built in?
Conversational Search will display only those vehicles that offer these features. At this point, Mary’s options should be narrowed down to only a few makes and models from a few specific dealerships. Conversational Search can then use a list of automotive technical specifications, and its vast automotive knowledge, to detail differences between various purchase options so that Mary makes an informed decision.
Once Mary has selected her ideal model, Conversational Search will then ask specific questions about available features that may be of interest. Heated steering wheels and seats, automatic high beams, dual-zone automatic climate control— Conversational Search will work with Mary at her own pace to customize a vehicle with features to her liking.
When this process is complete, Conversational Search will help Mary book a test drive at the closest dealership to her home. At this point, the conversational interface will remain the same, but the web browser behind it will display dates and times for available test drives. Mary will tell the interface which time works best, and the interface will book a test drive for Mary directly with the dealership—without Mary having to leave the interface, contact the dealership, or enter her personal information.
Don’t worry, Conversational Search will still be useful for the kinds of basic queries for which you use search engines today. The larger point is starting this year consumers will begin to reset their expectations for they interact with search engines, and companies will need to consider bringing a certain level of intelligence to how their products and services and delivered to consumers via Conversational Search.
Beyond the First Search
Search engines remember the terms for which you searched, the links on which you clicked and the products and services you purchased as a result of your search. However, with their call-and-response approach to results, they can only deliver information via links. With Conversational Search, your historical data can be used to refine future purchases, customer service interactions and more. For example: Mary used Conversational Search to book the aforementioned test-drive. She bought the car she tested and she’s owned it for 18 months. Unfortunately, the engine light is turned on and she doesn’t know what to do about it. With Conversational Search, Mary doesn’t need to type in a long-tail phrase, such as, “Honda Odyssey LX 2006 engine light is on what do I do?”
As Conversational Search worked with Mary to buy the car, Mary can simply type or utter, “My engine light is on, what should I do?” The AI will confirm that Mary is referring to her Honda, ask if she’s tried a list of known fixes, and if fixes haven’t worked, the system will help Mary book a visit back to the dealership. A resolution that might take a few minutes of conversation would have taken hours of clicks, research, phone calls and data repetition via traditional search.
The same is true for Mary’s second car-buying experience. Instead of starting from scratch, the process begins with an already-established wealth of knowledge:
Conversational Search: “What features would you like your new car to have that your Honda didn’t?”
Mary: “I’d like something less expensive, with better gas mileage and instead of blue I’d like red or black.”
Conversational Search has a detailed knowledge of Mary’s current car, and because it knows every feature available for the cars listed on associated websites, it can cross-reference and provide Mary with a list of suitable options.
A Realistic Projection for Conversational Search
Within the next year, you’re not likely to log onto your laptop and experience the kind of search Mary experienced in our example here. We’re still years away from a conversational interface for search that would span the entire public internet and utilize the kind of intelligence detailed in this post.
However, as we move into 2019, you should expect to see this level of sophistication on large retail, bank and automotive websites (among other large enterprises that have been investing in AI experiences). You will not be able to click on a search engine and immediately work with a digital personal shopper, but you will be able to log into your bank account, or your favorite e-commerce site, and begin having spoken or typed conversations with a digital assistant who knows your purchase history, has an expert level of understanding about the company’s product catalog, and can help you find exactly what you need—even if you don’t use the exact keywords.