Consumers recently reading People.com who noticed an ad from airline giant Emirates got a pleasant surprise: they could talk to it. What’s more, the ad responded — giving them answers to their travel questions and even recommending vacation packages.
The ad appeared next to an article about pop singer Jordin Sparks’ recent visit to Maldives (one of Emirates’ destinations). Readers could explain, for example, that they wanted to visit Maldives. The ad would respond with details — flight info, lodging, and what to do there. Then consumers could click through the ad to go directly to booking on Emirates’ website.
The ads are running in digital publications targeting travelers in nearly a dozen markets including Los Angeles, New York, Chicago, and Seattle.
Sure, travel brands have been using chatbots for a while now via their websites and Facebook Messenger to engage consumers through conversation. But Emirates is raising the bar and putting an artificial intelligence-powered bot right in its online ads.
“The overall goal for Emirates … was to deliver an exceptional digital shopping experience that improves online engagement,” says Noreen Henry, CEO of WayBlazer, the company that offers the technology behind the ads. “So, we developed a technology that let travelers find and book vacation packages more easily [and] created a first-of-its kind advertising campaign.”
WayBlazer comes with serious pedigree: founded by Terry Jones, who started Travelocity and Kayak.com, and Manoj Saxena, the former general manager of IBM Watson. The company began working with Emirates in 2017 to develop a “conversational recommendation engine” which it eventually named Emma. She’s powered by WayBlazer’s proprietary technology, which pulls from reviews, blogs, images, and videos to give travelers personalized and relevant recommendations.
It’s just one example of how artificial intelligence (AI) is taking advertising out of the annoyance column and putting it in the cool category. But even more important than cool is that the ads seem to be working. So far, these smart ads have driven an 87-percent lift in engagement for Emirates compared to traditional click-through ads.
What’s more, they’ve given Emirates valuable data about their customers, including impressions, interactions, clicks, bookings, intent, segment, and demographic information, like location and even the device they were using. Emirates is now examining how to use all the data to tweak future ads.
“Anything that removes the friction, we want to try,” says Ailsa Pollard, senior vice president of Emirates Vacations. “This is allowing them [consumers] to remain where they are … putting it as close to them as possible, with limited friction.”
AI’s ‘Tremendous Applications’
Henry says even though AI has been a buzzword in marketing for a while now, marketers still don’t tap its full potential with any regularity.
“Right now, AI is mostly used to automate simple functions like email marketing, but marketers need to understand that the technology has a tremendous number of applications across channels, including machine vision for image processing, online and voice-powered recommendation engines, and more,” Henry says.
One of AI’s most prominent and evolving applications has been conversational commerce, which combines conversation design and natural language understanding.
“Marketers are starting to use these functions throughout the entire customer journey, particularly to enhance the customer experience,” Henry says.
How does this look in practice? Henry says brands can gauge a shopper’s behavior and intent through each channel, whether that’s digital advertising, messaging, websites, email, in-app, etc., and then act on the data to deliver highly personalized experiences.
“Each touchpoint is aggregated in one stream and targets consumers with the right message and offers at the exact right time — and that results in increased customer loyalty,” Henry says.
The techie term behind all this is natural language processing (NLP) — a branch of AI that helps machines understand what people write or say. In essence, it’s what makes chatbots and voice-activated assistants like Siri and Alexa work.
Henry says one of the biggest opportunities with NLP is the new frontier of voice search (see sidebar, below). She cites research from comScore that says about half of all consumer searches will be through voice by 2020.
“To meet this trend head on today and optimize for it tomorrow, marketers should look to write search engine optimization content in a natural, conversational voice that answers the questions consumers are asking,” Henry says. “And they should look for ways to better understand the intent behind a question and how to deliver more accurate results based on that context.”
She adds “it’s critical” for marketers to view AI integration as a journey rather than a destination.
“Using AI across marketing functions is about constantly testing and learning, and then using that data to iterate and make better decisions for the customer,” she says. “Marketers also need to look for opportunities to bring in strategic partners that can extend the value of AI initiatives — partners who share the same mindset, passion, and agility of organization.”
Another AI expert, Paul Roetzer, founder of the Marketing Artificial Intelligence Institute, which works to teach marketers about AI’s potential, says a good place for marketers to start is to examine their data-driven and time-intensive activities.
Once they understand what AI’s capable of, they can then match AI tools to those tasks to increase efficiencies.
Roetzer developed what he calls the five Ps of AI to sum up where AI might help marketers:
- Planning: Using AI to handle search engine optimization, keyword selection, and topic clustering.
- Production: Having AI curate and create content to tell a data-driven story at scale hundreds or thousands of times instantaneously.
- Personalization: Allowing AI to set rules for personalizing marketing communications.
- Promotion: Giving AI a budget and the creative, for example, and let it handle all media buying. It can also run all the infinite variations and make changes based on performance data.
- Performance: Using AI to analyze and determine insights from data, and then figuring out what to do next.
T-Comm at Hulu
Another technology that’s making the grade among advertisers is TV commerce (T-comm), sometimes called transactional TV. A good example of T-comm at work came this spring when viewers of Hulu, the over-the-top (OTT) streaming video service owned by Disney, Fox, Comcast, and Time Warner, began to shop with their remote control and video game controllers.
It’s part of Hulu’s a new advertising unit developed in partnership with the interactive TV firm BrightLine. The unit debuted in March with a campaign for the Warner Bros. film “Tomb Raider.”
The ad featured a trailer for the film, paired with an overlay that asks the viewer to press a button to buy tickets. A screen then pops up with choices for date, time, and theater location (found using location data). Users can check out using their email address.
Jim Keller, vice president of sales at Hulu, says today’s TV is “a whole new game” with new rules.
“It requires a new playbook with a foundation in respect for the viewer,” Keller says. “By offering more interactive ad capabilities to brands, we’re able to do just that by delivering more choice and control to today’s viewer.”
He says success has been quickly realized, with viewers spending twice as long engaging with a brand when the ad is interactive, and some brands are seeing better than a 160-percent increase in purchase intent compared to standard ads.
“As one of the first OTT providers to enable interactive ads in the living room, we knew there was more we could give brands,” Keller says. “The launch of industry’s first television commerce ad was a giant step for not only Hulu, but also for viewer experience.”
Hulu says it doesn’t plan to stop at movie-ticketing. “We’re exploring how this technology can be applied to other categories as well, such as consumer packaged goods and quick-service restaurant advertising, among other categories,” Keller says.
Rob Aksman, BrightLine’s chief strategy officer and co-founder, says marketers who buy TV advertising need to understand that connected TV (CTV) is the new TV.
“The legacy way of delivering commercials will no longer cut it when viewers are watching TV via personalized, data-driven content experiences,” Aksman says. “BrightLine’s technology allows advertisers to make their CTV ads smarter, personalized, and interactive to capture and act on viewer intent.”
For example, an ad for allergy medicine would feature a dynamic overlay showing viewers their current pollen forecast. Or an auto ad lets viewers click to see the vehicle in different colors, angles, and configurations — all with the remote control.
“For years, the TV industry has had multiple fits and starts in search of a single, scalable standard for enhanced ads in the living room, and one now exists,” Aksman says.
The other critical point to know, he says, is that advertisers can develop an enhanced ad and run it across NBC, CBS, A&E, Fox, and Hulu — among other major broadcast and cable networks’ OTT offerings.
Sidebar: Voice Lessons
How to get found when consumers search with their voices.
When consumers speak, smart marketers listen. And smart marketers know more and more consumers are speaking, literally using their voices — instead of their fingers — to conduct their internet searches.
In fact, comScore predicts more than half of all internet searches will be done by voice in less than two years.
“We won’t type, we’ll speak, and every business needs to understand voice search is the single biggest change in the digital marketing landscape that businesses have faced in the past 10 years,” says Duane Forrester, vice president of industry insights at Yext, a digital marketing firm in New York. “Marketers are poised to see a big shift in user interface and data retrieval paradigms [and] they need the right facts in response to consumer voice queries.”
With voice search, powered by natural language processing (NLP), some say search will be better and far more intuitive and revolutionize how people seek information, conduct pre-purchase research, and find new products.
The key point? Voice search often delivers different results and ranks — so marketers need to ensure they can be found via voice and not just through traditional searches.
Backlinko, a search engine optimization company, analyzed 10,000 Google Home search results to better understand how voice search works. Specifically, it examined several potential voice search ranking factors like page loading speeds, content length, social media popularity, and more.
Here are the findings, along with some tips that might help marketers improve their chances of appearing in the results of a voice-based search:
- The average voice search result page loads in 4.6 seconds (52-percent faster than the average page) so to rank high in voice search, ensure your site is a fast loading one.
- 70.4 percent of result pages are secured with HTTPS (versus just HTTP) so using HTTPS may improve your chances of appearing as a voice search result.
- The typical voice search result has just 29 words, so Google likes brevity. Aim to make your answer snippets concise.
- 36.4 percent of voice search results come from pages that use Schema (schema.org), which is only slightly higher than the worldwide average of 31.3 percent. So, while Schema might help your overall SEO efforts, it doesn’t appear to impact voice search rankings
- Authoritative and trusted domains tend to produce voice search results significantly more than non-authoritative domains.
- The average voice search result has 1,199 Facebook shares and 44 tweets. Keep your content valuable and highly-shareable.
- The average Google voice search result is written at a ninth-grade level. Use simple, easy-to-read language.
- Very few voice search results had the exact query in their title tag, so you don’t necessarily need to create individual pages for each voice search query.
- The average word count of a voice search results page is 2,312 words, so it appears Google tends to source voice search answers from long-form content. Also, FAQ pages tend to perform particularly well in voice search.