It’s pretty obvious advertising is reaching the next level – AI, VR, AR, and countless other acronyms, showcase advertising capabilities nobody ever thought possible. Enter The Weather Company and Watson Ads. Watson Ads doesn’t just answer a posed question, it learns and adapts answers based on a number of factors, to deliver a personal experience. ExchangeWire speak with Jeremy Steinberg (pictured below), global head of sales, The Weather Company, an IBM Business, about the practical use cases of Watson Ads in the real world.
ExchangeWire: The Weather Company announced Watson Ads earlier this year, which sounds like a sci-fi level advertising capability – how does it work?
Jeremy Steinberg: Watson Ads create a new engagement paradigm between brands and consumers. It cracks the code of how to reach a massive audience and have a personalised, one-to-one conversation with each consumer. To do this, Watson Ads take the core elements of Watson’s functionality by training on a particular subject matter – in this instance, our beta partners’ body of content – and sparking conversation with The Weather Channel app and weather.com visitors. When Watson asks a user to “name any dish, ingredient, or occasion”, it is prompting a natural language engagement, reasons through possible responses, and offers thoughtful, personalised suggestions that drive deeper brand engagement.
Whether they’re used for food, consumer health, auto, retail brands – and beyond – Watson Ads will help marketers make the most of their moments with consumers, uncover consumer and product insights faster than ever before, and illuminate previously dark data. For consumers, it will be an intuitive experience that provides powerful utility within a trusted environment.
What sort of potential use cases could there be for a marketer?
Watson’s capacity to learn, develop expertise, and interact naturally make the use cases as broad as a marketer can imagine. In the CPG food category, Watson Ads can generate brand new recipes using core ingredients, and feed emerging food trends back to marketers. In the consumer health space, Watson Ads may help individuals make better choices for their common cold and flu treatments, help prevent and manage pain, and cope with allergy season – all occasions where weather plays a major role.
Looking at current Watson clients, like Macy’s and The North Face, the retail application could act like a personal shopper that makes smarter and more tailored recommendations. In the auto category, in particular, the use case could span vehicle selection and customisation to learning about staying safe on the road in difficult conditions. The way we think about it when approaching our marketing partners is: “What consumer challenge has been difficult to solve, even with research and focus groups?” Since each Watson Ad instance will be a personal dialogue, Watson Ads outputs and insights can act like focus groups, getting to the core of user-brand interactions, informing product strategies, and even expanding into creative strategies.
Take, for example, a consumer asking by voice interaction: “What can I make for dinner tonight?” Based on its machine learning and reasoning ability from the data is has ingested, Watson can sort through ingredient and flavor profiles to make smart recommendations based on the user’s input, like what kind of ingredients they have on hand, preferred cuisines, and the occasion hand – all surfaced via dynamics ads. Furthermore, Watson’s ability to process and create context from large amounts of unstructured data will help marketers provide consumers with meaningful, true brand and product engagement. In this example, creating recipes on the fly with ingredients consumers already have and like to use from trusted brands.
But you can extend this machine learning capability to almost any category. Watson can learn about any product – whether it’s dosage information for an over-the-counter medicine, creature comforts of a new automobile on the market, the flavours available for a certain ice cream treat, and beyond.
How would the ad respond to the user query?
Watson uses machine learning – meaning, it can ‘read’ volumes of information in milliseconds to learn about a given product, category, or industry. As time goes on, it also learns based on the questions it is asked or as new information becomes available to it. It would then respond with Watson’s ‘voice’ to answer the question based on its amassed knowledge.
First, Watson learns a new subject – in the case of Watson Ads, our beta partners’ brand materials, FAQ docs, recipes, and so forth. Then, Watson uses what it has learned and trained on to provide well-reasoned answers within the response field of the Watson Ad unit. When a user speaks or types their question or inquiry into the branded Watson Ad, Watson will search through everything it has learned to find possible logical answers, collects evidence and rates the quality of the evidence, ranks all possible answers, and surfaces the most logical one. Over time, user engagement further trains the response, getting smarter the more people use it.
Watson Ads could come across as gimmicky – are consumers receptive to this sort of advertising?
When an ad provides useful content – answers to people’s questions, interesting recipes, etc – they don’t look at it as gimmicky. It’s a welcome message rather than an intrusion or a ‘stunt’. The Watson Ads experience gives them the opportunity to get all of the information they need or want about a product without having to go anywhere else – whether it’s another site or product packaging or otherwise. It’s all right there within the ad experience. We believe people will embrace this as something that brings added value to our overall site experience. Initial user testing results underscore the value Watson Ads bring to their life, perceiving them to be like ‘an app within an app’, providing a convenient and easy to use experience, with the personalisation element resonating highly.
How would you measure the success of a Watson Ads campaign? What metrics do marketers have access to for campaign optimisation?
We will monitor engagement with the ad. Meaning, did the person view the ad? Did they ask it a question? How much time did the consumer spend with it? If we see our fans taking advantage of the opportunity to ask Watson questions and learn about the product or offering presented, then we’ll know we’re on to something. We know the longer the ad product is available, the better it will become based on Watson’s machine learning process.
All of that said, given Watson Ads is such a completely unique offering, we are in the midst of exploring what other metrics might be helpful to gauge the success of the individual ad and how it can or should be optimised based on performance
We’ll start with our partners’ goals and build experiences that aim deliver on these goals. The Weather Channel apps and sites already drive reliably strong performance metrics by leveraging our proprietary WeatherFX and LocationFX targeting on one of the most trusted consumer platforms in the US. Watson Ads will draft off of this to deliver high-quality, timely, and deeply personalised content that naturally deepens engagement and brand preference. Unique to Watson Ads are the actionable data and insights – in conjunction with established ad performance and reporting metrics – that can generate deeper insights about consumer preference, non-intuitive product questions, and emerging trends.
How will this develop marketing strategies and cross-channel attribution?
With The Weather Channel app and sites reaching hundreds of millions of users, Watson Ads can serve as massive focus groups to provide our marketing partners with information about consumer questions, preferences, pain points, and opportunities – all which can be used to shape media strategies, creative concepts, and future product decisions.
We’re excited to see our first cognitive-powered product come to life. The partnership with the IBM Watson team has been fruitful. In the future, we see cognition fueling more businesses, moving into broader categories, and expanding into different kinds of ad and consumer product experiences, both on the Weather properties and beyond.
IBM Watson can understand natural language, reason, learn and interact with humans. Leveraging Watson technology, Watson Ads will help marketers outthink their current strategies by harnessing the power of cognition and the learning to:
– Better understand brand perception and favourability
– Help consumers make more informed decisions at point of consideration
– Improve the customer experience
– Aid product information
– Inform and optimise creative strategies
– Help marketers use data more effectively