Earlier this year, several leading clothing brands saw an unusual spike in sales of tops—but no corresponding increase in sales of skirts or pants. The reason: amid the coronavirus pandemic, customers were adapting their work wardrobes from office attire to appearances on Zoom and other videoconferencing platforms.

This anecdote is just one illustration of a market-changing trend that has been accelerating for several years, beginning well before the pandemic: consumer preferences have been evolving rapidly—almost continuously—and have become increasingly volatile, mutable, and uncertain. And they have outpaced companies’ traditional abilities to track, anticipate, and respond to trends.

To adapt to what has become “certain uncertainty,” companies must find new ways to interact with consumers and gain insight into that uncertainty. It’s not as impossible as it might sound: they can rely on new “eyes and ears,” courtesy of the exploding availability of vast quantities of data from an increasing variety of sources as well as new capabilities afforded by technologies such as artificial intelligence to process, learn, and respond in near real time. These technology advances will enable a new kind of dialogue between companies and consumers that will lead not only to deeper insights into what consumers want but also to a proliferation of offerings from companies seeking to meet consumer needs. We call this emerging model the new consumer conversation.

The new consumer conversation will redefine the key success factors for consumer interactions and create substantial competitive advantage. But companies must move fast to be among the first: consumers can’t and won’t have room for conversations with everyone.

Evolving from Traditional Approaches to Tackle Certain Uncertainty

Traditional interactions between companies and consumers typically rely on a four-step process:

• Data Gathering. First, consumer data (mainly sociodemographic and behavioral data based on past transactions or interactions with the brand, complemented with consumer surveys) is amassed.
• Segmentation. Rule-based techniques such as multivariate regressions are then used to develop differentiated consumer segmentation models. Often, this segmentation remains static for a couple of years, assuming that consumer profiles and preferences conform to a relatively stable structure.
• Decision Making. Based on the segmentation models, marketers and consumer strategy teams make decisions tailored to each segment, from preferred channel choice to product recommendation to message customization.
• Implementation. Finally, the defined decisions are implemented, in the form of an adjusted marketing budget or a new campaign.

This approach is increasingly inadequate because it can’t keep pace with the growing volatility of consumer preference and the magnitude of the shifts that result.

A closer look reveals that companies seeking to understand consumer preferences today face challenges from three fronts:

• The unprecedented rate at which behaviors are shifting is weakening the predictive power of historical data alone. This means predictive models built on historical data must be replaced by near-real-time alternatives, acknowledging shifts and adapting to them as they happen.
• Uncertain times compound the difficulty of anticipating the pace and magnitude of behavior changes. History provides good examples. The 2008 global financial crisis, for example, led to rapid shifts in consumer buying behaviors, instantly invalidating the traditional equilibrium between price and perceived value.
• The value of data obtained via consumer surveys has been undermined by the compound effect of long (albeit necessary) time lapses between data gathering and insight generation and by consumers’ ever more counterintuitive behaviors in crisis times—behaviors that are often not aligned with their own statements. According to Jing Daily, a trade magazine that tracks the Chinese luxury industry, a consumer survey conducted at the peak of the COVID-19 crisis found that 56% of Chinese consumers said they would spend less on luxury—and yet, one month later, booking volumes for luxury cruises were up 9% among Chinese consumers compared with 2019.

With the emergence of new AI capabilities, companies can circumvent today’s growing limitations, opening up new opportunities for consumer interaction. Even just a few years back, companies were forced, for reasons of time and money, to make a tradeoff between the size of the audience reached and the variety and depth of the interactions that make up that reach. But AI is a game-changer that is transforming the tradeoff into a dynamic feedback loop. With AI at work, intelligent processes happen fast, at scale, and at marginal cost. Lessons learned from a greater volume of interactions enable companies to dynamically update the variety and depth of future interactions. AI is enabling new approaches to consumer interaction that are personalized, self-learning, responsive, and quickly scalable.

Reinventing the Consumer Conversation

The goal for companies now is to engage consumers in the continuously evolving, two-way, and personalized and responsive discussions that constitute the new consumer conversation.

Continuous, Iterative Processes. The new consumer conversation replaces the traditional interactions with iterative and interactive alternatives. The value of treating the interaction as a conversation is that it frames companies’ overarching goal—to analyze and respond to shifts in consumer behavior—as a dialogue that allows the company and the consumer to interact on an ongoing basis. This close and continual connection is vital now that consumer behavior has become so mercurial.

These ongoing conversations yield a myriad of fresh data on consumer behavior. With that insight, coupled with externally obtained data on macrotrends at work, companies get a more holistic understanding of consumers’ preferences and can seek out cues that signal subtle shifts in consumer expectations as they happen. This enables near-real-time self-updating of consumer interactions through continuous learning. Companies are then able to either perfect existing interactions or explore new conversation topics and modes suggested by the cues.

Spotify’s new Shortcuts feature is an example of how an ongoing consumer conversation can yield fruit. Spotify used vast amounts of data on listening habits, newly released music, and macro musical trends—and leveraged heuristics and machine-learning models—to upgrade the user experience. The result: easier access to an individual’s most-used content and a tailored selection of new music to explore. Frequent feedback loops make it possible to constantly improve the experience—recommendations evolve to reflect consumers’ listening habits at different times of day, for instance.

Two-Way Connections. Unlike the one-sided approach, the new consumer conversation gives consumers novel ways of interacting that not only allow them to share feedback, content, and opinions but also elevate their contributions by showing them that their input is genuinely in demand and taken into consideration.

Cosmetics brand Glossier has explored such dynamics. The company closely monitors consumers’ engagement and commentary on various platforms and constantly iterates with consumers based on these insights. The process unfolds as follows: Looking across platforms at consumer behavior—from purchasing habits to navigation to comments—Glossier identifies interests, whether stated directly or signaled more subtly. Glossier teams build on these insights to produce new content; consumers are encouraged to respond with related, self-created content, such as photos or videos. The Glossier teams then leverage this response from consumers to further refine and personalize content. It’s truly a two-sided conversation.

Personalized and Responsive Discussions. Static, preestablished rules once defined companies’ interactions with consumers, but the new consumer conversation changes all that through personalization that highlights the individual consumer’s wants and needs. This personalization manifests in the tailoring of messages, the choice of conversational tone, the push of services, and the recommendation of particular offerings and promotions.

Consider how a global beverage company is bringing personalization to vending machines. The company is collecting and analyzing massive data sets, including data gleaned from social media, to develop a nuanced understanding of where, when, and how customers consume its products. Then, it tailors ads and offerings for individual consumers depending on their current geographic location. The company also links vending machines digitally to the company’s smartphone app so that customers can digitally purchase drinks, redeem loyalty program rewards for purchases, and even preorder before arriving at a machine.

Redefining Success Factors

Implementing the new consumer conversation requires a paradigm shift along three fundamental dimensions: new data, new processes, and new decision making. (See the exhibit.) Such a shift redefines the key success factors for consumer interactions and emphasizes the urgency of moving from static communication to dynamic conversation.

New Data. The new data that fuels the new consumer conversation is mainly unstructured, coming in large volumes from an ever-broader scope of sources and updated as frequently as possible. As a result, key success factors shift from data quality to data quantity and freshness.

This new data provides companies with the necessary substance to feed the algorithms. They are able to identify even weak signals of consumer behavior shifts and to cope with rapid data obsolescence. For example, a Ben & Jerry’s market research initiative leveraged AI and machine-learning capabilities to crunch data from song lyrics and other content. The effort uncovered a new consumer trend: ice cream for breakfast. With this insight, Ben & Jerry’s quickly launched breakfast flavors, two years ahead of competitors.

In the course of embracing new data, companies need first to expand their understanding of what qualifies as a relevant data provider and develop a rich ecosystem of new data sources. This means that companies should look beyond digital retailers to access data from broader sources. For example, companies can leverage publicly available data including weather, news, public events, and the evolution of search topics. Aggregating these weak signals can be a valuable asset, helping companies detect strong trends at a fine granularity and then refine a more holistic and comprehensive view of customer habits, preferences, and needs.

Companies also need to ensure data freshness, which has implications for how they access data. That is, companies must build the right data infrastructure for near-real-time sharing with their ecosystem of partners. Finding the right ecosystem of partners is particularly important for industries with low interaction frequency (such as automotive and real estate), because the opportunities for data collection are scarcer.

New Processes. The primary new processes required for the focus on AI-enabled, near-real-time, adaptive reactions, taking into account the wide set of available data, identifying new patterns, and dynamically personalizing the immediate response to each consumer.

Machine- and deep-learning methods at the core of the new processes create a continuous loop of analysis and action. Each cycle produces new consumer responses and, therefore, new data about consumer behavior. When the system can in turn learn from this outcome data, that new insight fuels the next cycle of action and response. A simple example is Facebook’s text-understanding engine, DeepText, which has the functionality to contextually understand not only the content but also the emotional sentiment of thousands of posts to track new hot topics, shifting perspectives on topics, and early signs of discontent. The engine then suggests associated actions, continuously learning from previous interactions.

With these new processes in place, key success factors evolve from the accuracy of analysis to the speed of reaction and learning. With the ever-shorter shelf life of consumer insight, success will be a function of the speed at which companies derive new insights and act on them.

Companies thus need to invest dramatically in AI—from both the technological and the human standpoints. They must remain up to date with fast-moving tech advances. The ability of systems to continuously refine the seamless analysis and action loop is likely to become all the more powerful. Reinforcement learning is a good example: going a step beyond deep learning, it is able not only to draw insights from unstructured data but also to test the effectiveness of its previous insights and autonomously enhance its precision and quality. And from the human standpoint, skilled talent is crucial for companies to master the capabilities and harness the potential of AI.

New Decision Making. In the new consumer conversation, decision making is moving from the last step of the traditional sequential approach toward interaction in order to become an omnipresent and overarching environment within which AI can act in near real time. This new approach to decision making is shifting key success factors from case-by-case decision making to carefully considered and global framing in which AI capabilities and human creative intelligence are synthesized to set the “rules of the game.”

As important as AI is to the new consumer conversation, it is also critical to remain mindful of its limits. First, algorithms can introduce bias into consumer behavior analysis by compounding biases already present in input data. Second, AI-based systems can be perceived as offputting given their power to anticipate needs that consumers themselves may not yet be aware of. Finally, creativity is currently beyond the boundaries of what AI can do. Rather, it relies on the kind of counterfactual thinking—beyond existing data and frameworks—at which humans are best. Thus, AI must be used in conjunction with human input when exploring creativity.

Companies need to carefully track the evolution of the AI systems with human attention and intervention, ensuring the implementation of responsible AI. And companies need to implement the necessary transparency and “tact”—without which consumers might experience personalization as an intrusion or even as a violation of privacy. For example, even if telcos are able to spot a consumer’s interaction with a competing provider, they should refrain from making contact right away. Instead, they should wait a few days before reaching out subtly with new promotional offers.

Creating New Competitive Advantage

Embracing the new consumer conversation gives companies the opportunity to gain a substantial competitive advantage by breaking the traditional compromise between cost and connection, by fostering hyperresponsiveness and resilience even in uncertain times, and (it almost goes without saying) by creating a greater sense of belonging, affiliation, and engagement among consumers.

Beyond the Cost Versus Connection Compromise. In traditional approaches, and largely because of cost constraints, there was a necessary tradeoff between reaching many consumers with a standardized message and reaching few consumers with personalized messages. But with AI at the core of the new consumer conversation, companies can break the traditional tradeoff between cost and connection at the interaction level. AI eliminates the requirement to choose between the reach and the richness of interactions by making it possible to, on the one hand, identify new patterns of consumer behavior at unprecedented scale and accuracy and, on the other hand, respond to these patterns in real time. And both come with limited marginal costs.

For example, e-commerce platform eBay partnered with Phrasee, a pioneer of AI-powered copywriting, to generate millions of marketing copies at scale, in just a few clicks. Relying on natural-language generation and deep-learning models, Phrasee’s technology is able to generate human-sounding language, customized for eBay’s brand voice and adapted to the constantly changing behaviors and preferences of the platform’s 100 million email subscribers. Phrasee’s technology, operating at scale, also unlocks greater efficiency and lower costs, with each campaign setup requiring only five minutes. In the US alone, the initiative resulted in an uplift of almost 16% in the average open rate and a more than 31% increase in the average click rate, yielding consistent ROI on all campaigns.

Hyperresponsiveness and Resilience, Even in Uncertain Times. With AI at the core, the new consumer conversation is enabling a hyperresponsive learning loop. It enables companies that embrace it to detect and understand consumer behaviors shifts in near real time. It also arms companies to respond not only quickly but adequately to those shifts, pivoting to match new needs and expectations in the most relevant fashion. As a result of this near-real-time responsiveness, companies gain greater relevance and reliability, grounding their resilience in unpredictable times.

With its May 2020 search engine optimization update, Google demonstrated the value of hyperresponsiveness in the context of the high uncertainty surrounding the COVID-19 crisis. Driven by the never-before-seen surge in searches for a single topic over a sustained period, the digital giant updated its search-ranking criteria to reflect users’ new definition of relevant content: more-local information, especially on sheltering updates or the latest information on testing, for example.

Greater Consumer Belonging, Affiliation, and Engagement. With their opinions genuinely requested, their needs and preferences thoroughly considered, and conversation topics that extend beyond transaction to match their interests, consumers experience an advantageous “return on relationship investment.” They are thus more likely to prefer and proactively interact with brands that, by embracing the new consumer conversation, offer these benefits.

For example, UK-based clothing retailer Asos released an app that allows consumers to upload a photo of a favorite celebrity wearing a coveted outfit. Powered by AI, the app is able to scan the Asos clothing database against the photographed outfit to suggest similar but more affordable products. Such conversation between the consumers and the company resulted in almost 50% more product reviews and increased the likelihood of visitors returning by 75%.

Overall, consumers are more likely to engage when they feel understood, and companies are more likely to succeed when they listen carefully and then respond adequately to consumers’ shifting preferences. Such a dynamic creates a new kind of relationship, an evolutionary one in which interactions improve with each interaction. Companies are thus equipped to keep pace with consumer requirements as they shift.

However, consumers cannot and will not invest in too many two-way conversations at the same time. Given that, companies that quickly embrace the new consumer conversation can quickly gain a significant first-mover advantage.

To capture the new competitive advantage, companies must act now along the three dimensions of the necessary paradigm shift in order to successfully implement the new consumer conversation. With this transformation of their consumer interactions, companies will be able to seize a substantial new competitive advantage before others. Importantly, this constitutes an opportunity for incumbent companies to upgrade their consumer interactions to state-of-the-art standards that will no longer be the preserve of digital leaders alone.

BCGの戦略シンクタンクとして、アイデア創出に有効なテクノロジーを活用し、ビジネス、テクノロジー、科学分野からの新しい価値あるインサイトを探求・開発しています。ビジネスリーダーを巻き込んで、ビジネスの理論と実践の境界線を広げ、ビジネス内外から革新的アイデアを取り入れるための刺激的なディスカッションや実験を行っています。2022年7月に日本における拠点であるBHI Japanを設立しました。