Previous Submission

Predicting customer preferences at Starbucks and the challenges for the marketer of the future

Next Submission

Starbucks is using AI to understand customer preferences and offer recommendations even before they enter a store. The value for the company is clear, but is processing the data the real challenge?

Marketers are using AI to help them connect with their clients.  The marketing department is usually an early adopter of new analytic technologies such as machine learning and AI. Business in general use these tools to understand patterns in data and derive insights from it. As companies seek to understand better the drivers of their customer behavior, more marketers are now leaning on AI (Artificial Intelligence) to process data on their operations more efficiently. Essentially, marketing departments are using computers to scale up their analytical capacity and support the decision-making process with less human resources. Currently, it is estimated that 51% of the marketers use some kind of AI, while 71% high performers say they do [1], suggesting a correlation -but not necessarily a causation –  between analytical capabilities and performance. Some of the benefits that such technology presents to marketing departments could be generating new recommendations based on previous interactions, segmenting customers based on their shared preferences, hyper personalization of content to match their interests or predictive modeling of future behavior [2]. The increasing focus on customer centric marketing strategies, social media advertisement and virtual assistants is expected to bring this industry up to $40 billion by 2025 from $6.5 billion in 2018. This increase is majorly attributed to the use of Machine Learning to analyze big data to generate insights in a way not possible before [3].

Source: Salesforce.com [1]

Starbucks and the Digital Flywheel Program.  Based on more than 90 million transaction a week, Starbucks gather an insane amount of data and it is eager to extract value from it. In their minds, that data can provide insights on how the best tailor the user experience, independently from where they are [4]. The famous coffee brewing powerhouse Starbucks announced in 2016 that they were going to implement a new AI feature to their already consolidated Starbucks Rewards loyalty program. The technology aims to cross-analyze an enormous amount of data points such as weather conditions, time of the day, past purchases among others to start real-time interactions in the most convenient moment [5]. For instance, this cloud-based engine is able to recommend different combinations from their menu to customers that may want to try a coffee whenever they are close a Starbucks location [6]. As other digital features – such as Mobile Order and Pay checkout – of the Rewards program evolve, new opportunities to integrate the predictive technology emerges. So far, the success of the program is remarkable, with the company commenting in a report that it has “more than doubled customer response rates over previous segmented email campaigns, translating into increased customer engagement and, importantly, accelerated spend.” [7].

Generating impact from computer-based decision-making.  As the company explores ways to leverage their artificial intelligence and machine learning capabilities, the applications to generate impact are numerous. In the short term AI provides an endless source of opportunities to tailor the user experience for each client. From product development to promotions, some of the decisions usually assigned to marketing analysts can be in better hands if delegated to a well-trained machine. The computer can be more precise, agile and unbiased than a human if fed with the right data.

Predicting consumer preferences should serve as an input for the definition of an overarching strategy by the marketer, not the end strategy itself. The marketing department must see the technology as an ally to take over some of the marketing processes to let the team remain focused in the most important aspect of the job: understanding and elaborating how to implement an insight. For instance, as gaining share on user’s attention becomes more and competitive with broad adoption of tactics such as push notification on smartphones, Starbucks is bound to find new ways to bring its analytical edge to everyday interactions.

What future success looks like.  The move towards machine learning is inevitable and developing an expertise in the area is a must for the marketer of the future. However, perhaps as importantly will be ability to interact with consumers in a meaningful way. In a world where all our actions are collected and analyzed by computers, who would you pick to win? The best algorithm or the best execution? That the question that every marketers should be making themselves.

(706 words)

 

[1] Salesforce.com. (2017). Fourth Annual State of Marketing Report. [online] Available at: https://www.salesforce.com/content/dam/web/en_us/www/assets/pdf/datasheets/salesforce-research-fourth-annual-state-of-marketing.pdf [Accessed 12 Nov. 2018].

[2] Intelligence.businessinsider.com. (2018). AI will revolutionize the way marketers target audiences, but preparedness is key. [online] Available at: https://intelligence.businessinsider.com/post/ai-could-revolutionize-marketing-but-preparedness-is-key-2018-2 [Accessed 12 Nov. 2018].

[3] Artificial intelligence in marketing market to grow 29.79 percent CAGR to 2025. (2018). Business World, Retrieved from http://search.proquest.com.ezp-prod1.hul.harvard.edu/docview/2121518554?accountid=11311

[4] Boulton, C. (2016). Starbucks’ CTO brews personalized experiences. [online] CIO. Available at: https://www.cio.com/article/3050920/analytics/starbucks-cto-brews-personalized-experiences.html [Accessed 13 Nov. 2018].

[5]  (2017). Starbucks’ Digital Flywheel Program Will Use Artificial Intelligence. [online] Available at: https://www.nasdaq.com/article/starbucks-digital-flywheel-program-will-use-artificial-intelligence-cm824541 [Accessed 13 Nov. 2018].

[6] Marr, B. (2017). Starbucks: Using Big Data, Analytics And Artificial Intelligence To Boost Performance. [online] Forbes. Available at: https://www.forbes.com/sites/bernardmarr/2018/05/28/starbucks-using-big-data-analytics-and-artificial-intelligence-to-boost-performance/#444cae1765cd [Accessed 13 Nov. 2018].

[7] Starbucks. (2018). Starbucks Presents its Five-Year Plan at Investor Conference. [online] Starbucks Newsroom. Available at: https://news.starbucks.com/news/investor-day-2016-press-release [Accessed 13 Nov. 2018].

 

 

4 thoughts on “Predicting customer preferences at Starbucks and the challenges for the marketer of the future

  1. This is a fascinating topic, and your point that Starbucks has at its disposal data for 90mm transactions a week illustrates how well positioned the corporation is to employ AI. It would be neat if now Starbucks can take the learnings from their AI segmentation to inform other aspects of the business outside of push notification marketing, including their product offering (e.g., new product development, geographic product mix, etc.) or even ATL marketing efforts. An outstanding question for me: I wonder how much of the success of the targeted campaign (over the traditional segmentation email campaign) is due to (1) this is reaching people largely on mobile (as opposed to desktop) and (2) the ability to geolocate customers and ‘push’ them to a nearby location. Is the AI-driven, occasion-based segmentation really superior to its former approach, or is it the format that is in part assisting it?

  2. Erik – good to see you landed on your feet after Biometra. Thanks for sharing this insight into how Starbucks is using machine learning to improve their marketing and boost sales.

    I really liked your insight into the balance between machine analysis and human analysis. The recent explosion of machine learning has many people thinking that big data and AI are a silver bullet – able to solve any problem or improve any process. You are right to suggest caution. Just as IBM suggests using Watson as an input to the decision – not as the final decision maker – Starbucks needs to use machine learning as one more tool and not the only tool. The best analysis in the world is still ineffective if Starbucks lacks the ability to act on it.

  3. Love the question you raised around how a customer would decide on a brand in a world where every company has your data and knows your preferences. What I find super interesting, is that in that world, brand awareness once again becomes the most important factor. I think we will get to a place where, given the inundation of information, customers become increasingly picky as to what apps they have on their phone and rather than use many different brands, they will actually become more loyal to just a few. Therefore, traditional marketing once again becomes incredibly important as brands must connect emotionally with their customers and ensure they are one of these apps that is installed : )

  4. Great article Erik!
    As you mentioned, it’s a matter of developing a better algorithm than making a better decision. What could be the role of the marketers in the future then? Are they marketers or data scientists? I also liked your question about the competition. Once every company adopt Machine Learning and AI, what could be the company’s differentiation? Thanks for bringing a good subject which makes me think deeply.

Leave a comment