Personalization at scale: First steps in a profitable journey to growth

Personalization at scale: First steps in a profitable journey to growth

Personalized communication with every customer is the future of marketing. McKinsey partners Julien Boudet and Kai Vollhardt say it’s easier than many marketers think, if you begin with the data you have.

What is personalization at scale?

Julien Boudet: Let’s start with the personalization part. Consider your daily life as a customer: you don’t want to be bothered with irrelevant coupons, emails, or texts, but you do want to be informed of offers that meet your specific needs. Personalization is an approach to customer engagement—almost a philosophy—that focuses on delivering tailored, meaningful, and relevant customer communication. On the business side, personalization allows a company to activate all the customer data available to deliver more relevant experiences for your existing customers and for your prospects as well.

Kai Vollhardt: Let me take the “at scale” part. In today’s marketplace, it’s not enough to send each customer an e-mail that addresses her by name and offers a discount based on a past purchase. You have to design and deliver tailored messages to thousands of customers in multiple interactions. That’s where technology comes in. It makes it possible for companies to truly interact on a personal basis with all their customers. That’s why it’s important to always think and talk in terms not just of personalization, but personalization at scale.

Why should personalization at scale matter to marketers?

Julien Boudet: When done well, personalization can deliver impact and growth quickly while creating better experiences for your customers. Personalization plays a critical role across the full customer lifecycle—acquisition, customer engagement, basket size, frequency of purchase, cross-sell, and/or churn prevention, among other things.

Kai Vollhardt: Because the future of marketing is in data analytics, agile, and digital—and personalization at scale is where they all intersect! A majority of classical marketing disciplines—advertising, messages, prices, services—will become much more personalized. And it builds real value. Our research and experience show that personalization, fully implemented, can unlock significant near-term value for businesses—such as 10 to 20 percent more efficient marketing and greater cost savings and a 10 to 30 percent uplift in revenue and retention. What’s more, even though immediate results can be achieved in a matter of months, adopting personalization as a practice can have a long-term positive effect on customer satisfaction

The reality is that consumers want better personalization: approximately 80 percent of them say it’s important to them. But, while 95 percent of the marketing professionals we questioned at the 2017 World Retail Congress, in industries from energy to banking, said they recognize the need and potential for personalization, only 20 percent say retailers are doing a good job at it. For many companies, personalization at scale is still a mystery.

So what is the right approach to making personalization at scale work?

Julien Boudet: Understanding the importance of data and analytics is the key value generator for all personalization attempts, and it has to be at the heart of your thinking. First of all, you need to make personalization a priority and develop a strategy to build the right foundations and operational capabilities. Establishing the strategy doesn’t have to be a lengthy project that takes weeks or months, just a deliberate top-management decision to create a path to more personalized customer experience.

Kai Vollhardt: That’s true. Many marketers believe the first priority is to fully understand the quality of their data, build capabilities in analytics, or find the right tools. But most of them can start making personalization work quickly with what they already have. One of my core clients tried for a long time to build the perfect data cube in an attempt to harmonize data. But when he took a step back and rethought what personalization he could accomplish with the existing data, he decided to prioritize the analysis of data on customer interactions he already had. That made it possible for us to launch initial pilots within days.

Julien Boudet: Yes, I can’t emphasize enough that you can find real value—often a lot of it—by working with what you have.  Naturally you should strive for the full customer-data platform (CDP and a 360-degree view of the customer), but don’t wait for perfection. Even a first model can yield great results. You can start personalized cross-selling with basic information on past behavior; you don’t have to buy new data or connect systems. We often start with a simple customer-data set that combines basic demographic information with transaction history, product details, and maybe Web data to get a preliminary understanding of the customer. A “167-degree view” of the customer that enables the activation of a few prioritized consumer use cases is better than a long quest for a 360-degree view of the same customer.

The next step, decisioning, is also not as complicated as at first it might seem. We typically start by understanding who the customer is, looking at their behaviors, and identifying the key triggers we can act on, or markers of value.  This might come as a surprise, but a lot of the initial data mining is simply hypothesis driven, and a lot of the low-hanging fruit to drive momentum in the organization is common sense.

We worked with a telecom client that wanted to upsell a new service. Starting from a hypothesis about who should be targeted, we used frequency plots of customer value and past reactions to campaigns to identify and prioritize subsegments. This substantially accelerated the launch of pilot campaigns. Based on response rates, we refined the campaigns week by week, and we got progressively better response rates.

What prevents companies from successfully personalizing at scale?

Kai Vollhardt: We’ve observed four common roadblocks. First, many companies are collecting and storing massive amounts of data but are having trouble finding and merging the most relevant subsets. Instead of generating and assembling more and more data, companies should focus on identifying and collecting the right data. Sometimes less data actually put into action is more effective than adding the most sophisticated external data set.

Second, many companies still think in terms of seasons or general events rather than appropriate triggers. Triggers are the specific occasions when a particular message will be most valuable to a customer. A customer moving to a new home, for example, is a trigger for an energy company. In my experience, trigger-based actions have three to four times the effect of standard communication.

Third, personalization at scale requires agile, cross-functional teams, and many companies are still stuck in a siloed way of working. Running an agile project once is relatively easy, but making it stick and scale is difficult. Those cross-functional teams make it easier to apply a test-and-learn approach, as all relevant experts are in the room and insights can be shared instantly, which is a prerequisite for personalization at scale. As a result, the number of campaigns brought live can easily increase by a factor of ten or more. Test-and-learn or not being afraid to fail can be a significant cultural shift for traditional companies. And it’s a lot of fun for employees—something we always find amazing.

Finally, the right tech tools and infrastructure have to be in place to test successfully on a large scale across the entire customer base, and this can feel overwhelming. However, technology has advanced a lot, and there are several simple and powerful solutions available.

Julien Boudet: I’d like to emphasize what you said about cross-functional teams, Kai. Companies really need to engage with marketing, operations, and tech experts to build organizational capabilities that can sustain change and establish new ways of working. That comes from both training existing staff and recruiting new top talent. For companies that need to fill multiple roles with specific skill sets, specialist competence is essential. But it’s not enough. Specialists also need to be “translators,” who can communicate insights comfortably and effectively across business functions.

What technology is needed to successfully personalize at scale?

Julien Boudet: It’s understandably difficult to make the right technology choices, since the landscape is very dynamic, complex, and not particularly transparent, and it’s unclear what different providers are doing and what advantages they offer. In the 2018 edition of Scott Brinker’s Marketing Technology Landscape overview, he lists ~7,000 marketing-technology solutions.

But one way to start to simplify the technology side is to understand that it has to enable three things:

  1. Integration of consumer data to develop a clear and complete view of your customers, ultimately through customer-data platforms
  2. Decision making based on signals given during the customer journey. It can be as simple as an Excel VBA-macro that helps a call-center agent shape a personalized offer to a customer or as sophisticated as a centralized decisioning engine, or “brain,” that can interact with outlying systems to make real-time decisions based on consumer signals.
  3. Distribution of coordinated content offers across audiences and channels in real time, with room for teams to adjust them based on feedback

Kai Vollhardt: Simply put, technology is crucial to scaling personalization, and a customer decision platform has to be at the heart of it. We found that 50 percent of companies that outperform the market feel they have the tech tools they need, compared with only 16 percent of their poorer-performing peers. In the long run, the real value comes not just in developing the three elements of the operating model but in getting them to work together seamlessly. Once consumer data has been collected, it needs to be prepared and transformed to uncover additional insights: the more comprehensive the customer view, the more accurate the predictions of the decision-engine models will be. But while technology is very important, technology alone won’t solve all the personalization challenges.

Julien Boudet: Absolutely. Technology and analytics play a big role in driving impact beyond pilots and scale, but at the end of the day, companies don’t achieve the impact without changing their internal operating model to be agile, focused on key customer KPIs, cross-functional, and driven by rapid decision making.

About the author(s)

Julien Boudet is a partner in McKinsey’s Seattle office, and Kai Vollhardt is a partner in the Munich office.

Related Articles