The Challenge of Personalization at Scale

Personalization is a crucial weapon in a marketer’s arsenal in delivering a world-class experience for its customers and bringing value to an organization according to an article by McKinsey & Company called A Technology Blueprint for Personalization at Scale. Personalization at scale has the potential to create between $1.7 to $3.0 trillion¹ in new value across multiple industries from retail, and consumer packaged goods to travel and banking.

The value of personalization is in the trillions of dollars

The authors Sean Flavin and Jason Heller describe four critical challenges in providing personalization at scale:

Data foundation

Effective personalization requires data that is often trapped in silos. To build a Customer Data Platform (CDP) which typically consists of first-party data and combine it with a Data Management Platform (DMP) or a data lake requires data from first-party sources such as ERPs, CRMs, inventory management systems, Kiosk/POS, its website and third-party sources such as geolocation, weather, and time of day. Compounding the challenge of multiple data sources is its location: the data could sit in multiple places such as on-premises in a data center, in the public or private cloud or a hybrid environment. Once the data has been integrated, the DMP (or data lake) has to identify an individual across systems and tie them together as the identifier for a person is likely to be different from system to system.

Decisioning

For the data to be useful, data from multiple sources has to be evaluated and utilized in real-time so an organization can take the proper action. This is only possible through advanced analytics (predictive analytics) which usually requires the development of machine learning models with custom algorithms that can score a visitor’s behavior, predict their next activity and trigger an action in real-time to engage them in a meaningful way. McKinsey admits that such capabilities aren’t easy to implement and often require dedicated resources and specialized skill sets from data scientists. Not surprisingly, this is one of the most difficult challenges to overcome as McKinsey Global Institute predicts that the US economy will be short 250,000 data scientists by 2024². Thankfully, they predict that innovative solution providers (such as ZineOne) will fill this gap.

Design

The authors contend that Marketers are unable to match the pace of experimentation and the volume of data that is needed for effective personalization especially if it’s to happen in real-time which is most effective. One way to manage this challenge is to break up the experiments into small, manageable chunks and then expand from there to eventually cover the entire population of visitors, customers, and partners that interact with a company.

Distribution

The last mile in delivering personalization at scale is often the most critical and requires real-time orchestration between the multiple channels through which the customer interacts. Here, personalization at scale requires integrating diverse channels each potentially managed by a different owner, and connecting it together with third-party data. Then, the marketer can make insightful decisions, design the experiments, and connect it to marketing technologies for rich interactions with its customers.

In the article, the authors convincingly argue that CDPs and DMPs can help bring data together from a variety of sources and combined with machine learning models, allow marketers to tap into it to drive more effective engagement across all channels and at every interaction with a visitor, customer or partner. However, they also explain that creating a CDP and DMP is very difficult as it requires overcoming one of the biggest challenge, which is organizational. Many organizations do not give enough thought on how such systems can effectively share data across functions. Every function in an organization is set up such that the person in charge of that channel does not have any incentive or time thinking about how to share data with other channels. They end by suggesting that a cross-functional team that includes channel owners should be responsible for optimizing the customer journey across all channels.