By Ajay Khanna, vice president, product marketing, Reltio
How many applications does a typical marketing department use?
With marketing tooling getting very specialized it is not unusual for companies to acquire expansive toolset including customer relationship management, marketing automation, content management, email management, advocacy management, social engagement, online engagement, marketing analytics, and the list goes on and on.
All of these applications and customer engagement across multiple channels like online, email, phone, in-store, and social are generating information at a big data scale.
The good news is that we have a ton of customer data and very sophisticated tools to manage multichannel customer engagement. But the challenge is fragmented customer information that results in marketing organizations failing to provide relevant and personalized customer experience.
One of the reasons for poor customer experience or ineffective engagement is the lack of understanding of relationships between people, product, and places. The legacy applications like CRM are not designed to manage and visualize many-to-many customer relations across all business entities spread across systems.
Without understanding customer’s relationship with other individuals, products, content, and channels, we cannot get relevant insights into customer behavior and preferences.
Graph technology in marketing
A graph is made up of nodes (real-life entities like customer or product) connected with edges (relationships). You can continuously add entities and describe relations that link those entities.
This model can represent real-world relationships much more efficiently than relational databases with rigid schemas. Using graph, you can define unlimited nodes and relationships across people, places and products.
Many leading consumer applications like LinkedIn and Facebook use graph technology to manage complex relationships. For such organizations, graph technology can make it easy to describe and visualize complex relationships in an easy-to-use interface. As such, simplicity, elegance, and power of graph technology make it an attractive choice to solve many complex data problems.
When you start connecting all business entities such as customers, products, accounts, and locations in many-to-many relationships, you start getting newer insights into your customer’s needs and preferences. You can quickly identify channels that the customers prefer, content they consume, their influence in social media, products they endorse, and their locations.
Graphs can play an important role in entity resolution as well. For example, they can help you identify if a customer engaging with you through different channels is the same person or not. Graph brings additional evidence via relations to help decide if objects maintained in different systems are principally the same object or not. Extracting marketing insights from graphs
Graphs and big data analytics technologies such as Apache Spark can make for a powerful combination. Spark algorithms help determine product scores, network strengths, and the influence of customers relevant to your business.
Data scientists use Spark algorithms to rank entities, for example, to find the most influential media personality by analyzing their publications, social media followers, mentions in the news, and speaking engagements. Once complete information is compiled in the form of a graph, a PageRank algorithm can be used to calculate the influence score.
If you need to determine top-selling products in each of your target segments, you can use Spark SQL. If you want to group your customers based on their social connections, locations, and purchased products you can use Spark GraphX library, and you can use Spark machine learning for predictive analytics to determine next best offer to a consumer.
When combined with predictive analytics and machine learning, managing customer relationships become agile and real-time. Marketers can potentially uncover new relationships quickly and can use intelligent recommendations for effective engagement with the consumer.
Consider some sample use cases:
Key influencer management: Visualizing relationships is quite important, especially in B2B account-based marketing. Marketers must understand the relationships within an account, the key influencers, their locations, and the products they are interested in. If you know who the real decision maker is and how he or she is linked to the other stakeholders, you can design a personalized campaign targeted to that individual. Marketers may be hard-pressed to extract such information from traditional CRM systems.
Complex hierarchy rollups: Understanding account hierarchies is another challenge. CRM systems cannot represent complex organization structures easily. The hierarchies available are not in the context of marketing objectives. At best you get a view of legal relationships between a holding company, business units, or departments.
The graph allows you to create personalized hierarchies with product or revenue rollups. With a personalized view of hierarchies, you can identify business units in your customer account that are using your product and ones that do not. Once you identify such gaps you can target those business units to expand your coverage.
Householding: This refers to the grouping of consumers into family units. In any given family the financial buyer, the decision maker, and the user may all be different individuals. Father may decide to buy books for his daughter on her birthday but may use mom’s credit card.
Understanding this dynamic in family units and householding is important for consumer marketers. But how can an online merchant determine that various visitors and shoppers on the website are from the same household? The graph can help. Graph clustering algorithms can be used to infer new properties or relations to link family members to one household based on entity attributes like phone, address, credit card information, or last name.
Challenges with the graph
A word of caution: Many off-the-shelf graph databases may not be able to handle the customer information at big data scale. This scalability issue limits its usage for storing and serving petabytes of data gathered from e-commerce or social networks.
This limitation is leading to the emergence of modern data management platforms built on columnar-graph hybrid stores, and other forms of polyglot persistence. Such modern data management platforms offer the additional benefit of ensuring data reliability through proper data management.
Data is blended from multiple internal, external, third-party subscriptions, and social sources and then matched, merged, and cleaned to create a reliable data foundation. Afterall, the insights are only as good as the quality of the underlying data.
Benefits can be abundant
From better segmentation to consistent multichannel engagement, graph technology can dramatically impact the performance of modern marketing. Understanding customer interaction preferences can help ensure that your communications comply with all regulations.
Quick identification of the right influencers and early adopters can also help reduce time-to-market, ensuring successful product launches. Moreover, a thorough understanding of a customer helps you better design personalized interactions with relevant information and offers, improving customer experience.
The power of graph and analytical capabilities of Spark help you run a data-driven, closed-loop marketing organization where all decisions are based on deep customer understanding and insights from all interactions are brought back into the planning cycle to improve engagement and customer experience.