Are Data Relationships the Key to Love (and Business) Connections?
Online dating firms excel at managing data at scale and in real time, and their secret is a technology you need to have in your armory.
- By Emil Eifrem
- May 27, 2016
The Pew Research Centre reports that 15 percent of American citizens have used online dating sites and/or applications, up from 11 percent in 2013. Meanwhile, 27 percent of 18- to 24-year-olds have used such services, triple the level of just two years ago.
Online dating is a huge business success story. It's interesting to reflect on why. Does it reflect an increased need for dating help and resources? Not necessarily. There have been "Lonely Hearts" ads for more than 100 years -- and there's always been a role for a matchmaker.
Online dating's formula for success is applying data science to these mechanisms at a massive scale.
As the volume, velocity, and variety of data increases, the network of relationships we need to plug together is growing faster than ever. What we need in order to connect one set of daters with matches -- or customers with products -- is a data architecture that can excel at handling these large volumes of unstructured or highly connected data in real time.
Connections and Relationships
The database technology in question is NoSQL technology, particularly graph databases. Originally developed in-house by Google, PayPal, and LinkedIn back in the 1990s to handle data relationships for search algorithms, online payment networks, and social networks, graphs played a huge role in building the momentum of those giants, so it's no wonder that matchmaking sites are using graph databases as well.
The technology has moved into the mainstream, so you no longer have to be a big consumer brand to take advantage of it. There are commercial and open source variants available for all budgets.
As a result, more players are starting to emulate the Web giants and the dating sites by using graph technology to manipulate large sets of connected data to great effect. Forrester Research estimates that in a year's time, one in four enterprises will use graph database technology. Gartner reports graphs are the fastest-growing category in database management systems, predicting that 70 percent of leading companies will pilot a significant graph database project by 2018.
Why are graphs so great at working this way? Because graph database software gives equal prominence to storing data (customers, products) and to storing the relationships between data points -- such as who purchased what, who likes whom, which purchase took place first, and other useful data for a brand owner.
In contrast to traditional (relational, SQL) business databases, we don't have to live with a semantically limited data model and expensive, unpredictable joins. Graph databases can support many named, directed relationships between entities or nodes, for which they give a rich context to the data. Organizations using graphs can therefore learn a lot about a customer by mining these data relationships. Even better, the queries you want to run against all this rich data are going to be instant because there is no join run-time penalty.
Data Matching Performed Faster than the Human Eye Can See
The ability to map relationships with such facility results in hugely powerful and comprehensive profiles of your customers (or your potential new partner, in the case of online dating). This puts you in the position to match prospective customers with the products or services most likely to appeal to them in ever more tailored and immediate ways.
This is what the best dating sites do in exemplary fashion when they suggest a potential partner, and it's what businesses in all markets need to be doing to maximize value and increase up-sell opportunities. After all, offering the digital consumer of 2016 a promotion that's not tailored to their requirements is just not going to improve your bottom line.
Increasingly, brands need to abandon any general, one-size-fits-all approach and instead examine the customer profile and history, quickly query this data, and relate the customer to the people who are the closest match to them, both in their social network and in buying patterns -- in milliseconds.
Making suggestions as sharp and relevant as possible also requires the ability to instantly capture any new interests shown in the customer's current online visit. Again, graph databases are a great enabler, because they can effortlessly match historical data with live session data.
Retail Leading the Way
Many major U.S. retailers are already leveraging the power of the graph to drive sales. Walmart uses a graph database to combine information from customer purchases at physical and online stores in order to make real-time personalized recommendations. Global sports giant Adidas Group is using a graph database to offer enhanced features (such as product and content recommendations) to website visitors.
Whether it's directing a customer to their favorite brand and color of sneaker or making connections within a growing digital consumer dataset, NoSQL-based graphs are enabling retailers to deliver the super-focused, hyper-relevant features we've been discussing.
Graph databases offer a new way of connecting with customers on a rich, personalized level. It's time to make data-based "dating" your business, too -- whatever your actual sector -- so you can bring customer and product/service together quicker and make an ideal match.
Emil Eifrem is CEO and co-founder of Neo Technology, the company behind Neo4j, a leading graph database.
Previously CTO of Sweden’s Windh AB, where he headed the development of highly complex information architectures for enterprise content management systems, Emil famously sketched out what today is known as the property graph model on a flight to Mumbai in 2000. Since then, Emil has devoted his professional life to building and evangelizing graph databases. He is a frequent conference speaker and a well-known author and blogger about NoSQL and graph databases as well as co-author of O'reilly's Graph Databases.