Founded as a family business in 1935, Harel Insurance Investments & Financial Services has become one of Israel’s largest insurance groups. Their comprehensive product portfolio includes health, life, business, and motor insurance as well as investment services such as pension and mutual funds. Despite the fact Harel is now a publicly traded company with more than 5,000 employees, they retain their traditional family values including an emphasis on personalized service and reliability. Today, they’re delivering on these values with the help of the Elastic Stack and new insights into customer needs.
In 2015, Harel announced a new strategy to "recalculate routes" and transform for the digital era. Seeking to provide customers with simpler, faster, and more personal service, their strategy involved digitizing manual processes like settling claims to improve the speed of service as well as expanding their use of analytics.
The Elastic Stack has supported the delivery of Harel's strategy by helping the company extract and visualize the data needed to optimize digital processes and improve the customer experience. What’s more, it has helped Harel turn data into a competitive advantage without the need for costly and complex data projects.
Harel’s journey with Elastic began with performing log analytics on a single server. Their focus then shifted to analyzing events within their infrastructure operations, as Elastic replaced querying and sweeping thousands of machines using scripts.
However, Eyal Efrat, CIO at Harel, saw the potential to do much more with the Elastic Stack and has championed its use across the business.
"Elastic is one of the key building blocks in our digital and operational landscape. It allows us to constantly improve our services," said Efrat.
As Harel transformed for the future and began to digitize more services, the company needed to ensure the availability and performance of their infrastructure. In particular, Harel needed a quicker way to analyze IT events from many endpoints so they could get to the root cause of issues and address them right away. At the time, the infrastructure team would need to create scripts and harvest logs manually every time they wanted to explore a particular set of data. It could take days to resolve a single query.
Seeking a solution to these challenges, the company looked at various logging tools. However, they found many too expensive. Elastic offered the most value with its advanced security features, enterprise grade support, and ability to provide real-time business analytics rather than just IT operational metrics. It is also a fundamental part of the company’s culture to adopt open source solutions.
"Elastic stood out to us as a very easy, but powerful product to use. We wouldn’t need to hire a bunch of data engineers or build huge data lakes to make it work. Instead, we could give everyone the ability to search and visualize data themselves for insights to drive our digital strategy," said Raz.
Harel began using the Elastic Stack for monitoring and analyzing logs and, almost immediately, they knew that they had made the right choice. The Elastic Stack made it easy to extract data from multiple sources, providing real-time business analytics as well as insights that could be used to optimize IT. It also eliminated the hard and complicated work required to conduct queries for even the most basic analysis.
Harel signed up for a Platinum subscription with Elastic after only two weeks. They wanted to extend the use of Elasticsearch and Kibana right away to analyze the performance of their digital services. They also wanted to take advantage of Elastic security features such as role-based access control.
Fast forward to today. Harel’s digital service and infrastructure data are indexed to Elasticsearch and can be visualized in a way that helps the company improve service. And while their 150+ users are largely self-sufficient, Elastic support remains important and helps Harel to stay on top of new features and fixes.
Harel’s use of the Elastic Stack has evolved in phases. They started with ingesting infrastructure operations data into Logstash and then began using Beats to ship more and more of their data into Elasticsearch. Not long after, Harel connected the Elastic Stack directly to their relational databases in SQL and to every new digital service. Now, Harel has metrics on everything from product sales through to details on how customers are interacting with their products and services online.
The company has 420 dashboards and 3,108 data visualizations in Kibana — all of which helps them track the customer journey and identify trends. For instance, they can receive insight on when their 3 million customers are most likely to file claims and ensure employees are available at that time to support them. They can also analyze the completion of online forms and applications to understand how long these processes take.
“The best way for us to improve our processes is by analyzing them in Kibana. We can monitor things like the completion of online forms and applications to understand how long they are taking. Also, if we see process inefficiency, we can look at the data to understand why and then either adjust the form or reach out to customers to help them,” said Moshe Alaluf, digital product manager.
The use of Kibana Spaces helps to organize all of Harel’s dashboards and visualizations by relevant workspace and provide secure access to those who need it.
Harel began their journey using version 5.1 of the Elastic Stack, and they have upgraded to each new version as it has become available. They are now running the latest version of the Elastic Stack across nine servers and have indexed 25 billion documents equating to 16TB of data.
“We made the decision early on to upgrade with every version of the Elastic Stack and we have benefited immensely from immediate access to new features and greater stability,” said Raz. “In addition to these upgrades, we do a review every few months to ensure our architecture is aligned with best practices.”
Harel also makes very careful decisions about what data to ingest and how to best ship it to Elasticsearch and visualize it in Kibana. Harel decided that data being shipped must go through a grid of persistent queues on Logstash dedicated servers. This helps to optimize the performance of their cluster so that important data can be searched and analyzed quickly.
Harel is continuing their digital transformation and using Elastic to ensure the efficiency of new services. They are continuing to improve business activities to better suit customer needs. This includes detecting inefficient processes, churn rates, and how to create more customer-centric services. “Harel Insurance is going through a digital transformation as part of our customer experience evolution, focusing on efficiency.” said Efrat.
Moving forward, Harel plans to use alerting from Elastic to support faster resolution of any digital process or IT issues. Specifically, the company wants to use alerting to automatically detect changes in the baseline usage of Harel’s web services, and immediately prompt the infrastructure team to investigate the underlying cause. Raz anticipates this will help them to be much more proactive in resolving issues and make a huge difference to service.
The most valuable aspect of Elastic for us is that we don’t need data engineers to extract gold from the mountains of data we have. Instead, we can give everyone data analysis capabilities with just a few hours of training and then use that to benefit our customers.
We expect Kibana to offer additional functionalities like more visualization options and more analytics capabilities that will allow us to get even faster results.
- Number of kibana dashboards404
- Number of kibana spaces2
- Number of Clusters1
- Number of Nodes:9
- Number of LS instances4
- Number of LS Beats~2500
- Total Number of Documents21,158,338,893
- Total Data Size14.5TB
- Daily Ingest Rate5100/s
- Number of Indexes111
- Query Rate55/s
- Time-based indexes (e.g. one index per [hour, week, month])all
- Node Specifications: Memory (Total, CPU, Disk-type (SSD, HDD)memory: 64g/node
Disk: 20T SSD