The vast quantities of data that insurance companies accumulate can become complicated to handle, especially for organisations that have been around for more than 100 years and have made several acquisitions along the way. Zurich Insurance went through a transformation to simplify its data processes, and was recognised with a DataIQ Award, taking home the Transformation with Data prize.
Zurich Insurance Group has immense quantities of data. It has a workforce of 4,500 in the UK, operates in 170 countries, has made several acquisitions and holds transactional data on policies, policy holders, claims and brokers.
It comes as no surprise that the nearly 150-year-old company would have hundreds of legacy applications as well as different systems and different code bases. According to chief data officer, Anita Fernqvist, this meant that management information (MI) solutions were all over the place.
“You get a divergence between what is logged in the financial systems and the management information systems. It may not be an error per se, but not having absolute track of how information is derived, what logic is used and what transformation it goes through means it is very hard to reconcile the numbers between financial systems and the MI solutions,” said Fernqvist.
Essentially, there was not a ‘single version of the truth’. Fernqvist took over the data area in January 2016 after 11 years with the group and found that the MI team was working with heavy workloads, without sufficient prioritisation and in a business with insufficient consideration to data by design.
She said that there were several hundred reports going out on a monthly, quarterly and annual basis that required a lot of manual intervention and were not as detailed as the organisation needed them to be. There was also the problem of ‘key man risk’ with many old legacy systems that only some people understood.
To make this turnaround, Fernqvist led with the ethos of ‘doing the right thing in the right way for the right reason,’ and looked towards other data organisations and industry bodies for examples of best practice.
From the start, Fernqvist knew it was important to put data high on the Zurich agenda and get the leaders and executives of the organisation to buy into the importance of data, and therefore secure investment for it. She and her team did this by spending significant time with each member of the executive committee and their leadership teams to talk to them about what they were doing, why and how.
She began by focusing on people because, for her, team culture is of paramount importance. “The culture and behaviours piece plays absolutely as critical a part as the technical skill set.”
There was a small legacy team, some of which stayed and some of which moved on. She said the original team comprised MI analysts who ran regular MI reports, built new ones and consulted on programmes. In her view: “As a business we were reacting to the need for MI, rather than having the importance of data at the centre of what was being done.”
"We didn’t have a legacy of data architecture."
But Fernqvist was clear that the highly complex data landscape was caused by several factors. She said: “The reason we were in that situation was that we didn’t have a legacy of data architecture and of being a data-centric organisation.”
Therefore, strong data architecture had to be part of the solution, as did portfolio management, devops and a focus on data quality.
Fernqvist said that they had to architect systems at the outset from a data standpoint to drive value from the data in them, in addition to remediating those already live. The devops structure mean Zurich’s MI team built up its internal development capability and was able to build its own data warehouse, known as a strategic data asset, whilst working with strong third parties to upskill the team.
Portfolio management was a clear necessity because demand for reports outstripped supply and there was no mechanism for prioritising and tracking requests, and a data quality function was put in place to unearth and remediate these issues.
Now data quality is increasingly being dealt with at the source. “Often it might be a business process issues, so we’re able to work actively with the business and make sure that data in the systems at the outset is correct,” said Fernqvist. This has led to overall confidence in the numbers from the actuaries, the underwriters, the finance people, and the MI team, who can be sure they are correct and can understand how they were derived.
She also recruited “significantly different and new” roles into the team, both from within the organisation and without. She said they made sure they brought in people who had the right technical skills and the right mindset and behaviours. “So we have people who are really passionate about what they are doing and genuinely wanted to come in and help us solve the challenge,” said Fernqvist.
Recruitment involved finding the right people with the right skills, prepared to do things the right way in line with her ethos and help work with Fernqvist and the rest of the team balance that fine line between progressing at pace and maintaining quality.
They work very closely with suppliers who are seen as part of one big team. The core internal group increased from 10 to 35 over the course of the transformation, which doubles to 70 when accounting the suppliers.
Fernqvist is a big advocate of wide channels of communication amongst members of the team, including herself. These meetings are in the form of one-to-ones and regular all-hands meetings. There are also social events; some for just for Zurich employees and some for employees and Zurich partners to build camaraderie.
“We now have people excited about data and it’s only through hiring the right people that we can foster that and have a team who really get behind it,” she said.
One of the results of the transformation is greater job satisfaction. Members of the data team have moved from a focus on rudimentary, reactionary tasks like report generating and are able to retrain as data architecture specialists and data developers. This, and the team culture, has led to the team having one of the highest employee net promoter scores in the UK business.
The business is now increasingly able to have more meaningful conversations with customers and brokers because they have the data sets now to drive value. Data scientists are also now more effective in their roles, as rather than manipulating data, they can take accurate, granular data and actually drive value from it.
Fernqvist and the data team are now evangelising about data by doing a roadshow around the country. So far they have been to London, Farnborough, Swindon, Fareham, Birmingham, Manchester, Cardiff and Croydon armed with videos, exhibits and doing demonstrations of the technologies they have built.
“We want to really engage with people who already with us but also people who think that data might be complicated, highly techie or boring, and really draw them in and get them engaged in and excited by it.”