At Safety Insurance, projects rarely
are conducted on an ad hoc basis.
So, as third-party data continued
to grow through areas such as
telematics and social media, the
insurer chose to make a commitment to formalize the management of the company’s data.
“Everything here is done op-
erationally,” says James Berry, vice
president of insurance operations.
“If we didn’t operationalize the use
of data we would always be chas-
ing our tail. Doing everything as a
Safety worked with a consultant (Bill Jenkins of Agile Insur-
ance Analytics) to put together a high-level roadmap on what
a data governance plan would look like. Two challenges faced
the insurer, according to Berry. The insurer’s actuarial resources
spent most of their time assembling the data before they could
perform any type of analytical effort, whether it was rate-mak-
ing or general R&D.
“If they spent 20 hours on an initiative, probably 16 of that
was spent on data prep and at best only four on the analysis,”
says Berry. “Those people have a lot of talent and leveraging
their expertise on what pieces of information are in what files
was not a good use of company resources.”
As the carrier drew data from more systems, a second chal-
lenge involved the lack of documentation on where and how the
data was assembled.
“That was inefficient and we felt led to multiple versions of
the truth,” says Berry. “We would get reports on different initia-
tives and they wouldn’t tie back. We wanted a single version of
the truth whether it was operational or analytical. By operation-
alizing the data and lending some discipline to the process we
hoped to make it more effective.”
Safety relied primarily on its policy system as the sole resource
for information, but as third-party data was introduced and
transaction processing interacted with outside sources it became
an even greater challenge, according to Berry.
Safety drew members of its actuarial group along with busi-
ness users considered data friendly to work with Jenkins and
the IT department to develop a data warehouse solution. The
group considered the alternatives and ultimately settled on the
Netezza product from IBM as the data warehouse solution.
“We ran the gamut of ideas from hosting it or with other
trusted partners. At the end of the day IBM felt like a good fit,” says
The alternatives—buy vs. build—were considered with the
data model, but the group decided to use the OMG data model.
“The disparate groups had no existing position on data standards, so it was easier to consider newness,” says Berry. “We had
never done something like this before and all the players did
a good job of collaborating and listening to what experienced
people had to say. We’re early in this, but so far so good. [The
Jenkins] group helped us figure out the road map which we
presented to all the business groups and we have good collaboration between data and IT.”
Berry explains Safety is in the foundational stage of its data
project and the biggest initial engagement is to bring together a
single view of the customer.
“We run multiple systems so it is creating some great con-
versations,” he says. “Who is the customer? It used to be a car or
an address. How do you define your parties? As we go through
those exercises to deploy that single view to the business units
we expect that will dramatically change the conversation. The
potential to price our products and to go through exercises such
as lifetime value of a customer are very exciting.”
The hope is that for the agents selling Safety Insurance
products, the data brings added value to help them with appli-
cations, according to Berry.
“Agency management systems actually do a better job of
giving a single view of the customer than what most carriers
can provide,” he says. “We will be able to do a much better job
with our interactions if our billing and claims people know a
particular customer has two cars, a home, and owns a business.”
“We expect to take advantage of that as more data comes
available on a daily basis,” he says. “Our industry is evolving and
the more ground we can cover with our analysis the better the
products we can provide and the more accurate we can process
A new data warehouse and data standards allowed Safety Insurance to opera-
tionalize its data operations.