Many carriers are worried about the amount of data that is entering their systems on a daily basis. While some worry more about unstructured data—video, texts, email and such—at least one industry observer feels it is more important for carriers to deal with the data quality issues around their structured data.
“It’s not uncommon for us to go to insurance companies and they cannot tell a person’s first name from last name,” says Samir Ahmed, senior architect, X by 2. “They have a name field that someone typed in first name, last name or last name, first name. It may be a mainframe system with two name fields.”
Ahmed realizes such situations sound comical, but he insists it is a situation that is both real and pervasive within the insurance industry.
“In our experience as a service provider to insurance companies, [carriers] are not worried yet about how to make sense of unstructured data. The truth of the matter is insurers have bigger fish to fry, so to speak.”
The ability to move content throughout the enterprise usually begins with the preparation done beforehand, explains William Montefusco, business systems delivery manager for AmeriLife.
“If you are working for a company that has a [content] management strategy, the way they house data is handled needs to be conducted in a way the business users can work with,” says Montefusco. “When there is no strategy or things come from the bottom up, there tends to be a lot of variety. In such cases you have problems because you have to adapt to a lot of forms. It also depends on whether a customer has invested in forming some form of architecture in respect to how the data is housed.”
Ahmed sees some carriers looking to modernize their policy systems in order to enforce some of their internal conventions and rules. This includes steps such as putting controls up front where customer information is routed through a CRM system rather than keying in the data directly into the policy system. Another technique involves putting data cleansing into place after the fact.
“Once the data is recorded, you do an extract of it and run it through tools that can parse out and try to make sense of the data,” he says.
Carriers then can perform a statistical analysis on the data they have.
“Say they have 80 percent confidence that the name is correct the way it is recorded,” says Ahmed. “When the tool says it is 80 percent confident that might be good enough. There are also tools that say they are 80 percent confident that it is incorrect and that kicks out the name for someone to review and fix manually.”
Ahmed believes the starting point on their confidence in the data’s accuracy actually is fairly high.
“You could say you would only accept a positive result if the tool gives back a confidence level of 95 percent,” he says. “That creates two standard deviations of confidence. You start [at 95 percent], but that means the number of exceptions is going to be high because you are rejecting things that actually are correct. Then you start looking at training the tools, improving the processes, and then you start ratcheting it down, slowly but surely. We’ve seen cases where insurers are running into fewer problems because of these tools.”
The social media content integration that Ahmed sees among insurers is purely at the Web content level.
“We haven’t seen it at the full enterprise content scale, primarily because enterprise content includes things like internal documents, manuals, reports, and workflow cycles for how to create them,” says Ahmed. “Social media doesn’t really play a role in any of that.”
What Ahmed has seen is carriers interacting with their customers so they can tailor their content to known customers more than to the anonymous public browser.
“We’ve seen [social media] definitely complicate things because a lot of the traditional content management systems don’t have social media capabilities. [Carriers] have to look at how to integrate that content. Do they upgrade to newer versions that now include social media modules?”
Celent senior analyst Mike Fitzgerald believes insurers need to be careful in dealing with the information spread through social media.
“That’s just starting to evolve,” he says. “A company should not turn [positive comments] into an advertisement. That can turn on you very quickly.”
The best strategy that Fitzgerald has seen with consumer-based products is to identify influential people who are current customers and have had positive interactions with your company.
“Offer them some additional assistance and make sure they know what is going on and see if you can add them to a special network and then see where that goes,” he says. “You want to cultivate [influential customers], but not so it looks unseemly. You need to see who is influential as a satisfied customer.”
Carriers also need to identify when positive information is posted and to leverage that content in a way to gain some visibility, according to Fitzgerald. That includes determining how social media data can be used to help risk selection and benefit people who may display positive risk characteristics, which traditional underwriting hasn’t discovered.
“On the life side—which appears to be extremely active—there are companies doing innovative things around risk control that the underwriter wasn’t aware of,” says Fitzgerald. “The ability to use that information for better underwriting is still to come, but the information is out there. It’s a question of whether companies want to access that and how quickly they can monitor it and eventually take advantage of it.”
Another area that makes the picture more complex is content on carrier Websites, which is increasingly moving in the direction of being user generated.
“People are providing reviews and comments and that is incorporated in the actual write-up,” says Ahmed. “For instance, if a carrier has a page about what type of homeowners’ products they are offering, they may want to link it with some basic home information and provide tips from other customers. These are types of things that customers are initiating and they get added to the write-ups. At some point, people like these tips enough that they are incorporated in the main article.”
Ahmed believes insurers generally are taking it slow when it comes to integrating social media content.
“There are not many doing a full integration of user-generated content into their main articles,” he says. “What is more common is the ability to comment on articles, to “like” something, and tweet a link to something. We’ve seen a few that did go that route and look at positive customer reaction and incorporate that content.”
Where that impacts workflow for carriers is they have to incorporate the monitoring of comments and tweets in their workflow and then they have to extract the content from the comments, incorporate it into the main article, and publish it. There’s time and effort going toward that.”
Ahmed doesn’t feel those not integrating user-generated content are seeing an impact in their content workflow.
“What they have generally done is create rules for social media marketing,” says Ahmed. “They use different terms for the position and they rely on the typical social media tools rather than content management tools to monitor what people are saying and writing.”
From a p&c perspective, social media has become useful because in managing content, according to Montefusco.
“Something as simple as Google Earth can give you a feel for what’s going on,” he says.
Still, Montefusco believes integrating social media into content management is not being done at many insurance companies.
“We utilize social media in an industry perspective, but we haven’t set up anything that integrates it,” he says. “If we smell something fishy or it seems odd, we’ll use that as a means of investigation, but we haven’t made it a standard yet as a catch-all. We’ve created relationships with third-party companies that run credit reports or do background checks. We provide them information, grab the results, and pull it into our environment. With respect to Facebook or Twitter, I’m not sure how we would do something like that yet.”
It is particularly important to monitor social media content for insurers operating in the highly-regulated lines that involve some sort of regulated securities products, points out Fitzgerald.
“Any of the comments made around a product are considered advertisement and fall under the same auspices as other advertisement and has to be made by a person licensed by FINRA,” he says. “Those rules impact the content and what actually goes out. The difficult thing for insurers is to make sure either independent brokers or captive brokers are aware of the regulation and don’t make a product endorsement unless it is controlled by the same rules of standard advertising. It’s a content issue as well as a management issue and being able to manage what goes out, even if it’s not under their direct control.”
The monitoring of social media content already is being used in claims on the p&c side as a standard practice for investigations.
“Claims adjusters say [a social media site] is the first place they go when an SIU gets a file,” says Fitzgerald. “The officer will go to Facebook and Twitter and look for the same or general names in the right geographical area.”
The problems claims people face is the search technology isn’t as strong as investigators would like.
“It’s almost all manual searches,” says Fitzgerald. “It’s proven its value and admissibility in court tests. Courts look at this as discoverable evidence if it is treated appropriately. Do you try to mechanize it and improve its usefulness and efficiency in which the data is brought in? That’s still under development.”
Kay Haupert is applications director in IT for Sentry Insurance and one of her responsibilities involves working with the carrier’s electronic publishing group. She helps build all the publishing systems and form templates.
“They are responsible for getting data and templates merged and distributed to customers through a variety of channels,” she says.
Haupert also works in the operations area where Sentry has two print-to-mail sites, which conduct the heavy printing and inserting.
“What I provide is more from a publishing perspective on content management,” she says.
Sentry is focusing on going paperless so the content that is received in various formats—unstructured objects—can be stored in the carrier’s archives.
“Some of our business areas are trying to go paperless so the objects can be stored and retrieved without having to manage paper files from desk-to-desk,” she says. “The claims area was one of the first onboard with the paperless environment. Even if there is printed material, we are scanning it and putting it in an archive so [business users] can retrieve it without touching any paper. There certainly is a push for that, but I see a lot of areas that still use paper and have to manage a lot of content.”
Publishing for Sentry means creating documents—although Haupert points out that doesn’t necessarily mean printing on paper.
“They become an object that can be distributed in some fashion,” she says. “When we say publishing we are getting a piece of data or a request for a transaction from one of our business applications.”
That might be a single transaction or the department might send the publishing group something in batch, explains Haupert.
“With a single transaction we would get a request from a business system and in that request it would give us enough information that we know what type of form to publish in that package and they provide all the data,” she says. “Our job is to take that data, marry it with forms, and based on what the users ask for, the publishing group distributes the content.
That means a variety of options—email, fax, print it, a PDF or just store it in the archives.
“Once we get the request, where it goes and when we reply depends on what the business unit requests,” says Haupert.
The Oracle system Sentry uses is Documaker, which is Sentry’s publishing engine. The carrier builds templates with the software and then uses the publishing engine to build and publish the template.
“[Oracle] has provided some other pieces around [Documaker] that allow us to accept the transactions, process them, and publish them using Documaker and post-processing—the actual distribution of the content,” says Haupert.
Sentry has had several software packages for the publishing process—one for batch print, and another that did user interaction.
“What we are able to expand on now is to go beyond batch,” says Haupert. “For our point of sale documents, we can actually create the documents and return a PDF immediately rather than waiting for a batch process,” says Haupert. “There were ways to do that before, but it was convoluted. This has opened that avenue as well as other distribution. With Documaker we are able to publish quickly and when they want it distributed. They can tell us in the transaction and we can react to it. Before it was either you get batch or real time, but you can’t tell us at the transaction time what you want.”
Montefusco has worked for multiple insurance carriers and thus has worked with three different content management solutions in his career.
AmeriLife uses the ImageNow solution from Perceptive Software and has seen an increase in work productivity among the business users with the product.
“It typically took 40 minutes to process an application through its life cycle, but we’ve been able to cut that to 32 minutes,” says Montefusco. “That’s a significant savings. We use that savings to support attrition and we can grow our business without increasing staff.”
Flexibility is a key for any content management system, according to Montefusco, particularly for carriers operating under the burden of legacy mainframes.
“The majority of our legacy systems are post-application,” he says. “In the insurance industry there is a lot of green screen. We use ImageNow with accounting and other areas. In our environment, we use ImageNow with CRM, Excel, some Web applications, and a proprietary application. It’s extremely adaptable and can handle working with a multitude of systems. Some companies have more programming involved, but most can support multiple legacy systems.”
Ahmed doesn’t see a carrier’s legacy system being an impediment to the quality of content for carriers, at least on the outward facing side. With inward facing content, it is a bit of an issue.
One element of ECM that is different from Web content involves documents that are relevant to the customer—something they can see on the portal.
“Not all systems have the ability to provide a Web consumable document,” says Ahmed. “There are systems that produce print screens that a printer can understand and you have to put conversions on top of that to convert the print screen into a PDF file and then find a new repository for the documents to make them available for customers and agents. In that sense the legacy system is definitely impeding that aspect, but in the Web content aspect, there’s not as much dependency on other backend systems.”