"Carriers often look to automate flow when they are leaving quotes on the table because their manual processes do not allow them to scale up for increased quote volume," Gorman says.

Most carriers that have enabled real-time quote flow also have taken the process to the next logical step: allowing the agent to accept the quote once received and provide confirmation, such as a policy number, that the policy will be issued. However, only a small minority of companies that have automated commercial lines quote and application flow have taken the process one step further to make actual policy delivery a near-real-time, one-visit process–less than 10 percent, estimates Stephen Forte, principal research analyst in the P&C sector at Gartner.

One company in that minority is ACUITY. In its flow solution, called ACUITY ASIST, the carrier accepts applications electronically via its own agent portal and through bridges with various agency management systems on BOP policies (as well as all personal lines), connecting to an internally developed rating and policy administration system on the back end. ASIST incorporates ACUITY's rules-based underwriting system, based on Blaze Advisor from Fair Isaac.

Available for all lines of BOP since last October, today more than 40 percent of bound applications pass through ASIST with no manual intervention. As soon as the agent submits an application, it is sent to the policy administration system, which produces a PDF of the complete policy within about 30 seconds. A link to the PDF is e-mailed to the agent as soon as the PDF is available, giving the agent the option to print the policy and hand it to a new customer at the point of sale.

"Because a PDF can be quite large, we prefer to use e-mail link rather than automatically displaying the actual document in the agent's window," explains Marcus Knuth, director of information systems at ACUITY. "We give agents an underwriting decision instantly, but even though the policy generation process takes less than a minute, it's longer than you want to have an agent wait for a response after clicking 'submit.'"

The print-on-demand option currently is used more heavily in personal lines than in commercial lines, where most of ACUITY's agents still choose to have the carrier mail a paper policy to customers. However, the insurer sees on-demand, point-of-sale policy print increasing. "It will be a larger advantage in the future as people's expectations for the insurance transaction change," says Neal Ruffalo, ACUITY's vice president of enterprise technology. "Already many of our agents have turned off policy print for their agency policy copy because those records are available online 24/7."

IF YOU BUILD IT . . .

Because the penetration of commercial lines into comparative raters and agency management system bridges is nowhere as great as personal lines, the question arises whether time-strapped producers are willing to use proprietary front ends and essentially become the carrier's data-entry staff. The success of an automated flow strategy therefore starts with a front end that is intuitive and easy to use.

Currently, The Hanover Insurance Group's flow system handles BOP, auto, umbrella, and workers' compensation commercial lines. Having already achieved agent utilization rates of its portal for 90 percent of BOP and about 70 percent of other commercial lines accounts, the insurer targeted the site's appeal and flow as it planned to expand automation into larger commercial package policies.

"We're constantly refining the question flow to address the big qualifiers upfront so that our agent partners get a quicker 'yes or no' answer on basic eligibility," explains Jay Coon, assistant vice president of small commercial at The Hanover.

In May 2007, The Hanover completed a revamping of its point-of-sale system that focused on easier navigation and restructured the system to follow an account view rather than policy view. For instance, the system now identifies data elements on an account that are common across all policies a customer might have (name, address, etc.) and passes that information to subsequent applications in the package to eliminate rekeying.

"We tried to mirror the experience agents or CSRs are used to from within a typical agency management system, where there is an account [structure] that provides a consolidated view of information and allows them to drill down into account information, messaging, and other activities," notes Coon.

The company also is piloting a new front-end system to a group of agents for commercial package policies. What The Hanover found in creating the pilot is flow systems for package policies create usability challenges that increase proportionately to the complexity of accounts.

"The package policy is more intense in terms of data input. Some agents are naturally willing to do the extra work required because they value the ability to provide more immediate feedback to their customers, but some are reluctant and instead continue to rely on the traditional paper-based ACORD application process," states Scott Shader, the insurer's chief operations officer for commercial lines. "That is why we developed a tool that is easy to navigate and doesn't require much of a learning curve for CSRs new to the system."

To achieve and improve flow automation in commercial lines, carriers also have applied the lessons learned in personal lines. "The key difference [in commercial] is the amount and variability of data," says ACUITY's Knuth. "It is a lot of work to codify the underwriting decision and determine how we want to handle a particular type of situation every time without saying 'it depends,' because when you say 'it depends,' you have to find a way to define and capture the data for what it depends on."

This challenge is exacerbated on larger accounts. "The consistency of the data is a huge challenge" to automating the flow of those accounts, Gorman points out. "If we look at personal lines auto, the data is very consistent from one application to another. Larger commercial risks can be totally unique."

Safeco, which had deployed an automated flow system for property, general liability, commercial auto, excess liability, and workers' compensation, addressed that challenge when it added commercial package policies for midsize accounts in 2005.

"The challenge is to create enough rules to handle a diverse range of accounts but not to create too many so that the system is no longer usable. Also, you can't get caught up spending too much time trying to automate a risk fully you'll see only once or twice a year," says Kirt Lenard, assistant vice president of product modeling at Safeco.

Front-end data collection for the flow system starts at the company's Web portal, Safeco Now. Applications pass to a custom-built rules engine to determine whether the risk matches Safeco's market appetite and is properly classified. Additional data from third-party sources, such as motor vehicle records and Dun & Bradstreet ratings, are assessed by the rules engine, as well.

Then Safeco's flow system invokes a predictive modeling engine to perform a multivariate analysis and determine pricing based on forecast experience. "That's where the complexity really comes into play on larger accounts," Lenard says. "When you have diverse exposures, you have fewer data points from which to model."

While the surefire way to solve that problem would be to ask more questions to collect data, doing so would impact both the design and usability of the system. "We do ask [application] questions differently from what the standard ACORD forms do that are specific to particular types of business, but we also continually look at our question set to determine whether we're asking high-value questions," Lenard says.

If a risk delivers enough data, agents are presented a firm quote and can choose to bind the policy and obtain a policy number. Other applications may generate a tentative quote that is displayed to the agent but subject to underwriting approval, while others create an advisory price range that is provided to underwriting but not the agent.

"For those risks [subject to approval], we stake out a position and indicate a price band; it's then up to the agent to provide additional risk information to the underwriter and work to arrive at the correct price," Lenard says. Agents can decide whether or not to send risks that need referral to their underwriter or cancel the quote transaction.

Important to Safeco has been the knowledge gained from collecting and analyzing data over time for the multivariate modeling process. "The insight the process gives us allows us to broaden our appetite for risks we might not otherwise have written simply because we hadn't studied them enough before. Now, we can charge an appropriate price and be a market for more types of business," adds Lenard.

A BETTER WORKSTATION

On the transactional side, the focus today is on putting more information at the fingertips of underwriters through workstations, using rules engines and other subsystems of flow technologies to automate report ordering and administrative tasks, and leveraging decision automation technology for decision support on complex risks.

"We received a lot of inquiries, starting in the latter part of 2006 and continuing into 2007, about underwriting workstations," says Gartner's Forte. "There are a lot of manual tasks that can be automated so that underwriters can focus on underwriting and not get bogged down in collecting information. The ultimate goal is to gain efficiency due to better routing, workflow changes, assigning work to the right underwriter, as well as by automating information gathering."

Companies typically target at least a 25 percent increase in revenue per underwriter due to these types of initiatives, according to Forte. "There are different technology approaches companies are using for this, whether buying a stand-alone underwriting system, building one, or connecting a bunch of different systems with a BPM [business process management] framework," he says.

BPM-based systems also provide control and visibility into the underwriting process. "Sometimes underwriters do things they don't have the authority to do, but without a workflow system you don't have the ability to detect that until after the policy is issued," Forte says, adding these systems also can guide the decision-making process. "They can tell underwriters whether they have enough information to evaluate the risk."

For its underwriting workstation, The Hanover uses a BPM system from Accenture, which Shader says was "heavily customized" by the company's internal development team. With the system in place since 2002, the company has worked to streamline the transactional process to move accounts needing underwriting review back quickly into the flow system.

"For agencies writing small commercial lines, they want to get the answer as quickly as possible. Our goal is to create a pass-through experience even for business that kicks out" from the flow system, Shader continues. "We're constantly looking at every 'stop' in the system from the front to back end, whether the stop involves people, process, or technology, and what we need to do to make it a better experience."

Since deploying the workstation, The Hanover has seen its time to process transactional BOP business trimmed to just one-fifth its previous requirement. "It's been a major savings," Coon attests. "There is a consistency benefit, as well–the agents know what to expect and they get into a regular mode and method of doing business with us."

Part of The Hanover's workstation is an "underwriter action report" that serves up a snapshot view of underwriting information on quotes as well as a price bracket based on 56 underwriting variables, which are presented to underwriters for their review. "We don't let the machine make the pricing decision; we want to provide underwriters only guidance in their decisions," says Coon.

Creating workstations and workflow systems that guide and complement, rather than restrict, the way underwriters process business is a challenge. Additionally, as the size of an account increases, so does the volume of information that is typically available and that needs to be digested by the underwriter.

"Historically, underwriting data was captured on paper and the analysis was done on paper, and the ability to do more of a portfolio analysis of the information was inhibited," says Gorman. Simply imaging those reams of paper doesn't necessarily make them any easier to digest. Therefore, the current edge of leading development in underwriting support technology is extracting structured data from the bevy of unstructured documents that comprise an account. That can include evolving to dynamic PDFs, applying richer metadata to photos and scanned documents, or applying text mining to documents.

"That technology has advanced significantly lately. We're seeing new players emerge in this space that have the technology in both workflow and rules engines to capture what historically has been unstructured data and make it easier to analyze," Gorman remarks. "As they capture [application] data electronically and then drive toward derived data from unstructured content, the opportunities for the flow of business are enhanced."

Another challenge in deploying new underwriting technology is cultural. "If underwriters have been doing business a certain way for years, introducing a Web-based system when they are used to green screens is a cultural shock," Forte says. Carriers need to strike a balance that delivers consistency and control without over-regimenting the process on accounts that can't be automatically passed through.

"Change management is an issue," Gorman agrees. "We often have found the business process issues are bigger [problems] than integration of new technology."

"Since the system just serves up data to underwriters and makes a suggestion, and since underwriters still make the decisions, acceptance was quicker than it would have been had we tried to control the decision," Coon maintains. "Although it still took a while to get everyone used to the workflow, if we pulled [the system] out now, there would be mutiny."

Safeco's system has "changed the underwriting culture," Lenard asserts. "It has put more discipline into the culture, and we've spent a lot of time showing people the benefits of giving them the decision support to do a better job and be more productive. Today, underwriters like having a consistent, factual basis [by which] to talk to their agents."

Whether they like it or not, underwriters will have to learn to adapt to new flow and transaction technology as carriers continue to pursue solutions to increase efficiency and market share. "Agents will send business to companies whose processes are more streamlined," Forte contends.

However, unlike in personal lines, where companies have exceeded 90 percent pass-through rates and the role of personal lines underwriter often has evolved into agency relationship manager, commercial lines likely will retain its high-touch, transactional business for the foreseeable future.

"In large commercial, you'll never get to straight-through processing because there are too many variables," Forte claims.

"The areas of gain [through automation] in commercial lines are in quoting precision and predictability and increased capacity and scale–being able to focus personnel on more difficult quotes," Gorman concludes. "Companies making the investment [in underwriting technology] see that as an opportunity, an area where whether you're a national carrier or a midmarket company, you still can compete equally and there's still chance to be a market leader."