An independent insurance agent can be a carrier’s best friend. But sometimes that same agent can also be a thorn in its side. In some markets, agents can discount the rate from the technical price developed by the insurer. This blog addresses those situations and the need for insurers to use predictive analytics to guide them in controlling their agents’ behavior.

Take the efforts on the part of many auto insurers to optimize their renewal pricing. The insurer may want to take advantage of the latest advances in prescriptive analytics to set rates, but how will the company’s agents respond? The insurer’s agents may feel that by driving away some customers with rate hikes they will lose commissions, so they continue to offer discounted renewal rates to all their customers—regardless of their driving records. That has led some carriers to try and predict their agents’ behavior and monitor the discounts that they offer. But this type of micromanagement is counter-productive for most insurers; it’s awkward to manage and runs the risk of alienating the agents—who might respond by shifting their portfolios to another carrier. A better, more cooperative approach is for the insurer to demonstrate how predictive analytics and dynamic pricing can benefit its agents’ business.

The question you should be asking yourself is where you are on the agent/insurer relationship spectrum and how can analytics improve your agent relationship and ultimately your bottom line? Let’s take a look at this spectrum in more detail:

  1. Guessing what actions your agents will take
    There is a general assumption that agents reduce prices up to the level of last year’s paid premium, or very close to the strongest competitor offer in case of new business. So insurers assume that agents will give discounts based on this method.
    To be able to control the level of discounts that agents give, some insurers implement reward programs for agents based on loss ratio and retention metrics, while others have a budget of discounts per policy/per agent, but this needs to be monitored on a regular basis. These approaches are inefficient, as they assume that all agents and customers behave in the same way and do not take into consideration the unique needs of each segment. So the next level of sophistication is to predict agents’ behavior.
  2. Anticipating agent actions
    Using analytics to analyze agents’ previous actions can help insurers to anticipate how agents will react to a change in pricing strategy. This can help the insurer put the pricing initiative on a more effective/efficient footing and seek out their agents input on what actions to take. This approach can be effective in the case of tied agents, who can be the insurer’s employees or at least resemble that relationship. In the case of non-tied agents, there is always a risk in changing a pricing strategy, as agents could take their portfolio to another company. So a better approach is to cooperate with the agents.
  3. Cooperating with agents: predictive analytics
    Predictive and advanced pricing analytics is not a zero-sum game for insurers and their agents. When they share their ideas and cooperate, both parties can get maximum value from their customer relationships. In practice, this means demonstrating how pricing based on risk levels can help the insurer’s agents retain their most profitable customers and attract new ones—without cannibalizing their commissions or the insurer’s profits. The carrier should consult with its agents, listen to their concerns and suggestions—and then act on some of their ideas. For most insurers it’s their agents, after all, who are most familiar with their customers.
  4. Instructing agents: prescriptive analytics
    Quite recently, a large European carrier took this approach when familiarizing its agents with the Earnix pricing analytics software. While the carrier wanted to improve its loss ratio, it also wanted to maintain market share and ensure that any pricing adjustments wouldn’t increase churn.. To achieve this balance, the carrier initially proposed curtailing its agents’ ability to offer discounts and relying entirely on the Earnix software to set premium levels.

The insurer’s agents responded that until they gained more familiarity with the system and had a chance to see how it dealt with customer needs, they should still have the latitude to offer discounts to policyholders that they knew to be high-value customers. Recognizing the wisdom in this, the carrier agreed that moving forward the analytics platform would determine rates for 40 percent of its policyholders, while the agents would still set prices for the other 60 percent. In addition, the preferred strategy was to maximize retention instead of other KPI’s, so that by having more customers, the total agent commission increased as well.

The insurer’s decision was in line with the results of a recent Earnix customer survey where it was reported that carriers and their agents that use Earnix software have seen their customer volume grow three percent on average without undercutting their profitability. Furthermore, by using Earnix machine learning classification models, the company identified agents who had a high ability of converting and retaining customers. By implementing a prescriptive algorithm that identifies the best pricing for each segment, the company managed to retain more customers without increasing the loss ratio.

It is my belief that working hand-in-hand with agents brings optimal results. This can only work in situations where trust has been built up between the agent and insurer to the point where the insurer can reach the ultimate goal of being in the position to instruct the agent on pricing without having any resistance from the agent because goals are aligned for both parties.