Client Allstate Insurance Corporation
An insurance company that offers a full range of products, including protection for vehicles
My Role Lead researcher (contract) — planned and conducted all user research
Team Collaborated with a UX Designer
Project Timeline 3 Months
Deliverables Research report with recommendations, Appendix: techniques for influencing expectation and perception (cognitive biases and heuristics), Video highlight reels with multiple insights tagged per clip
When an insured vehicle is damaged so severely that the cost of repair is greater than the vehicle’s value, it is considered a “total loss”, and customers are offered a cash settlement instead of having their vehicle repaired. These customers are often dissatisfied with the amount of money offered for their unrepairable vehicle.
This is a significant issue for Allstate:
- 20% of all auto claims are a total loss
- 8% of customers reject the cash settlement value.
- In total, 1 in 4 customers leave Allstate within 18 months of filing a total loss claim.
There is an opportunity to improve the customer experience and explore new and better ways to communicate the value of their vehicle in order to retain more customers in the future.
- Learn more about how customers think about their vehicle’s value and determine how much they think their car is worth
- Determine how and when to best communicate a vehicle’s value with customers
- Identify opportunities to increase customer satisfaction by improving the total loss process as a whole
- Increase for customers
- the percentage of customers who accept the initial cash settlement offered
- trust that an offer is fair
- satisfaction of the total loss experience
2. Decrease for customers
- surprise at the amount of money Allstate offers
- confusion of how valuation is determined
- instances of contesting the amount of money offered and further negotiation beyond the initial offer
PROCESS AND APPROACH
I met with the Claims eCommerce Program Manager, Automation & AI Strategy Lead, and other members of their teams to share our hypothesis and assumptions, as well as risks for the project.
Our first meetings took the form of a project-canvas workshop, where all individuals stated their goals, hypothesis, and potential risks separately, then we prioritized and organized our thoughts as a group.
Our team was conflicted about how to best address these problems. The main disagreement was whether or not notifying customers about depreciation would be beneficial, harmful, or have no effect at all. This would be a main area to focus our research on.
- By providing customers the ability to input condition details about their vehicle when getting insurance and update those details upon renewal: valuation calculations can be calculated more quickly and customers can get paid faster.
- By notifying customers of their vehicle’s depreciating value over time, customers will be less surprised and disappointed in the event a total loss ever does occur.
- Informing customers that their vehicle’s were depreciating at various points during their insurance policy may cause unnecessary points of contention with little to no upside
- Customers aren’t likely to remember or care about past vehicle value estimates if their vehicle is ever damaged, they care about what the actual valuation offer is in the present.
- Many customers are inherently distrustful of their insurance company and don’t expect fair valuations
- The stakeholders would not be available much during this project, and didn’t know if they could participate much outside of responding to project updates and milestones.
- It would be difficult to get permission and schedule interviews for people in other departments in within the company
Understanding how claims work
To gain an understanding of how insurance claims were calculated and processed before I started the project. I collaborated with stakeholders and subject matter experts to visualize the steps taken by Allstate during a total loss to determine the scope and complexity of this process. We identified the steps customers go through, the average timing of each stage, and the backend tasks required to process a claim.
Note: When an insured vehicle is damaged so severely that the cost of repair is greater than the vehicle’s value, it is considered a “total loss” which an insurance term used throughout this case study.
There are six main steps during a total loss which takes over 3 weeks to complete on average — most of which are invisible to the customer. Main pain points for customers are the delays in this process and not knowing what the status of their claim is.
In order for a claim to proceed from one step to the next, aspects of vehicle’s condition are first manually assessed, then fed into many interconnected systems that calculate how much money a vehicle is worth.
*Image above is blurred to conceal confidential information
Framing and choosing problems
Prior research review
At the beginning of the project, the stakeholders hypothesis was that the primary problem customers had with a claim was the amount of money offered for their unrepairable vehicle — known as the settlement. While this seemed likely, I wanted to verify we were focusing on the right problem. After networking across various departments, I discovered a survey created by the marketing team that revealed why, and to what degree customers were displeased with their total loss experience.
Multiple choice question results shown. The most severe problems customers had were based on the settlement, which is the actual cash value presented at the end of the claims process. Note that there are a lot of opportunities beyond money to improve the customer experience.
Open ended text responses shown. The size of each box shows the frequency of themes; the larger the box, the more frequent that theme appeared. Note that the settlement and valuation (money focused) issues appeared less often than timing delays and poor customer service. There are many areas to improve the total loss experience.
I extracted relevant information from the survey and confirmed that indeed, the amount of the settlement was the primary problem customers had. Customers were also dissatisfied with the way in which Allsate explained how their vehicle’s value was calculated. Other Issues that scored low customer satisfaction ratings revealed themes I would explore further using other research methods later on in the project.
Secondary research meta-analysis
To put Allstate’s finding into a broader perspective and provide context, I looked beyond the company to the entire auto insurance industry. The main findings from literature review echoed what we found in the survey — the perceived fairness of a settlement is the most important factor for customers, followed by the explanation and transparency of the claims process. However, It’s not enough that the experience is fair and efficient — resulting in logical satisfaction. The customer must also feel cared for in an effortless process that they understand — resulting emotional satisfaction. This makes holistically improving all aspects of the customer experience as an integrated system paramount.
I also Identified 15 relevant psychological heuristics and cognitive biases that are likely to affect customer’s attitudes and behaviors during a claim, which should be considered when designing future service solutions.
I read and synthesized findings from market research journals, insurance industry white papers, and behavioral economics literature to identify other opportunities and models that would help inform possible solutions.
- The industry average customer satisfaction for how customer claims were handled is 7% better than Allstates. Aclear opportunity for improvement.
- Allstate customers appear to be more loyal than the industry average. Whereas the industry average of 1 in 3 customers likely to defect in the next 12 months, Allstate’s defection rate is only 1 in 4 in the next 18 months.
- The claim is a very strong trigger for switching insurance carriers. Customers that have submitted an insurance claim in the past two years are almost twice as likely to switch providers than those that have not.
Interviews 1 — holistic claims experience
I spoke to ten people who were currently going through a total loss or recently experienced one within the last six months. Using themes identified from prior and secondary research, I wrote a discussion guide that helped reveal why and how participants think about the value of their vehicle. A nuanced, first-hand view of how insured customers experience the Total Loss claims process emerged, and revealed new pain points and opportunities while enriching understanding of previously identified themes.
I conducted remote semi-structured interviews with participants. I asked them to walk me through their total loss claim experience while discovering what other insurance companies did well and areas they failed to meet expectations.
In these interviews, I focused on the emotional highs and lows of their journey, and used a psychological tool known as Plutchik’s Emotional Classification Wheel to aid participants in articulating their attitudes and reactions to their total loss claim.
Interviews 2 — valuation focus
I co-designed three stimuli with a UX designer — rough wireframes were used to discover how participants would react to valuation at various times in their insurance journey, how they process valuation presented in various formats, and gain insight into how they might respond if our valuation doesn’t match their expectations. The stimuli were roughly based on existing Allstate wireframes, but were streamlined and altered so that only necessary UI components relevant to the scenario were present, and took liberties in order to simplify screens.
Initial thumbnail sketches of the scenarios and screens
Final stimuli for each of the three scenarios: 1. Valuation estimate at the time of generating a quote, 2. Valuation estimate during renewal of an insurance policy, and 3. Valuation estimate at the first notice of loss (when a claim is filed)
We showed the stimuli to six participants who recently experienced one within the last six months. In order to encourage more authentic responses from them, each stimulus was customized to show the vehicle each participant’s currently owned, and valuation estimates and settlements were tailored to reflect their current vehicle’s value based on real-world calculations. This made each scenario easier to understand, more realistic, and increased the sense that there were specific and tangible monetary outcomes to react to.
Each participant reacted to stimuli that showed the vehicle that they owned, and actual valuation estimates based on the make, model, year, and self-reported condition of their vehicle.
A participant reacting to a stimulus that explained how the final valuation of their vehicle was calculated. This screen shows one of several ways in which the transparency of communication could be presented.
To aid participants in articulating what elements of each stimulus they liked and why, I utilized a set of adjectives based on the Microsoft Desirability Toolkit, which added words indicating brand and experience attributes that we wanted to emphasize.
Based on reactions and comments from participants, I identified elements from each stimulus that were most successful at informing participants enough to feel confident that they were being offered a fair value for their vehicle without being overwhelmed or confused.
After notes were collected and organized for the interviews, I looked for themes to determine what the most important and frequent issues that arose.
I reviewed notes taken during interviews after each session to consolidate findings and takeaways, making analysis of multiple sessions easier and more efficient later on.
Affinity clustering — these were the final themes and sub-themes which arose from both rounds of interviews.
Customers are often unhappy with the cash value offered for their unrepairable vehicle because their expectations are based on resale values, which the insurance industry doesn’t use.
After the previous rounds of research and testing it became apparent that when customers aren’t pleased with the amount Allstate offers them for their car, it’s not necessarily the final dollar amount. It’s the value offered compared to their expectations, which many form based on Kelley Blue Book. KBB is oten customer’s primary point of reference for determining a vehicle’s worth. If Allstate influences customer expectations, then they are likely to accept an offer. Reactions to stimuli showed that valuation reminders were not helpful, which led our team to rethink the rationale that customers use to think about the value of their vehicle.
Almost all participants compared the valuation of their vehicle provided by Allstate to Kelley Blue Book, rather than past valuation estimates.
Valuation isn’t regularly top of mind or memorable because it isn’t regularly useful.
Customers have little reason to think about the value of their vehicle on a regular basis.
Where and how we talk about value will drive our ability to set expectations.
We should talk about value in the right places by paying attention to when vehicle value is useful in a customer’s journey. We should talk about value in the right way by taking note of customers’ value-related questions/concerns, and by building our experience so customers receive an accurate valuation, understand where it came from, and understand what will cause it to change.
We need to set and re-set customers’ value anchors.
For customers to trust our valuation, our experience must acknowledge that individuals will question our valuation and compare it to Kelley Blue Book. In turn, our process should make the setting of an anchor point a collaborative process, one that a customer can be invested in.
Not everyone can (or will) answer vehicle condition questions accurately and honestly.
Customers may struggle accurately answering vehicle condition questions.
Insurers are often not given the benefit of the doubt. Beware unintended consequences!
Customers question how their interactions with an insurance company will affect their financial bottom line. When we ask customers vehicle value-related questions, they worry that their responses may hurt them down the line.
We need to reset customers’ value anchor away from Kelley Blue Book in order to set our own. For customers to trust our valuation, Allstate’s total loss claim experience must acknowledge that individuals will question our valuation and compare it to Kelley Blue Book. In turn, our process should make the setting of an anchor point a collaborative process, one that a customer can be invested in through education provided by and communication with Allstate.
The concept of Anchoring is essential to creating a better customer experience. An Anchor point bias in decision-making where the first pieces of information presented to someone become a reference point for making decisions. Once an anchor is set, other judgments are made by comparing new information to that anchor. Anchors that come from trusted sources (most notably Kelley Blue Book and are provided most recently are the strongest. Allstate needs to control the anchor point so a customer perceives their offer favorably.
Reset customers’ value anchor away from Kelley Blue Book in order to set our own.
I created a list of insights which were organized around seven key themes. Each theme contained several sub-themes that clarified and elaborated on each topic in greater detail. Recommendations followed which provided specific guidelines on when and how to communicate vehicle valuation, as well as how to improve the total loss projeccess in general. Every recommendation was paired with the corresponding themes that would be impacted if a recommendation was followed to make the impact of each (right hand columns with orange and green icons).
The interviews revealed seven overarching themes that were captured in video highlight reels, where clips and moments within them could be found and organized via a tagging system. These videos were shared across the entire UX Research department to create foundations of a video research repository in hopes that teams could learn from one another.
Customers expect different timelines during their claim process, and those expectations are influenced by the needs and responsibilities in their day-to-day life. Depending on the context, faster is not always better.
2. Personal Impact & Control
Customers are ultimately focused on how their decisions, actions, and interactions with the insurance company will affect the financial bottom line.
3. Input Accuracy
Very rarely do customers express a desire to be dishonest, but they do have a shifting frame of reference for the material they are familiar with.
4. Trust & Fairness
Customers are wondering if they can trust the insurance company to make fair decisions on their behalf. They want to make sure that the outcomes are fair, and don’t trust valuations blindly.
5. Guidance, Support, & Communication
Customers expect assistance and status updates throughout their Allstate experience. The importance of individual steps is contextual, and customers should be guided through each one.
6. Comprehension & Understanding
Customers need a frame of reference for assessing the condition of their vehicle, since the “condition” is a subjective evaluation.
7. Unintended Consequences
When condition and valuation are introduced into existing touchpoint (such as quote or renewal) will alter how customers behave in various ways that are difficult to predict.
When and Where to talk about Valuation
Recommendations organized by theme
Proceed with Caution
- Avoid Risky Touchpoints
Be wary of communicating valuation at touchpoints where customers believe their premium is at stake and is being determined based on factors beyond their control. Times such as during quoting, changing a policy, and renewal are places where customers may feel more vulnerable. New information will likely distract users from their objectives at these touchpoints, which may alter their behavior in unpredictable ways at moments where they may be triggered to contact their agent or consider changing their insurance carrier.
- Determine Ideal Touchpoints
- Leverage Proximity and Recency
How to talk about Valuation
Recommendations organized by theme
- Stay One Step Ahead
Whenever we state value, answer the questions we know users are asking, and remember that we’ll be benchmarked against Kelley Blue Book or a similar tool. Preempting these questions will increase our ability to create new, more powerful anchors for customers to adopt.
- Set Dynamic Expectations
Consider how we might inform / educate customers about how their car’s value will change over time, what the expected / projected rate of depreciation will be, and why these changes occur. Explain that while there are industry-wide standard models for depreciation that each insura
- Format in Ranges
Show value as a range when appropriate, not a single number. Determine the most advantageous low and high points of the ranges.
- Frame Valuation as Beneficial
Help customers understand why they should care about value — “why should I think about my vehicle’s value if I’m not looking to sell it?”
- Be Transparent
Explain how their valuation will be or was calculated, and what the impact of the valuation will be.
Recommendations organized by theme
Proceed with Caution
- Remember that Information Provided by Customers may be Unreliable
Be wary of relying on customers’ ability to accurately describe their vehicle options and condition. Think about how we can provide a value estimate with minimal information gathering, how to induce customers to provide information to refine our estimate, and how we can gather information in a more reliable way. Not everyone can (or will) answer condition questions accurately and honestly. When you aren’t a car expert, or are asked subjective condition questions, you may be unable to accurately answer detailed questions. When you question an insurers’ intent, you may not want to. If a customer sets their condition inaccurately, then the estimate becomes an ineffective and potentially detrimental anchor.
Consider Doing, If you allow user Input
- Set Expectations
Let customers know upfront what information will and will not impact their premium. Tell them how personal data will be used and/or why you are asking for it.
- Frame Information Sharing as Beneficial
Frame the sharing of information as beneficial to customers, and explain why sharing information is advantageous. Customers are more likely to share information if they’ll receive value in return.
Provide guidance to help customers accurately provide value-related information.
General Claims Experience
Recommendations organized by theme
- Help Customers Form Realistic Expectations
Help customers form realistic expectations regarding what will happen during a total loss, especially with regard to time and touchpoints.
- Help Customers get a 360° Perspective of their Claim
When claimers can view content in multiple ways, such as by type (claim), action (task to be undertaken), time (when the event occurred) or player (the participants or vehicle involved), they gain confidence that a resolution is in progress. Provide more self-service options, as well as personalized service when needed.
- Provide Guidance
Guide customers and walk them through each step needed in the process, not only the Allstate interactions. Consider extracurriculars — what customers are doing already and would do on their own anyways.
- Be Consistently Proactive
Keep customers informed of what’s happening throughout the entire process, especially while they are waiting. Update them periodically as to the state of their claim’s progress, even if that update is that there is no update. Identify most common issues at milestones and key touch points, and provide easy access to answers, explanations, tips, and other helpful information.
I created a presentation slide deck with appendixes when presenting to stakeholders in person at the end of this project.
Presenting findings to stakeholders in our final meeting
Video Highlight Reels
In addition to the report, the interviews revealed 7 overarching themes that were captured in video highlight reels, where clips and moments within them could be found and organized via a tagging system. These videos were shared across the entire UX Research department to create foundations of a video research repository in hopes that teams could learn from one another.
Shared foundational elements of a video repository with colleagues that would allow researchers to easily discover relevant moments within footage gathered by other teams.
More interviews were planned than we were able to conduct
In my initial research plan, I proposed talking to Allstate employees who have direct and frequent contact with customers going through a total loss, including Express Agent Call Center Representatives and Total Loss Field Adjusters. This would have allowed my team to benefit from those employee’s vast experience with a larger number of customers. Unfortunately, due to scheduling issues, time constraints, and other logistical issues, we weren’t able to speak with those additional people.
- Created guidelines for how and when to best communicate vehicle valuation with customers that served as the basis for further prototypes and pilot studies
- Made recommendations on how to increase customer satisfaction during the total loss experience beyond just the valuation of vehicles
- Discovered how customers form expectations for their vehicle’s cash value
- Determined ways to positively influence customer’s perception of cash settlements (actual cash value) offered
- Prevented stakeholders from moving forward with risky solutions by invalidating concepts at an early stage.
After presenting this research my contract ended, so I created a list of suggested next steps my stakeholders and team could take further action if deemed necessary. Next steps:
- Conduct a survey In order to determine how much of the larger population relies on Kelly Blue Book as their primary way to determine their vehicles value via a questionnaire. I’d like to ensure that the findings from this project’s sample are reflective of the general customer base.
- Turn stimuli into concepts for testing UX Designers should utilize the guidelines from this project while considering relevant cognitive biases and heuristics mentioned in the final presentation and report to design higher fidelity mockups in order to determine how participants react to new proposed solutions.
To solidify what we learned from this project, I conducted a post-mortem retrospective workshop with my teammate and stakeholders. Here are the main takeaways:
- Importance of stakeholder involvement From the beginning of this project, it was very difficult to get our stakeholders to participate and actively engage. I learned how important stakeholder buy-in and continuous involvement is. I now make communication and collaboration a more formal part of my research plans by explicitly including communication goals and guidelines when working with a new team.
- Adapting to involve more participants in the research
In the end, we conducted fewer interviews than initially planned due to logistical issues. If I were to continue this project, I would have interviewed more participants to generate a more robust and representative findings and insights
- Reaching out to collaborate
As a contractor, I’ve learned to network as soon and as broad as possible to develop connections and relationships that will make more comprehensive research more likely.
- Articulating decisions better In addition to the three scenarios we developed stimuli for, I unsuccessfully advocated for a fourth scenario, at a time that is not linked to any official insurance activity. I thought this stimulus would have provided more realistic responses from participants, but my team and stakeholders were not convinced. I’ve learned the importance of articulating the reasoning behind my decision making, and acceptance that sometimes my ideas just won’t move forward.