ºÝºÝߣ

Editor's Notes

1. HistoryWe began this project in 2010. At the time, the Liberty Mutual website had transitioned from a small site with limited content to a large website with multiple lines of business. So our old-style navigation was no longer well-suited for users visiting the website of a major insurance companyIssuesLimited number of navigation optionsUsers could access only nine categories in the top right corner of each pageFour of the nine categories took users to a different websiteSome options were not top-level items that users wanted (such as Member Rights)And we lacked a traditional navigation bar with drop-down options, which would give our customers access to so many more categories
2. ScopeWe needed to create a new navigational hierarchy, one that would replace the current one and allow:Current and potential customers to easily find specific options under main categoriesHigh-Level ObjectivesTo focus on the navigation and organization of the website, including first- and second-level navigationTo make sure our users could understand the navigation labels we wanted to use (keep away from insurance jargon)To make the site attractive to customers (delightful, satisfying experience)Again, all with to goal to makecontent discoverable and easy to find
3. So we employed this methodology that combined online card sorting, physical card sorting, and focus group discussion with 26 participants. To prevent any bias, we ran the card sort at a neutral off-site facility. We recruited a mix of customers and non-customers, following our standard recruiting criteria. We conducted one session with 13 participants in the morning, and a second session with another group of 13 in the afternoon. Each lasting two hours. Individual, Online, Unmoderated, Open Card SortHere¡¯s how each session went. First we started with the individual card sort. Here participants completed an online, unmoderated, individual, open card sort, using WebSort. This took 45 minutes. Team-based, Physical, Moderated, Open Card SortNext, we asked participants to take a 15-minute break, and then we set up for the next part of the process, the group card sort. During the break, we removed the laptops and laid down brown paper to cover each of the tables. We brought our participants back in. Next, we handed out Post-It Notes that had the same 79 labels we used in the individual card sort. We asked participants at each table to work together and complete an open card sort as a team. As each team worked together, our facilitators moderated each team¡¯s activity, and a note taker captured qualitative feedback and group discussion.After they finished the group card sort, weasked each team to nominate a spokesperson to give a short presentation, explaining how and why they organized the information the way they did. The group card sorts and presentations took approximately 40 minutes.Focus Group DiscussionFor the final 20 minutes, we conducted a focus group discussion. Participants sat in their original seats, with the completed group card sort results still in front of them. Members of our research team took notes. Examples of discussion topics included:How would you like to have your information organized?How many categories should there be? How many are too many? Too few?Were there any items that were difficult to categorize or name?
4. After the 26 individual and 6 group card sorts were completed, we began analyzing the data. Three members of our researchteam analyzed the individual card sorts, and three other members of our research team analyzed the group card sorts. 1. Standardized category labels and combined similar categories2. Compared grouping of labels3. Created and compared site maps4. Finalized navigation categories
5. Our 26 participants created 210 unique category labels during their individual sorts. Our 6 groups of 4-5 participants used the same cards for the group sorts, and created 46 unique category labels. Some participants and groups used the same labels for categories (for example, ¡°About Us¡±), while others used different labels for similar concepts (like¡°About Our Company¡± and ¡°Company Info¡±). As suggested by Donna Spencer, we first standardized the main categories and applied consistent naming conventions. We gave categories with similar names or concepts a consistent name and combined groups where participants used the same basic concept but a slightly different label. For example, standardizing a category name of ¡°About the Company.¡±
6. Second, each member of each team separately combined similar standardized categories. Each member then presented and explained their categorization, and then each team agreed upon and arrived at our totals of standardized categoriesFor example, in the group sorts, participant groups put an average of 13 cards (40/3) into the ¡°Auto Insurance¡± category and used 18 different cards. This category has an agreement number of 0.74, meaning that 74% of participant groups put the same 13 cards in this category.
7. Third, we compared the grouping of labels between the online, individual studies and the in-person group studies to see if there were any patterns. We generated ¡°tree diagrams¡± or dendrograms, which display not only the relationship between items but the strength of that relationship, or how frequently items are associated. Both individuals and groups wanted to separate information from tasks. For example, information on benefits and features ¨C such as our labels for ¡°Accident Forgiveness¡± and ¡°Unlimited Car Rental¡± ¨C were isolated into a separate information category, ¡°Coverages & Benefits.¡± Alternatively, the Auto Insurance category was focused on tasks, such as ¡°Tool to Estimate Car Insurance Coverage.¡± This revealed a key finding that both the individuals and the groups organized the content by information AND by task.
8. Next, we created site maps to organize the results from the individual and group card sort exercises. What this visualized was that individuals had sorted items into a narrow and deep hierarchy. The site map created from the individual card sorts consisted of only 3 major categories ¨C (1) Our Company; (2) Policy, Billing, and Advice; Types of Insurance.
9. But what the group card sort revealed was that our groups had sorted items into a medium navigational hierarchy, with 6 major categories, or double what our individual card sorts showed us ¨C (1) About Us, (2) Auto Insurance, (3) Home Insurance, (4) Insurance Basics, (5) Insurance Products, and (6) Claims Center.
10. Finally, we created a 6-category navigational bar.We moved About Us into a top-level navigation bar that all Liberty Mutual websites share. We moved Life Insurance to the same level as Auto Insurance and Home Insurance, based on the objectives of both our participants and our stakeholders, who wanted all insurance products at the same category level.?
11. The combination of individual card sorts, group card sorts, and focus group discussions allowed us to:Obtain qualitative and quantitative data in one dayRecruit a large sample sizeSave money and timeThe benefit of this unique method is that the individual card sort activity enriched the group card sort activity and focus group discussion, aiding in productive negotiations and discussions for the group exercises?
12. Two years later, the site navigation has had a positive effect on the businessKey performance indicators and web analytics demonstrate that customers are actively using the new navigationIn multiple usability tests since the navigation update, users have commented positively on the navigation, rating it as one of the best attributes of the siteIn its 2012 Insurance Website Evaluation, J.D. Power and Associates stated that ¡°Liberty Mutual provide[s] menus that [allow] shoppers to easily find relevant shopping tools on any page¡±