This is the process I used to improve search engine results for IWIN (I Want Information Now), an intranet website for financial services professionals.
Phone reps can’t search for answers using the search engine easily.
The primary audience of IWIN is financial services professionals with 0-1 years of experience. A secondary audience is 2-4 years of experience. Most phone reps with more than five years of experience have learned by answers to common questions, and they don’t need to access IWIN frequently.
To solve this problem, I worked with the developers, business SMEs, and my own team to improve content and findability.
- There wasn’t a strong governance model with a clear content owner.
- Another constraint was the system, which was an old version of Teamsite. The pages were XML files uploaded and published through the CMS, so making structural changes to the content was a difficult process.
- The search engine was federated and Sharepoint indexed Teamsite’s web pages.
Step 1: Establish a baseline
The image below is an example of the analytics that helped me understand how the search engine performed and how the audiences interacted with the website.
By using the top 1,000 searches or a random sample, I measured search term relevancy (Did the user see the top 5 most relevant results delivered by the search engine) and precision (Did the user use a precise keyword? Was the keyword relevant to the page being returned?)
Step 2: Research
This step took a lot of different forms, including cardsorting or usability tests. Other times, I relied on contextual interviews, observed phone reps searching for content, or listened to customers.
Step 3: Create new pages, rewrite existing ones
Sometimes, I rewrote a page. Other times, I edited it. For example, one preferred term by Schwab was Transfer of Account. However, reps often searched with the words “transfer account,” which impacted the top 5 results. I optimized the page by adding this phrase.
Other optimization efforts included linking from the topic pages and other highly trafficked pages, using more or fewer acronyms, or adding keywords to the title.
New ways to solve problems
Sometimes nothing worked, so I had to get creative.
- One way to bulldoze the constraints was to develop “synonym rings” and leverage Best Bets. These were tools available in SharePoint’s taxonomy system.
- I also analyzed the traffic patterns. Webtrends (think a more expensive Google Analytics or Adobe Experience Manager) was the tool I used to measure forward/reverse path or to analyze top pages, the bounce rate and more.
Using content heuristics
Content heuristics was a favorite method to improve findability. Here’s a before and after example below.
Outcomes and Lessons
For each content project, I used variations to improve the search results. Each time, I saw a findability improvement of 10-25% per quarter. Besides the findability, the readability also improved, as shown below.