Have you ever used search on a company’s website and received a page of odd, irrelevant results? Welcome to enterprise search – the type used on corporate websites. Enterprise search is not the magical answer-anything-machine that Google is. It requires regular care and feeding to be effective.
In 2013, I led the first implementation of search at Vanguard’s international division for their Canadian financial adviser website.
Designing the user experience for search seems simple. Start with a search box. Accessorize with filters and a drop-down panel. Arrive at the results page. Voila!
Much lies below the surface, however. Three major components are: the content, the search engine, and the user interface.
The content consists of all the text, images, files, pages, and data available through a website. The search engine performs two primary tasks. First, it “crawls” the file repositories associated with the website and organizes the file data into a directory called an index. When a user enters and term into the search box, the search engine, then, references the index and retrieves a relevant set of results (or not).
Words are the currency of search. Much maintenance involves the use of language – both inside and outside of the content – throughout the site.
Every business and industry uses a particular set of words that are specific to their domain. However, to the delight of poets and lawyers, words are slippery. For effective enterprise search, words must be “fixed” into (what is known as) a controlled vocabulary.
This controlled vocabulary then informs a variety of tuning and maintenance techniques so that the content of a website is understandable to both people and machines.
Maintenance extends beyond the website. Search requires ongoing cooperation between design, research, technology, content, and business teams. Most importantly, it requires a commitment of resources.
Many parameters framed this project. Vanguard aspires to be a thought-leader in new markets, so articles and commentary needed to be prominent. The IT group set the schedule and chose the technology. And design needed to respond on various devices.
But the long-range implications concerned me most. The International group had 17 (now over 30) websites that would require search after this one. That is a lot of maintenance and support for a small team in a new division. I presented a minimal design and championed focusing our efforts on configuring the index properly.
I hypothesized that a balanced, well-structured index would reduce the amount of tuning, modifications, and maintenance required downstream. That old chestnut, “Garbage in, garbage out” comes to mind. To this end, I came up with a relevancy formula. However, we needed to ship and did not explore this idea fully.
We succeeded but is this a success? As with most projects online, but particularly search, time will tell. Vanguard employs great people who know what to do. But I feel unresolved about this project. Search is a long game and I am an independent consultant. I am not involved in the products and features I design after the contract ends. And my relevancy hypothesis remains untested. Still, I’m hooked.
Search is business-critical. Content will only grow, often in unwieldy forms like unstructured social media data. Customers will continue to have high standards for search. The job for businesses will only become more complex.
Search touches every aspect of a system or product. It engages many groups within an organization. It demands collaboration and alignment on multiple fronts. Effective search and efficient maintenance indicates a high-functioning organization. Sign me up!