Finding IA at the Enterprise Search Summit Ð Part 2

In May, I attended the Enterprise Search Summit East in New York City with our illustrious Operations Manager for the IA Institute, Noreen Whysel. We wanted to see the role of information architecture in Enterprise Search. As a search engine optimization (SEO) professional and an information architect, I have a passion for effective search interfaces, ones where the searcher experience is supported via both querying and browsing. I have known for years that an effective information architecture can actually help deliver more accurate search results. And my personal quest was to see if other search professionals validated my research.

One of my areas of interest was governance and information architecture (IA). Dr. André Schaefer, Project Lead at Raytion, defined enterprise search governance as, "The roles and processes that facilitate the operation, maintenance and extension of organization-wide search platforms in a goal-oriented and structured way." Who has the power to define labels on a website? Who verifies the performance of search labels and has the authority to change them when necessary? How should web content be archived? Labels include (but are not limited to):

  • Document titles
  • Document headings
  • URLs (web addresses)
  • Embedded text links (i.e. text links within the main body area of a document)
  • Formal navigation labels
  • Annotations/snippets

How do these labels appear in site search engine results pages (SERPs)? Who monitors search results to better accommodate searcher behaviors in future interfaces? Schaefer recommended ongoing analyses of searcher behavior and regular maintenance of content to ensure search quality.

In fact, in another session entitled, "How to Make Your Silent Search Users Talk: Actions Speak Louder Than Words," David Gaulin, Vice-President of Professional Services at Norconex, provided some actionable, practical tips to improve a search interface:

1. If a site search engine is used over and over again to find the same document, it might mean that the document is not prominent. You might need to re-label the document to make it more findable, or provide easier browse access to that document.

2. Monitor queries with no document views. If the correct title (and link) and document link are displayed in search results, maybe the labels are not effective enough to elicit a click. Or you might need to add the right content.

3. On an enterprise website or system, searchers tend to type in more keywords. You can adapt the search box to accommodate the average number of characters per query.

4. Monitor what interfaces are using -- mobile, tablet, desktop/laptop. Evolve your site search engine to adapt to specific interfaces.

5. When is the search feature used? When is the search feature used most frequently? Schedule work on search functionality to do "little" damage.

6. Monitor for acronyms and new/updated content. Do searchers use jargon? You should include common acronyms in your thesaurus as well as the spelled out meaning of the acronym. Is new or updated content appearing in search results when it is available?

Smart website owners do not leave labeling or governance decisions to technical teams, whose mental models rarely match user mental models. Nevertheless, technical teams do end up labeling website content. Getting everyone on the proverbial same page is key to an effective governance policy. Lisa Welchman, Founding Partner at WelchmanPierpoint, LLC, recommended to, "Épush results-oriented standards until they are a part of the business culture and process. Emphasize the value of standards and de-emphasize the control. People don't want to be or feel controlled."

Mobile search is another of my areas of interest, and I was very pleased to listen to Greg Nudelman, CEO/Principal of DesignCaffeine, Inc. and author of Designing Search: UX Strategies for eCommerce Success. "Design for mobile context," he said. "Mobile devices come in all shapes and sizes, and that is a big challenge for mobile search interface design."

"Most mobile devices do not have the luxury of having the space for a left navigation that houses the filters in desktop faceted search," he continued. "Even without showing the filters, most mobile apps devote as much as 24% of the available screen real estate to chrome and navigation. This takes the focus away from finding and interferes with making mobile search a fully immersive experience. I've identified some mobile search design patterns that work well on screens of different sizes, and developed some alternative approaches to mobile searching and filtering. A full third of my book is devoted to presenting immersive mobile search design patterns, including augmented reality, gestures and multi-touch."

(Note: I have purchased Mr. Nudelman's book, and I highly recommend it, as well as Lou Rosenfeld's new book, Search Analytics for Your Site: Conversations with Your Customers).

Nudelman presented 15 lessons about mobile enterprise search, and here are 3 of his most important recommendations:

1. Start from zero. Your zero results page is not an error, but a natural outcome of difficulty of typing on mobile devices. Focus on providing a way out. Don't leave searchers with a blank results page with no help whatsoever.

2. Mobile search can be, and should an immersive experience -- we need to look beyond simply using the standard tool kit. Mobile apps don't need to be complex or sophisticated to produce immediate business results. Make customized sorting, filtering, and preferences easy to accomplish (and save).

3. For Mobile, browse is better than search. Take full advantage of on-board mobile sensors (GPS, previous search history, etc.) to create personalized, in-the-moment, pertinent search results without having the user enter anything at all.

Although I loved attending the sessions at this conference, I often find great gems coming from places (and people) I don't expect. One of my favorite encounters was with Avi Rappoport of SearchTools.com and one of the Google representatives in the exhibition hall.

"There is a big honking difference between filters and facets, and Google is being disingenuous when they gloss it over," said Rappoport. "The difference between filters and facets is the difference between diving into a muddy river and diving into a clear one. If you're used to the muddy river and have familiar features to guide you, it isn't so bad. But in most cases, we'll all go for the clear river, because we can tell where the rocks are and make informed decisions."

"I think that the most important advance in enterprise and vertical search in the early 2000s was the systemization of faceted metadata search and browse, pioneered by Professor Marti Hearst of the UC Berkeley School of Information, and also the founders of Endeca," she continued. "These facets show how documents found with a particular search fit into the formal structures of the information repository. This had been done to a certain extent before, but never on so many different schemas of organization, from conceptual to mundane attributes such as date, size, and file type, which offer surprisingly valuable context."

"By including the analysis of metadata fields in the search indexing architecture, Professor Hearst was able to handle huge amounts of data quickly, so the results could show not just the categories in which some of the search results would divide, but how many would be in each category," she said. "This provides hugely valuable data to the searcher, who is performing 'information foraging' as they work, allowing them to make significantly better choices as they interact with the system, essentially removing the mud and making everything clear. By seeing the number of matches in each facet value, the searcher can avoid 'dead ends', waste many fewer clicks, understand the data source better, and locate the information they need faster, with higher confidence that it is the best available."

"Here is an example from REI on a search for shoes, with and without the match counts for the values of each facet," she continued."

Shoe height

  • Ankle
  • Knee
  • Mid-calf
  • Over-the-ankle

Shoe height

  • Ankle (302)
  • Knee (1)
  • Mid-calf (5)
  • Over-the-ankle (104)

"In short, the difference between a filter and a facet is context," she concluded. "This only applies to site, enterprise and vertical portal search, where the search engine can count on a lot of reliable metadata. Web search engines can't do this: spammers abuse metadata as they do everything else."

Recent News

Feb 9 2019

Delivered in email on February 7, 2019

Greetings,

We appreciate your patience and understanding while we worked together as a board to launch elections and figure out our immediate path forward. There are many good things that have emerged from this situation and the board has...

Jan 26 2019

Greetings,

As you may, or may not, know the Information Architecture Institute has recently been called out publicly for a Board decision relating to the handling of a Code of Conduct violation reported at IA Summit 2018 in Chicago. There have been some disturbing accusations made in...

Dec 18 2018

All funds will support World IA Day 2019 worldwide organizations

Buy this great shirt to support Wear your support for World Information Architecture Day 2019 and its theme Design for a Difference . As designers, we are responsible for creating worlds of experiences, environments,...