I just spoke to Martin White about his up-coming enterprise search workshop and while we were ostensibly working out the logistics of the day, Martin gave me an introduction to the complexities of intranet and website search.
As a previous intranet manager, and current consultant, I have heard myself promise better search to stakeholders without knowing what the heck that even meant. Martin’s five crucial points fill in the gaps and explain why it is incredibly hard to make incremental improvements. Martin believes holistic change is needed to deliver a search experience that creates business value. The first one (below), I totally get.
Five vital search matters
1. Quality content
A poor search engine is very capable of finding poor content. An excellent search engine is even better. To deliver more relevant search results, without the duplications, variant versions, and irrelevant conversations and unspecified, untitled reports, the organisation must get control of its content creation and maintenance practices. Not only do the official intranet contributors / publishers require training, but so does anyone who works with documents, presentations, and files.
Search is not independent or above content. Garbage in, garbage out.
2. Indexing tactics
How is your search engine configured to crawl, deconstruct, and index the swathes of unstructured and semi-structured material across the intranet, shared drives, document control repositories, and discussion forums?
The technical configuration is not a ‘set it and forget it’ matter, as the complexities of context (words mean different things in different settings) means that appropriate metadata is vital to help the search engine prioritise relevant content, based on author, location, org structure, date, common words, keywords, tags, and synonyms.
As a content contributor, you and I can help by writing truly appropriate titles for our articles and documents, but the search team and the IT team have to work together to make sure that content is crawled on a timely basis. You expect content to be available as soon as it is published, but until it is crawled and indexed it will be invisible to anyone searching.
3. The search dialogue (UX)
How people make their initial query and then refine their input is a dialogue between the person and the system. It is now quite common for the search results page to offer faceted search or filters, but are these appropriate? Offering a list of departments as a facet is only helpful if you know which department published the information you are looking for. File type information can be misleading. A PowerPoint presentation might have been saved as a PDF, so looking at the PPT file filter will not find it. Of course if you know something exists you will probably persevere until you find it. But what about when you don’t know. If you can’t find it could it be because it does not exist, or it exists but has not yet been indexed, or it has been indexed with incorrect tags?
Different people have different search needs depending on their objectives, and context within the org. Google, for example, offers many separate search experiences, with specific functionalities. Take a look at Google Scholar, Google Finance, and Google Shopping. Many
of us use the regular ol’ Google to find research reports when we could use Scholar; it offers a different set of filters down the left of the results page (published date, for example) and identifies publicly available versions of subscription content.
The regular ol’ Google offers simplicity, something everyone seems to shout for, yet Google offers some incredibly powerful (and non-intuitive) features that you must learn to use. Try Googling ‘Google tips’ and try ‘Google tricks’ if you have time to spare). So, if you can learn to get so much more out of Google, shouldn’t we help people learn how to use our intranet search? Martin says everybody needs training. Do you even have a search tips page? Roche Pharmaceuticals has a comprehensive Intranet Academy site to support its 90,000 intranet users which includes pithy search tips.
4. Relevance scanning
How do you, as an individual, recognise what is relevant on the search results page? (You know what they say about page two on Google it’s a good place to hide a body. What makes you click a result and what makes you start your search query all over again? So what is it about the results display that makes a person feel confident in clicking and opening / viewing a piece of content? Going back to the first point, titles are so important. Clickable author names can be valuable as often we want to find a person rather than a document. ‘People search’ is also very important and very challenging. You are trying to find “Michael Slanofski”, but will your people search engine find Michał Szklanowski or just tell you there is no such person?
The search engine can only surface (show) what it has access to – it cannot re-interpret your document’s title or the rambling opening paragraph of your news articles. Does your content carry its last modified date on it, or is that date merely when IT migrated the content to a new server?
5. Search support team
Search applications are at the intersection of database design, computational linguistics, the mathematics of probability, and user interface design — creating a decision-support application, not just a search tool. IT and business teams have to be brought together by the search manager. Most organisations have a webmaster, but never a searchmaster. Without search team members tracking search logs, training users, and developing better relevance tuning, all the technology will be wasted.
I very much look forward to learning more from Martin during his search workshop in April, and I hope you’ll join me in attending. Martin will cover website search, intranet search, and enterprise search. It’s a semi-structured hands-on workshop, not overly technical, and Martin says that you’re welcome to discuss the agenda with him right now, should you wish to contact or tweet him.
- Martin White’s enterprise, intranet, and website search lab – April, London.