Library work can sometimes position you heavily in one part of the organization – among the researchers – and disconnect you from other parts, especially the “back office” areas. My work has, like many librarians, moved from your traditional ask-and-answer support to broader information projects. In my case, it includes managing the Web content management system and working on some of the related technology, like site search.
One of my roles is to be a throttle, mostly the kind that brings a focus to resources not the hands-around-the-neck kind. Like many organizations, the Web site is designed with external visitors in mind but it is supported by internal users. If someone wants information, they often call before even visiting the Web site. Staff then attempt to walk the person through how to find the content on our Web site, which is like a rather cluttered filing cabinet (yes, I realize that’s a problem).
A couple of times now, staff have asked for a link to a resource to be placed “on the home page”. The resource is one they say they frequently refer callers to and it’s hard to find for staff. Another kind of request I field is the “when I search for X, it should be the first result” kind, where someone wants a particular piece of content to appear higher in the site search results.
The first kind of request is an easy one to solve. You place a link to the result on the home page. That was our process in the past, which led to a landing menu on our old site that had over 130 links on it arranged hierarchically in submenus. It couldn’t have been more unwelcoming.
But if your process needs to change – so that you can say no or, better, yes but – it’s hard to have a discussion based on anecdote. On the one hand, the team requesting it should be able to give some context to what’s driving the need:
- are they getting 50 calls a year or 500? and how does that relate to their annual call volume? In other words, how big a problem is this?
- what isn’t currently working: they can’t browse to it but it’s the first result in site search? the resource is called one thing by staff (like an acronym) but something else by visitors so staff can’t find it because they use a different term?
I’ve been trying to come up with some data as well, to help determine the context for when we should throw resources at fixing it, and whether those are within my team (a technology-centric approach) or whether there are content or other approaches that are less resource intensive. This can be fuzzy but I’ll usually look at:
- the number of times the page or resource is viewed on an annual basis. Focusing on a shorter period of time can miss spikes that occur only once a year. Google Analytics or whatever site statistics tool you use can help with this. Ideally, you would be able to filter so that you can see how many are internal requests for the page and how many are external;
- the search terms used in site search to reach that page. Again, Google Analytics can help, as it supports capture of site search queries. This can highlight differences between internal and external vocabulary. For example, we had a request to change a navigation label to “Legislation” by a staff team. But after a discussion where I showed that the word “legislation” had never been used by anyone looking for that page, but that nearly 80% used the word “rules”, a word included in its current label, we left it alone.
Without having some data-driven context to have the discussion, it becomes an argument over anecdotes. “My team is flooded with requests” or “we can never find” or “most people use Google”, and so on. My library role has meant that I’ve experienced both the business area and IT-esque responses. Data won’t create a bright line but it can help to create a frame within which to view the need.
A recent example involved a highly political initiative that staff would send visitors to. Staff felt that it was hard to navigate to. Site search didn’t help because the first result, although using the initiative’s name, was a news item, not the initiative itself. How do we make this easier for staff to find, so they can tell visitors how to get there?
Ideally, you wouldn’t have staff giving visitors voice commands on how to get around the site. The ideal Web site will have simple navigation, a heavily edited and limited content set, and a well-integrated site search. But the reality is that many sites aren’t ideal and we’re among that group.
But there are still ways to focus on content that needs improvement and figure out what the options are. In this particular case, where we had two identically named pieces of content in the top two search results, and staff were clicking on the wrong one, we could do a number of things:
- the content owner of the #1 search result could rename it so that it reflected more clearly that it was a press release, not the initiative itself;
- alternatively, the content owner of the #1 search result could delete what was now a multi-year old news item;
- the content owner of the initiative page could improve it. The document used an acronym for the initiative heavily but not the actual name of the initiative. Site search relevancy wouldn’t necessarily treat the two the same, so the news release – which used the full initiative name – ended up as more relevant than the one with the repeated acronym;
- the content owner could also add or improve the document’s summary in the CMS. This summary is often blank on our content, which was migrated in en masse and not created fresh. But the summary is ingested by site search and impacts relevancy as well. It is also the search snippet, so it is a great way to improve the clarity of the search result itself.
There were some non-content options as well. The one I was least open to was to customize the search tool itself in order to make this particular link rise from the second position to the first. It’s hard to be sure if one is entirely open-minded about this sort of thing, but I have a tendency to resist system customization when there are other options. This can be a struggle if your organization prefers to customize software rather than make other changes.
There were two staff options as well, which would have required internal staff to have a bit of education or guidance. The first would be to take advantage of the friendly URL or alias that the initiative’s Web page carried: www.our-site.com/acronym-for-initiative. The acronym is only 4 characters, so it would be easier to explain over the phone than having the person navigate, and could end the call more quickly. Alternatively, staff could bookmark the page and then, when talking to a person, could go to it and e-mail a link to the person that they could then follow.
But what about the data?
This resource was visited roughly 1,800 times in 2013, on a site that receives millions of page views. So it was not a heavily trafficked page. The search term, and variations on the term, that staff used to retrieve the page was used 150 times the year. Only 50 of those were by staff; the team who fielded the calls didn’t keep track of how many they handled. My assumptions – which could be totally wrong – were that the problem was much larger for internal staff than for external visitors. It’s hard to know, but if there has been a sudden spike in calls, then a perception can arise that something is more problematic.
In a distributed content environment – in ours, Web content management is a system function, with content creation and ownership happening and overseen elsewhere by dozens of staff – it can be hard to make the system respond to every request. In fact, system customizations that are tweaked for one item may have unforeseen consequences for others.
Again, my personal bias is to avoid system customizations unless it’s warranted. In other words, there’s a way to know that it’s necessary. I get a niggling feeling, though, about whether it’s just shirking change but with data I feel like there is less grey area for me to make that choice. In this case, it seems pretty clear cut that we can make (a) content changes and (b) process changes to help staff get to the content more effectively. The content changes should have the result of improving the result for everyone, which in itself is a benefit. It may also encourage staff to create new, or edit current, content with these ideas in mind. Sometimes it is the process that needs tweaking and not the system but there has to be an environment where the data is gathered so that discussion can happen.