Manipulating WordPress Canonical URLs

For many people, it’s not necessarily a big deal to move a blog from one domain to another, and losing social sharing metrics as a result (at least when you’re like me, and most shares are single digits, anyway). This commonly happens because the share counts get tied to each unique URL. If that changes, then so does the count. Virtually every social platform, however, allows you to specify the URL that’s used when sharing via special <meta> tags. This whole issue becomes much more important, though, when you’re a company with many hours and lots of marketing dollars vested in social metrics and the value they drive to your business.

Read More

OTC Goes Bold With Redesign

I want to extend a sincere congratulations to the folks at Ozarks Technical Community College on their redesign. It is probably one of the single most brave things I’ve seen a college do with their homepage in quite some time, for better or worse. And that’s good, because that’s how everyone learns. Someone has to take a chance once in a while. What especially caught my attention though was that they basically did something I never really thought would be possible. Back in 2009, I wrote a bit on the principles of IA in large sites like a university. Several conversations ultimately were spun off that article, one of which involved talking about the idea of driving a university site’s navigation entirely through search.

Back then, it was little more than a pipe dream though. Random musings about a “what if” scenario. There’s so much to consider for it – and I’m not even talking about things like the political side of university sites – that as neat as the idea seemed, I never thought it could be done. And while I applaud OTC’s attempt, I still think the approach is not really ready – though it could be with just a little more work. Here’s why.

Majors Search Results

Majors Search Results

Probably the most important thing is SEO. If you are going to lean so heavily on search, that means your site – all of it – needs to have pristine SEO so that everything can be found and located properly. We’re talking meta data, keyword density, link text, the whole shebang. OTC is using a Google Appliance of some kind, which can afford you a lot of power (sadly, Google discontinued the Mini this year, leaving only the more expensive GSA on the product line). You can see some of that power in action if you do a search for “programs.” Note at the top, you specifically get the keymatch that they manually entered to make sure that a search for “programs” always results in the right page first thing. That’s good. Now do a search for “majors.” No keymatch this time referring the visitor to the programs page. The top matches aren’t relevant at all, as a matter of fact. That’s not to say the results are consistently bad, but in this approach, there’s just so little room for error.

PSU’s unified search

PSU’s unified search

Another pain point for me here is the use of the stock results page as well. It’s bland, uninteresting, and doesn’t invite the user to explore the results. They have added additional search options above the box, but they aren’t integrated at all – each is a different landing page that isn’t necessarily search related. Lastly, they don’t seem to be taking advantage of collections, which can make a GSA or Mini so powerful in getting users into the right “bucket” of information. Collections are a way of filtering content into logical categories of some kind. For instance, you could have a “News” collection that keeps all the press releases searchable and separate from the normal search. At PSU, their search is an example of both unified search and collections (seen to the right). Things like “athletics” and “classes” are collections, while “people directory” is actually a separate system. But it all works through the single interface (though the people results do go to a different results page, so it’s not entirely unified).

Something else, and this is specific to the GSAs still, is that they don’t appear to be using OneBox modules either. That’s the perfect way, for instance, to try and pull in some of those external searches from the result page header, like departmental and contact searching. For instance, do a search for “NUR 230,” a nursing program course. Using a OneBox module, they could instantly provide course information, schedules, associated books, teachers, etc. If you want more examples, the OneBox is what gives you instant results in Google when you do things like typing in a FedEx tracking number, looking for movie times, checking the weather, and so on. That’s the trick here. If you’re going to go all in, blow it out of the water. Universities have TONS of structured data that could be presented this way, to fantastic results. Won’t someone think of the user’s clicky finger?

From an article at the Community College Times:

OTC Chancellor Hal Higdon said a review of the college’s website using Google Analytics showed that more than 80 percent of site visitors find what they look for through an outside search tool or OTC’s Google search server. Often, visitors skip the front page and go directly to the search box to quickly find the information they need.

Google Analytics Site Search Usage report

Google Analytics Site Search Usage report

Admittedly, I know nothing about just what went into this research (and if anyone at OTC reads this, I live in Pittsburg, KS, about an hour and a half from you – let’s talk), but I would caution any school interested in this that analytics alone will absolutely not give you the full picture here. It can give you a lot of information, to be sure, but context and intent are intimately important to this particular endeavor. For instance, it’s easy to say that people may search a lot on a site because the the navigation or IA sucks – something analytics alone won’t tell you. So it would seem reasonable that going all search would avoid that problem, since search is designed to do an end-around on such things (this is, of course, assuming you aren’t considering things like nav and IA in your search logic). But maybe they search simply because your content sucks, and they’re trying to find something more informative. That’s a content problem. My point is, know your problems and know your goals. Have a plan for each, isolate your success metrics, and have a maintenance and measurement scheme ready.

And there certainly may be something to catering to users that search. A quick look deeper into the Google Analytics report sampled above (you do look at your search reports, right?) revealed some extremely interesting metrics. For instance, the average user spend 4:33 minutes on the site, as opposed to 11:48 minutes for users that searched. Users that don’t search viewed on average 2.73 page compared to 8.28 for searching users. But, what the analytics here don’t tell me is why. But hopefully, if your numbers are similar, you would want to know the answer to see if there’s something valuable there to be leverage.

There’s something else that bugs me, though. While I don’t want to nitpick, I feel the need to point some of this out.

“Start Here” navigation

“Start Here” navigation

In trying to mimic Google, they also used a “services header” on the homepage. That’s fine, go for it. But, I gotta admit the logo really bugs me. It just looks stuck on and clip-arty. But more than that, I am really bugged by the “Start Here” link. First off, “Start Here” isn’t at all descriptive about what to expect when I click on it. And once I did, I was confused that I was looking at a page with a careers based URI, but the content seemed to be related to academic programs. That’s just a labeling thing, but it’s a pretty major one, since it’s first chair in what little navigation they have. They also added a “more” link. While I know this is in line with mimicking Google, it smells too much like rebranded quick links. As a user, if the goal is to have me search, why would I click the “more” link rather than just type in the keyword for what I want? From the very start, you’re already inviting me to break with your intended navigation scheme, and that’s a dangerous game.

At the end of the day, I still think there’s something to this. Every university struggles desperately with IA and navigation. Awesome, global search just seems natural. The barriers that will most commonly prevent success are technology that can’t deliver, and the politics of university web maintenance. If you’re considering it, keep this stuff in mind:

  • Hire a full time SEO person. Period. Don’t be cheap here.
  • Don’t abandon navigation all together. Consider your “services” that require fast access. This requires a shift in thinking, making your homepage that of a “service provider,” rather than whatever you are now.
  • Spend six months on taxonomy. Card sorting. User research. Whatever people call something, make sure those keywords are mapped and accounted for
  • Make use of autocomplete and dynamic results (again, both things Google does). Save your users as much time as you can, and help eliminate mistakes.
  • Utilize tools like OneBox or similar systems to provide enhanced result data for commonly accessed, structured data.
  • Make sure you have a reporting system on pages. A “Was this what you were looking for?” flag people can click that will report the page and search that sent them there.
  • Accept the fact that you may have to take away a lot of editing rights from people to prevent pollution of your results. Two words – Quality. Control.
  • You might consider splitting the site into a sort of “gated” and “ungated” area, where the gated area is vetted, approved, specific info. The ungated section is everything else that no student ever cares about.
  • Respect the results page and how important it is
  • Unify your search platforms
  • Measure and track everything. Can you tell me the most viewed, but unclicked autocomplete keywords? Most common misspellings? Keywords most likely to result in an application? Bounce rate after a search result? And these are just some of the easy ones.
  • Your search needs to be smarter than your users. It should know what they want, regardless of how they ask for it. It needs to deliver, accurately, without question. It needs to adapt incessantly.
  • Hire a full time SEO person. Period. Don’t be cheap here.
Edinboro’s keyword autocomplete

Edinboro’s keyword autocomplete

Oh, there’s one more important thing here. I don’t care if your homepage is a Google knockoff or not, you should care about search. Almost all of my bulletpoints above hold true no matter what your web strategy entails. Edinboro University is one I credit with putting a ton of work into mapping keywords for things on their site to an autocomplete feature for their search. Their keyword system is a completely secondary system too, it’s not in a GSA or anything like that, but they unified it properly so the user’s experience is seamless. All they know is they are getting good recommendations that can save them keystrokes. But otherwise, Edinboro’s search is implemented just like any normal search, nothing else special about it. But the details, the little things, that’s what can matter the most.

Good search is like a life preserver. It can save a visit. It doesn’t matter if it’s just a tool, or your entire navigation. Bad search frustrates users and drives them away, and I don’t know anyone in that business. At the end of the day, I have no doubt OTC will continue to improve, and for a community college I have a ton of respect for the effort they’ve put forth here. I’m damn interested to see how it evolves.

Are You Being Used?

Have you heard of Fiverr yet? Fiverr is a service that launched back in February of 2010 as a tool for people to sell simple goods and services for five bucks. Maybe that’s planting a tree in your honor in the rain forest, or sending a letter to a random soldier, or belching your name on video. Pretty much anything goes. It’s not a terrible idea, strictly speaking, and is a nice way for people to make a little extra money doing something they’re good at.

So, what does this have to do with higher ed, and why should you care? Well, simply, this.

Search Results on Fiverr for “edu”

Search Results on Fiverr for “edu”

It’s no secret that there are plenty of black hat SEO techniques for link farming. This is also far from the first time someone tried to leverage the .edu TLD for link relevancy (Note: it seems pi.edu has finally gone away, without much fanfare. No one misses it.). On top of it, odds are you can’t make Fiverr stop these listings. Because screw you that’s why. At least, I suspect that’d be the subtext of the answer you’d get from them.

How Does It Work?

Simple, spider services have created lists of things like blogs and wikis that have unmoderated change or comment systems. The people offering these services buy or pirate those lists. In some cases, they have tools that automatically submit to sites on the list. Then you watch the spam start coming in. Anyone that runs a WordPress site understands how much trouble spam can be. If you’ve ever wondered where it comes from and why, this is a pretty good start.  In the end, the provider or their software tries to pass as a legitimate commenter and includes a link in the post text or author site (if you include the author’s link on their name) which then shows up, they get paid, and you get polluted.

This is a much less offensive and less dangerous version of account hijacking that we’ve seen in the past, where faculty, staff, or student web space hosted by the university is taken over and used as a landing page host or to drive backlinks and keywords.

What Can You Do?

Shut. Down. Everything. Okay, not really. But seriously, do review your moderation and approval processes for your blogs and wikis. Anything someone can contribute to should be reviewed to make sure you haven’t created a target. Keep some of these in mind (adapt to your environment):

  1. Don’t ignore your sites and security settings.
  2. Try simple steps like requiring at least a first post to be approved before users are whitelisted.
  3. Look at third party commenting services like Disqus or Intense Debate which have tools for addressing this that are better than yours.
  4. Many CMS’s have plugins that can provide more robust comment protection. For instanceAkismet is common for WordPress. I’ve had success with Spam Free WordPress.
  5. Add moderation or extra steps to comments containing links.
  6. Make sure links in comments are set to come through with rel=”nofollow” enabled.
  7. Limit faculty and student abilities when it comes to setting up and configuring sites, blogs, wikis, etc.
  8. Allow visitors to vote down or mark comments as spam.
  9. Turn off commenting after a certain length of time or when a blog is discontinued but still available.
  10. Set up a routine to audit your sites for this kind of spam ever X months.

None of these suggestions will likely work on their own. Some may or may not work at all in some cases. There’s no real silver bullet to the problem, as long as humans are willing to do the work manually for companies for $5.00. But, you can at least try to minimize your risk of exposure by making the effort for the spammers cost more than the time it’s worth. When they get through anyway, if you’re monitoring properly you should be able to delete the comment and blacklist the user or IP quickly enough that it becomes apparent you aren’t a high value target. The bottom line is to be vigilant, active, and take responsibility for the sites and services you’re offering that could be targets for these types of tools. Fiverr is far from the only way to accomplish this (see?), but what really matters is preventing the end result.


Photo Credit: cc icon attribution small Are You Being Used? Some rights reserved by 666isMONEY ☮ ♥ & ☠