Is it that we’re spending so much time focused on more complex user-friendliness issues (ie., can users navigate the site, is our online application usable, etc) that it’s almost like we’ve forgotten one of the fundamentals?

Dressing up link text with hover styles does nothing for the scanning eye. Finding what words are links on a website shouldn’t be an easter egg hunt. We should know *immediately* when we glance at a page that a word or phrase is a link, not after we move the mouse over it.

Take a look at these examples. Can you tell where the link text is immediately? Now imagine these little snippets were part of a much larger page with much more text. Do you still think you could pick out the links without really staring at the page – or worse, without “mouse hunting?”

There is 1 link in the above example.

There are 4 links in the above example.

There are 4 links in the above example.

There are 2 links in the above example.

There is 1 link in the above example.

Compare the examples above to these.

It’s interesting that almost all of the websites that don’t underline links use a different link style on hover, usually a vastly different style (different background color, normal-to-bold, underline, italicize, etc). Some of the hover styles add in the underline when the user hovers over the non-underlined link. What sense does this make? It’s not apparent that it’s a link until you hover? I don’t get it. There’s nothing wrong with hover styles. It can’t be the only (or easiest) way to find a link, though.

How could we measure the effectiveness of this? Good question. Can we really measure the effectiveness of changing our link styles? An easy way, obviously, is to do a quick usability test. Can we do it using our analytics tool, though? I think so. If your site uses Google Analytics and uses Site Overlay, you could measure before and after changing your link styles to see if making them more visible increase clicks. I bet it will. I think that would be an interesting test.

Let’s get back to basics and start underlining our links again. Who’s with me?

Why do I think this? It became apparent after a couple conversations during the HighEdWeb conference I recently attended in Milwaukee. First, let me say that this has nothing to do with the conference. It was a fantastic conference and I learned a ton from the great presenters and attendees. I would go again in an instant.

More than once, however, I found myself in the midst of a conversation about how using web analytics is pretty much “useless.”

I doubt that the people in the conversation had any idea who I was (who would?) and that I was actually presenting at the conference about web analytics (both conversations took place before the presentation). So, I decided to keep my mouth shut and just listen. I wanted to see *why* they thought web analytics was useless on their sites before I jumped in to defend the practice.

Conversations like these happen all the time. They could have very easily happened at any of our campuses. In any of our offices. I’ve heard it all before and I’m sure I’ll hear it many times again. So, I’ll bring up some excuses I heard and offer some recommendations.

So … here goes …

We always meet our enrollment goals, so what’s the point?

The first excuse was that the university/college had a waiting list a mile long to get into the school, so why bother using web analytics to make the site more useful?

I’ve heard this excuse many times before, even by some people at my own institution. This, beyond anything else, is the most frustrating of excuses. Why? Because it screams, “I don’t care about my customers!”

For the sake of argument, let’s forget about using web analytics to optimize marketing efforts. Let’s just talk about usability in general.

To not use a tool that can help improve the usability of a website *and* help save money just because you’re meeting your [insert goal here] goals just doesn’t make sense. Without even mentioning the fact that you may not always meet your goals, let’s go straight to customer service. Do we not care about customer service anymore? Should the fact that “we aren’t Best Buy,” mean that we completely ignore the fact that our website might suck and that users might be frustrated?

But we run usability tests. We care about our users! Yes, usability tests are fantastic and we all should run them. But usability isn’t 1 and done. After your usability tests are done, you make your design changes, how do you know it worked? Making sure your website is useful is a never-ending task. Your content changes constantly, right? Then how do you know it’s continuing to be useful without continuing to look at the stats?

Secondly, web analytics can save you time and money. Do those in higher education care about whether they are losing money (hard and soft costs!) due to poor design? Your enrollment goals might be met, but how much money are you wasting on staff time when users call in because they couldn’t figure something out on the website?

We looked at the analytics, but all we saw was that *everyone* was clicking on [insert link here].

This is where segmentation comes in. When someone says, “everyone” did something, I always question that statement. Who is everyone? How many is everyone? Did you use segmentation? For example, did you segment out first time vs. returing visitors? Chances are, if that was a “current students” or “log in” link, the vast majority of the users clicking it were returning. So segment that group out. What were the *rest* of your users doing?

If you aren’t segmenting your data, then yes, in almost all circumstances the data will be pretty much useless.

Here is a very simple example:

So if I didn’t segment new vs. returning visitors here, I’d think that my bounce rate and average time on site might be ok (although 39% is pretty high for a bounce rate). I wouldn’t have a clue that we need to dig deeper with our new visitors and find out why their bounce rate is so high and they are not spending a lot of time on our site.

In essence, this data in aggregate is pretty much useless. When you segment, you begin to get great insights.

We don’t have time to really use web analytics.

I’ve written about this one before as well. If you truly want to be user-centric, you find the time. Period. Do you have *time* to run usability tests and analyze that data? Do you have *time* to read blogs, articles, and tweets about web design, development, or whatever, for your own professional development? No. But, you find the time. Why? Because it will benefit your users. The usability tests will help you find website issues. The blogs, articles, etc. will help you improve your skills thus helping you to create a better experience for your users. Isn’t that the idea?

That’s why I don’t buy the “I don’t have time” excuse. Nobody has time.

Recommendations

So none of my ranting probably changes the fact that some people still don’t think web analytics is useful. So, how do we solve this issue? Here are some quick recommendations:

1. Segment Your Data – As I mentioned before, if you are not segmenting your data, you will either a) make a potential wrong decision based on the *average* data or b) think the data is useless. Neither conclusion is good. Segment your data!!
2. Focus – Know your business goals and website goals. Develop key performance indicators. This will save you time and give you focus. Know what reports to look at and forget the rest. If one of your goals is to achieve better search engine rankings, know what reports will help you with that and only look at them for that goal. If one of your goals is to increase online course registration, know what reports drive users to that goal and look at those. And so on … forget the rest.
3. Create a Dashboard – This will also save you time. Again, look at your goals, know what metrics to look at to see if your website is achieving those goals (KPIs!) and create a dashboard using those reports.
4. Look at Your Quantitative and Qualitative Data Together – When you’re running your usability tests, surveys, focus groups, remember to look at your on-site web analytics as well (your quantitative data – the *what*) to see if it jibes with your usability/survey/focus group results (your qualitative data – the *why*).

What I usually find is that most people who think web analytics is useless don’t know exactly how to use it effectively. That’s fine. There’s nothing wrong with that. Don’t make the mistake of thinking that it’s useless, however, without finding out more about it – without making sure you’re getting the best insights you can possibly get out of your tool.

Customer-focused. Remember that is our mantra. Web analytics can help you get there.

]]> http://www.trendingupward.net/2009/10/frustrating-conversations-we-dont-need-web-analytics/feed/ 11 http://www.trendingupward.net/2009/07/4-quick-ways-to-clean-house-using-web-analytics/?utm_source=subscriber&utm_medium=rss&utm_campaign=rss http://www.trendingupward.net/2009/07/4-quick-ways-to-clean-house-using-web-analytics/#comments Mon, 06 Jul 2009 04:54:50 +0000 Shelby Thayer http://www.trendingupward.net/?p=1567
Photo by:  Marcin Wichary

So I’m taking most of this week off to clean house. Yes. I’m serious. The *real* reason for the house cleaning is that my family is coming down next weekend with a toddler and someone who is allergic to cats (not the toddler) and I have two (cats, not toddlers).

In the middle of writing my lists (yes, multiple lists … I am a list writer) my mind began to wander to work. What are some of the “low hanging fruit” that we could clean up pretty quickly on our websites?

How can web analytics help us with this?

Let’s explore some ways to quickly clean up your online campaigns and/or websites using analytics as  a guide.

1. Identify your top entry pages with the highest bounce rates.

Stealing an Avinash idea – This is very easy to do and is very actionable as well. Remember that just because a page has a high bounce rate doesn’t necessarily mean that’s a bad thing. If your page is  ”portal” page or a “non-sticky” page by design, then you can expect a high bounce rate.

If not, then you need to find out why users are bouncing. Is the page confusing? Are you attracting the wrong audience? Is it relevant for the wrong keywords (this can also be “wrong audience”). Dig deeper and find out the reason for the high bounce rate.

This could be as easy as changing terminology and re-structuring the content of your page. It could also be as complex as re-writing your page to be more relevant to what the users want to see on the page.

2. Dump your bad campaigns.

Look at the campaigns that are total losers, including PPC. This is where you need to look beyond the click-through. Using segmentation, look at the behavior of these low-converting campaigns. What *do* they do on the site (if anything)? Even if you’re getting incredible click-through rates for these campaigns, if they are not converting or engaging in your site in any way, there’s something wrong. Are they the wrong audience? If so, get rid of them.

There is another side to this, though. Do you have campaigns where there is low conversion rate but those users happen to be engaging in your site in other ways (using search, knowledgebase, clicking out to your social media sites, etc.)?

These users may be in the research stage of the buying cycle and may not convert right away. Heck, they may not convert for 6 – 12 months (and by that time their cookies have definitely been deleted!).

Obviously the endgame is the conversion. My point here is only to say that you should take a look at more than *just* conversion when it comes to your campaigns. Other metrics can give you some key insights as well.

3. Clean up those 404′s.

There are a number of ways to clean up possible 404′s (Google Webmaster Tools, Xenu Link Sleuth), but my favorite is to find the referrers to your custom 404 page. What’s great about the 404 page referrers report is that it can tell you what the most popular external links that go to 404 pages. If you have a site with a ton of pages (including legacy pages), then you may need to prioritize. This is a great way becuase it shows you the most “popular” links that go to 404s.

We’ve found that a lot of these outside pages are actually pages within our larger school website that point to either very old/outdated or wrong URLs. These are simple to clean up, just pick up the phone and give someone a call.

4. Give the people what they want!

Ok, so this isn’t really “cleaning house,” but it can be a quick way to give your users what they want. Obviously we want to make it easy for users to get to the information they need.

What about when those *needed* pages change depending upon the time of year? Do your most popular pages and top internal search terms change depending on the time of year? It always seems that about a month before finals, our numbers for certain terms (and the popularity of those pages) skyrocket. For example, right before finals the instances of the term “proctors” jumps as does the popularity of the page that explains about getting a proctor (our website handles our online degrees, so this a relevant term for our current students).

So, if you have a similar scenario for different times of year, make it easier for your users to get to those pages. Put links to these pages in prominent places on your most popular landing (entry) pages (updates section? news and events section? etc.) Then simply remove them after the fact.

These may or may not be events. Of course we always put up links to upcoming events, but what about easy ways to get to needed tools or pages at certain times of year?

So take a quick minute to make a list. What kinds of things can you clean up quickly on your website?

This issue is always brought up at semester-or-fiscal-year-end because it’s the time that a lot of policies change.

Case in point – we recently changed a policy university-wide. A communication went out from the university to change the policy on individual college or unit Web sites.

Just for kicks, I ran a quick check for the policy on our university internal site search to see how many places the policy showed up. 1,119 internal pages (most at different internal Web sites or sub-domains) contain the name of the policy. Granted, some of those probably link directly to the main university page that contains the policy. What I found, though, is that a *lot* don’t.

I checked out the first 3 results pages of that search. The first 2 results didn’t even belong to the *main* policy page , but that’s another post for another time. Almost all pages that included the text also included the policy itself (of those on the first 3 results pages). So, there are literally 10, 20, 30, 100? iterations of the same policy within the different university Web sites. 

This doesn’t just happen at Penn State, obviously. I looked on 6 different Big Ten school Web sites for the same policy. The same thing is happening on *all* of those Web sites I checked.

Yes, the policy makers (or messengers) sent out communications for Web sites to update the policy. Do you really think every site will do that in a timely manner? 

Are there university standards saying that you have to *link* to policies, not duplicate the policy on your Web site? Maybe. Maybe not. 

Where there seems to be a lot of this happening is on different college Web sites. For instance, the College of Engineering or the College of Mathematics Web sites will re-iterate an admissions policy instead of linking to the policy on the admissions Web site. 

Those same college Web sites say the same policy on an admission requirements *page* within the Web site and also an *FAQ* page as well. So the duplication may be happening twice within the same college or unit Web site.

Then, when the policy changes and they aren’t all updated, there is inconsistent content and confused users.

Obviously everything isn’t cut and dry. For instance, a policy may be slightly different depending on the user (audience). 

There are *many* policies that are the same for everyone (or the same for *big buckets of users* – for instance, all undergraduate or all graduate), however. 

For our sanity and for our users sake, shouldn’t we house policies in one appropriate place and link to those policies instead of duplicating them on each sub-site within the University?

Why the meta description is still important. The meta description won’t help your ranking in the top search engines, but it’s still very important.

It describes your page. According to an old SEOmoz post, the meta description is used:

1. To describe the content of the page accurately and succinctly
2. To serve as a short, text “advertisement” to click on your results in the search results
3. To display targeted keywords, not for ranking purposes, but to indicate the content to searchers

The meta description is usually listed beneath the linked title in most search engine results. Key for click-throughs.  The keywords users search for in the top search engines are also bolded in the search result (the meta description and title tag). Again, key for click-throughs.

Quick example. If I’m a user searching for liberal arts colleges in Wisconsin and I happen upon the below search listing, I have a nice little blurb that tells me what the page is all about. Since this is the homepage, it tells me what the school is all about. I know nothing about this college, but taking 1 or 2 seconds (and that’s what users take) to read the description helps me decide if I should click through or not.

Search Result #1:

Compare that to the listing directly below it. I’m deleting the school and names.

A Great College
A Great College work pref 414.555.1212 work 1.800.555.1212 4400 South 1st street … A Great College named one of the nation’s 10 leading schools for …

Or the one directly below that:

A Great University Home
Smith joined the physics department in 2006 as part of the Great University Fellows in the Liberal Arts and Sciences program ..

Do you have any idea what “A Great College” and “A Great University” are all about? We know the phone number of one and the recent hire of another, but we still have no idea about the homepage or the school. Don’t be fooled into thinking users will click through to find out, either.

Of course we’re talking about the homepage here, but most secondary pages should have a meta description – a *different* meta description that describes those pages as well.

First impressions count. For many schools, users searching may have never heard of the school. If they happen upon the school in a search listing, that title and description is the first impression – make it relevant.

Leaving the meta description blank so the search engine can scrape your page and use what *they* think might be relevant doesn’t make much sense. Writing a good meta description is not easy, but it’s worth it. Take the time and write it!

Kyle James just wrote a great SEO post on .eduguru last week about optimizing higher education websites for search engines. This is an area that higher education sites really need to focus on and, as a whole, we haven’t.

SEO is so important to get users to your website and the meta description is important as well – just not for page ranking. It’s one thing to rank high. It’s another thing to entice the user to click through to your website.

Obviously the higher the rank, the more users will click through, but if your school is listed on a page with 6 other schools (or 6 lead generation sites who spend way more time on SEO than you do!), the meta description plays a key role in that click through.

Directory Listing Descriptions. If your website is listed in the Open Directory or Yahoo Directory, Google and Yahoo both can grab your directly listing from DMOZ or from Yahoo Directories and use it as your “description” in the search engine results.

To get them to use your meta description tag, simply use the following snippets in your header (<head></head>):

<meta name=”ROBOTS” content=”NOODP,NOYDIR”>

This tag tells Google, Yahoo, and MSN to use your meta description instead of the directory listing description.

Of course, once you get users to your website, it’s your job to *keep* them on your site. That’s another post for another day, though.

What about writing effective internal emails, though?

I know, I know. It’s not as important as our external emails and so, it takes a backseat. Understood. I have to tell you, though, we all need to have a refresher course in effective internal email writing.
My top 5 internal email pet peeves:

1. The email goes on and on and on and on and on and on. And on.
2. Entire paragraphs are bolded or emphasized.
3. The font colors, sizes, and styles are ever-changing.
4. The thread is 20 emails long.
5. Clip art is used when it doesn’t need to be (and, I’m sorry, it very rarely needs to be).

Ok, before I get yelled at, I fully acknowledge that I’m certainly not perfect. Far from it. The above pet peeve I’m usually guilty of is #1.

Will you join me in making an effort to write more effective internal emails?

Here are some tips.

• Before you write a dissertation, ask yourself:
  • Do the recipients of this email *need* to know everything I’m writing here?
  • Is there somewhere else this information resides that I can link to instead of reiterating it in an email? (Intranet anyone?)
  • How can I structure this long email better to be more readable – bulleted lists, correct emphasis, breaking up paragraphs.
• If you bold more than 3 or 4 words in a paragraph, the emphasis loses all effectiveness. Only emphasize necessary words.
• Multiple colors, fonts, sizes, and styles are rarely effective.
• If the email thread is more than 5 emails long, delete the unnecessary threads. If they’re all necessary, maybe email isn’t the right place for the discussion. (Again, Intranet anyone?)
• People might disagree with me here, but, like the fonts and colors, clipart is very rarely necessary or effective in internal emails. All it does is add to the file size.

Hopefully if your university or unit has an Intranet you’re internal email is getting better. We just launched our Intranet at the beginning of this year and I think it’s already working to help the email madness.

Am I overreacting? Possibly. Maybe it’s my old journalist roots coming out. I’m not sure. I just think writing ineffective internal emails can lead to lost productivity and frustration.

… and don’t get me started on replying-to-all …

The bottom line is the online course catalog is essential for higher education websites. Creating and maintaining a *usable* course catalog is as important. So how do you know if the catalog is usable?

Before going on, I’ll mention that it’s important to not only track the form page, the *results list* page, and the details page, but it’s also a good idea to track what users are entering into the search form.

Another point I’ll mention quickly is that the search engine for our course catalog is independent of the search engine for the rest of our website. For this reason, in our analytics tool, although they are within the same report, we use a pipe (|) delimiter to denote course catalog search results (the pipe separates each input field) while the overall site search results have no pipe delimiter so we can quickly and easily separate out the two results. This technique is probably not the best, but it works well for the time being.

Hard decisions about the course catalog. A little more than a year ago, we decided that something needed to be done about our online course catalog. How did we come to this decision? We not only looked at the analytics, but we also listened to the customer. We found that:

1. We had a low “search conversion rate.”
2. Our search form was confusing and thus users were abandoning it.
3. Students and prospects were calling in with questions that could have been answered using the course catalog.

Low search conversion rate. What do we mean by that? We’ve talked before about conversions not being only for e-commerce sites. In this case, we wanted to figure out what percentage of the time users viewed the search form, searched and got to the results list page, and then clicked through to the course “detail” page?

For example:

• catalog-search-form.asp –> catalog-results-list.asp –> course-detail-page.asp

We were finding that the conversion rate was quite low and, further, the conversion rate between the form and the list results (the first step) was low. This means that a high number of users were taking a look at the search form and never actually searching for anything.

Because our search form was quite complex, our assumption was that users were getting confused when confronted with the search form.

Failed search results. Looking at failed search results is more complex. We’ve tagged our search form such that we knew the specific criteria users were inputting into the form to search. At the time, our search form had many input fields:

It was basically an advanced search, but it was our default course catalog search. Our analytics tracked not only what users entered into each field, but how many results came back. What we found was that users were entering too many criteria, thus narrowing their searches so far that no results came up.

Voice of the customer. We were also experiencing a high number of calls that were what we’d call *simple* calls. Questions that the caller could easily help themselves to on our course catalog (if they knew how to find the information).

Time for some user testing. Next, we decided to do some user testing to see if we were, in fact, correct in our assumptions taken from our analytics. If possible, it’s important to user test and not just *assume* user behavior is one way for a reason. This is another reason voice of customer is so important. Basic web analytics can tell you the what, but they cannot tell you the why. That’s where voice of customer (surveys, feedback forms, etc.) and user testing come in. We had our assumptions about what we *thought* was the reason for the poor performing course catalog, but we couldn’t be sure until we saw users struggle with our own eyes.

After testing 8 users, our assumptions were confirmed:

1. When users were faced with the search form, it took them all quite a long time to search for a course – meaning, they had to think about how to do it. They had to think hard. Obviously we didn’t get the abandonment rate that we saw in our analytics because we were forcing these users to find a certain course so they couldn’t abandon the form.
2. When users typed in criteria to search for a course, more often than not, they entered too much more criteria than was necessary (guessing at some of the criteria). Because of this, they were getting no results when, in fact, the course they were searching for was actually there.

A facelift and more analysis. After a redesign of the catalog (the form, the results list page, and the detail page), we were careful to analyize the data to see if things improved.

Before the redesign, we found from the analytics that the most used fields were the keyword search, the semester pulldown, and course abbreviation pulldown. We decreased the number of fields on the search to 4 fields (including the top 3 most popular from our analysis as well as a field for schedule number since that’s is  how our current students are used to searching for courses.

We  did include an *advanced search* so if users were looking for something more specific, the functionality was there for a very focused search.

The course catalog redesign worked. We saw signifant increase in conversion from the search to the results list to the details page. We found that the amount of criteria users are inputing has decreased dramatically and thus, users are getting results to choose from. Further, we found that users were not only getting search results, they were also clicking through to the detail page.

All-in-all it had all the elements of a successful redesign – evaluation and analysis, user testing, redesign, more evaluation and analysis, design tweaking.

When redesigning or updating your websites significantly, it’s important to not stop there, but to continue to analyize the data to make sure the new content and/or design is working.

If not, tweak and repeat.

To troubleshoot, I open up the mouse properties and click on the hardware tab thinking that I need to reset some preferences. I click “troubleshoot” and it sends be through the little troubleshooting wizard. I hate these wizards, but I want to get to the bottom of the problem so I answer the first question.

And then the usability issue is screaming at me … the next question looks like this:

Notice anything weird? The “next” button is on the left. I pushed “start over” twice! Yes, twice. The first time I pushed it and realized what I had done and made a mental note. The next time I was presented with this screen, I still hit “start over.” Mentally, I knew where the buttons were and I still hit the wrong button. I just automatically hit the “start over” button without even thinking.

Remember Steve Krug’s book, Don’t Make Me Think? This wizard is doing just that … making me think.

How about this:

Such a simple thing can make so much difference.

If you’re advancing to the next screen, the “next” button should always be on the right. Likewise, the “back” button should always be on the left. There is no reason to have a “start over” button as long as there is a “back” button (especially if the wizard is short). This is like the “reset” button on forms. Jakob Nielsen was calling to get rid of reset buttons back in 2000.

Isn’t this a no-brainer? Why do we have to fight simple usability issues like this in 2008?

I have two words for you. Card sort.

So what the heck is a card sort and why should we use it? A card sort is a usability test of sorts. It’s a technique used to help organize a website. A user is given a stack of cards with labels on them. The labels are usually navigation labels – page names (or potential page names) of the website. They don’t even have to be that granular. The cards can show topic areas instead of specific page names.

There are two types of cards sorts, open and closed.

Open card sort. This type of card sort allows the user to group the cards into buckets that makes the most sense to them.

Closed card sort. This type of card sort gives the user the predetermined group name and allows the user to then place the cards into the already existing groups that makes the most sense to them.

Why use card sorts? They’re quick, they’re easy and they’re cheap.

Example. We just finished the second phase of what I like to call a navigation re-labeling and re-organization. The first phase was user testing on the section. From that testing, we found that users found the section very confusing. What’s interesting is that users mentioned the content within the section was very useful, but they couldn’t actually find the information easily because of the link labels.

We then used a card sort to give us the first step in determining better organization of the section.

We took our existing section navigation and put each label on a card. Using the open card sort method, we gave the users the stack of cards and said to organize them in the buckets of their choice. If a label didn’t make sense, the user was told to write what they thought the label should be.

The result was eye opening. We found that many of our labels made no sense to the user. Not only that, the users grouped them differently than we originally had them grouped.

In that test we used good old-fashioned index cards. There are many web-based card sorting tools as well. Among others there are Websort and Optimal Sort. Our unit is actually using Optimal Sort on project we are working on right now.

So, if you’re thinking your site could use some re-organization or your launching a new site, try a simple card sort. It’s definitely worth the $4.29.

To tell you the truth, this is really a rant about why we should still be underlining all content links. If links *can’t* be underlined for some weird reason, the contrast should be *dramatic* between the color of the link and the color the non-linked text. Dramatic. I’m not sure why content links *couldn’t* be underlined, but I’m sure there is an exception out there (maybe your Web Style Guide says don’t underline? Even then, I think I’d defy the style guide).

Remember, “don’t make me think.” You can argue that users are more used to seeing non-underlined links now, but it still makes users *think* about what’s a link and what’s not a link.

According to Jakob Neilsen’s alertbox about visualizing links, there are two exceptions to underlining links:

… navigation menus and other lists of links. However, this is true only when the page design clearly indicates the area’s function … Users typically understand a left-hand navigation rail with a list of links on a colored background, assuming it resembles the navigation areas on most other sites.

Incidentally, also mentioned in the Alertbox … all green or red links should be underlined for colorblind users.

Anyway, I was just thinking about link styles today. Any strong opinions about underlining links?

Update: I actually just realized that my links are a dark green. The contrast should be more (even though the links are underline). So, I’ll be changing that to a more *contrasty* green.