What Social Media Analytics Includes … And Excludes

The purpose of social media is to interact with people, thus building relationships. From a business perspective, there are many reasons to do that, chiefly that people buy from people they know and like and because it opens opportunities – hence social “networking” – you’re getting the same benefits as networking, just online. (That’s the ROI on social media.)

If we manage and optimize our use of social media, we need to measure results.

Q: What should we measure?

A: The paradigm for how social analytics should measure relationships is described in The 7 Habits Of Highly Effective Teens; relationships are bank accounts whose balances change as we interact with others. By tracking our interactions, we can understand what our bank balance is.

Q: What might that look like?

A: Aldon Hynes did a good job articulating what information social media analytics should give users:

“I want to know when I’ve actually contacted someone in my online social media meta network. When did I last sent them a tweet, retweet something they wrote, write on their Facebook wall, send them a message or give them a call? Whom haven’t I contacted recently that I really should contact? Do I have other notes about them that are important? Who are their friends and relatives online? When are their birthdays? How do they fit into whatever my goals are in using social networks?

“To push it even more, I would like to add information such as whether I’ve visited or commented on their blog recently, or whether they’ve shown up on my blog, as noted by tools like dropping EntreCards, or showing up in recent readers lists like those provided by MyBlogLog or BlogCatalog.”

The following are a sample of tools currently on the market for social media analytics, which get at this kind of functionality.*

These tools measure your interactions with contacts on social media. The value in using them is that you can tell at a glance how strong your relationship is, and take action to strengthen it or ask for their help, accordingly.

Buzzstream

Provides two versions of its tool: originally a link building CRM (Customer Relationship Management tool), the company now also offers a social media CRM.

Buzzstream's contact profile view shows and allows editing of tasks, contact info, notes and history.

Both versions of the tool integrate with your email and Twitter accounts.

Under the heading “IMAP” on the settings page you can connect your email to Buzzstream. BuzzStream will then automatically associate new emails with your contact records, updated hourly. As well, it currently retrieves emails with any contacts going back 14 days.

Alternately, by BCC’ing a Buzzstream email address unique to your Buzzstream account, your email with a contact is directly imported into that contact’s history in Buzzstream.

With regards to Twitter, just add your account and your contacts’ accounts into their profiles on Buzzstream and your relationship’s tweets will automatically be imported to Buzzstream from then onwards. It should be noted that previous conversations aren’t imported.

Buzzstream History View shows tweets and email (not shown) from both parties, as opposed to just your own messages, but only from the moment of adding a contact onwards.

A related function is the auto-discovery of contact information by crawling contacts’ sites, once imported.

Finally, Buzzstream offers the ability to add notes to a contact, which means you can keep track [albeit manually] of offline interactions, phone calls and social networks not tracked by Buzzstream.

Starts at $19/month. 

Pros:

• Integrates link building metrics and buzz measurement tools.
• Shows both sides of a conversation so you tell what your contacts said and make sense of your response.
• Scrapes sites for contact information which helps build richer profiles.

Cons:

• As of now, no automated contact import from email or social media – requires more data entry.
• Limited to Twitter for now; no support for other social networks.

ConnectedHQ

It’s interesting that ConnectedHQ doesn’t describe itself as a CRM tool, but as a contact manager.

Connected HQ - Contact View shows interaction history for a quick overview of the most recent interaction.

Accordingly, ConnectedHQ brings together all your contacts into one system, integrating with Gmail, Facebook, Twitter, LinkedIn, Outlook Contacts, Mac Address Book, Google Contacts, Google Calendar, Google Voice and Evernote.

ConnectedHQ auto-imports contacts, including your history with them and also imports and displays their recent Tweets, immediately. Over the next few days, the service imports the history of your relationship going back about 18 months.

One feature I’m particularly appreciative of is that Connected sends me weekly emails (ironically called the Connected Daily) with tips on who’s changed jobs, based on LinkedIn information. This is like birthday notifications in Facebook – it enables you to make a little investment in the given relationship.

Starts at $10/month. 

Pros:

• Auto-import is a huge time saver.
• Sync between email and social media profiles is another big time saver and quite useful.
• Full contact histories.
• Integrates with multiple social networks including Facebook and LinkedIn.

Cons:

• Only shows your own messages to a contact, which is like watching a sitcom where half the dialogue is muted out.
• Doesn’t integrate SEO metrics or other tools.

Related Tools

The following tools also show you contact information while you’re logged into Gmail, which helps craft better emails and perhaps refresh your memory where the relationship is at. But they’re more neat assistants to building relationships than social media analytics tools.

Rapportive

“Rapportive shows you everything about your contacts right inside your inbox.”

You can immediately see what people look like, where they’re based, and what they do. You can establish rapport by mentioning shared interests. You can grow your network by connecting on LinkedIn, Twitter, Facebook and more. And you can record thoughts for later by leaving notes.

They also recently integrated with Aweber and have a MailChimp integration, which helps email marketers know about the person who they’re replying to, what lists they’re on, and other info that may be in your email marketing software.

Starts at free.  

CureCRM

CureCRM’s Magnify for Gmail does roughly the same. In their words:

“Need to create Salesforce Leads or Contacts from Gmail? CureCRM has a solution for Gmail and Google Apps email to show recent sales conversations and optionally create Contact entries as you send your sales emails.

“View social data from Gravatar, LinkedIn, Facebook, etc as you navigate Gmail. “View conversation history with your sales customers.”

Starts at free. 

Finally, here are three rejected possibilities for social media analytics.

1. Since FriendFeed aggregates our social media lives into one site, shouldn’t that be enough?

No, FriendFeed is sadly insufficient. While the site makes it easy to find content friends are enjoying, and thus, might be a meta-index of social media links to rival in quality the [non-social] web’s link index – FriendFeed doesn’t organize the data by contact, an essential feature if you’re to measure any given relationship.

2. Can you do-it-yourself and create your own a dashboard? Sites like DoshDosh (no longer online) have suggested organizing blogs’ RSS subscriptions into folders according to the quality of a relationship, and then interacting with a blogger accordingly.

But this moderately useful technique is limited to interacting with bloggers (as opposed to non-bloggers who tweet, facebook etc) and requires complementary or manual tracking. Creating Facebook and Twitter lists according to relationship may be helpful, but they’ll always be limited to very recent timeframes.

3. Gist from Blackberry and CoTweet from Exact Target integrate with Twitter and have been touted in some blog posts for purposes similar to ours, but they don’t let you see tweets by relationship, instead showing you a contact’s latest tweets. That’s basic functionality lacking for a social media analytics tool.

Conclusion

The field is still in its infancy, but already the tools are making big strides towards showing the current status of the relationship and doing so across multiple communication methods.

In the future, I hope that we’ll see greater use of automated contact imports, and wider integrations with older social media networks like Digg and StumbleUpon (useful for linkbaiters).

Perhaps we’ll even see tie-ins with instant messaging systems and VOIP (ex. Skype). Some status bars on overall relationship health (think PageRank bar but indicative of frequency/quality of interactions) would be very useful, as would the ability to group contacts.

From my personal experience, both companies (Buzzstream and ConnectedHQ) are very receptive to feedback and are frequently tweaking and improving their services. So if you’re interested in particular features and in helping shape the future of social media analytics, go ahead and reach out to them!

Footnote:

*I’ve deliberately ignored social media buzztracking tools because they measure brand awareness, and while brands are the collective relationship the public has with a company, my focus and interest here is in tools that measure 1-on-1 relationships, not a collective average. The tools in this article measure relationships as they’ve been built to this point, and as they’re built into the future.

Also excluded is a new sort of Facebook fanpage analytics tool, currently under development by BlitzLocal. It shows the percentage of people interacting with wall posts and related time-based metrics, as well as fans’ most common interests, which helps target further Facebook ad buys to grow the fan base even more.

The functionality is related to relationship building and does measure social media, but it’s not on a 1-to-1 scale like the above tools and provides specialized functionality placing it in a separate category.

Service description: Usabilla shows screenshots of your choosing to testers and asks them questions which they answer by clicking and/or annotating over the image.

You choose what questions to ask from a preset list and/or provide your own.

You can provide the screenshots or just input the URL and Usabilla will take the screenshots for you.

Usabilla hosts the test (it appears in an overlay).

What the service isn’t: Usabilla does not recruit users for you from a panel that they manage. Instead, they give you a link to send testers to, or offer some JavaScript code to place on your site, which will invite users to take the test.

Here’s what the marked up screenshot looked like, with points (circles) and notes (squares).

Value of insights obtained: Pretty high in many cases, perhaps 8.5/10.

Some examples:

1. The header graphic, which I took to be just ordinary branding, was seen as displaying products from Strike Models’ catalogue. Likewise, an image of the products in action lower down saw notes attached suggesting that the boats be linked to associated products.

2. The basic navigation menu above the header graphic that is standard on the Thesis theme is subtle and takes a while to notice (the average time recorded by Usabilla was 15.8 seconds. Accordingly, a few people suggested making it larger and one thought moving it under the main header graphic would be helpful.

Nevertheless, the large majority of users clicked on the menu item labeled Products when told to find product info and try to order.

3. One tester said he was a news addict and wanted more news about the model warship field. This obviously presents an opportunity to become the content authority in the field and build up an SEO advantage accordingly.

4. Another tester was interested in seeing more featured products, and several users generally commented that they would like it to be easier to find information about the products.

5. Product images and lighting could often be improved, and when they were this induced interest in finding out more about the products. Action shots in particular were really interesting to several users, since after all the point is to have the battleships duke it out on a mini-pond.

6. Remove pictures of products that aren’t carried.

Usability of Usabilla: It’s a little ironic, but there were a few stumbling blocks to making use of this tool.

First, once you upload your pictures and get started, it’s not clear that you need to check the checkboxes near the questions you want to ask. I thought I was going to ask all of them, and wanted to click the Activate button, but couldn’t when I had no questions yet selected.

Second, Usabilla provides no panel of users, which means you need to recruit people yourself. In this case, I was fortunate to be able to rely on Keri’s network of contacts, but otherwise this can be problematic.

For example, a usability client has a target demographic of women in their 50s to early 70s. I know few women that meet the description, so recruiting subjects has been a challenge.

Third, the results are displayed in a way that isn’t intuitive and instead relies on mystery meat navigation. You need to hover the mouse over the squares to read the attendant notes… but you don’t know that to begin with, let alone that squares different from circles because they represent notes.

Similarly, the full notes themselves are displayed after a large whitespace area that creates a false bottom impression, in IE7 (and IE8?).

In fairness to Usabilla, their support was very fast and helpful in clarifying things, and I was able to figure out the whole notes thing after a little initial frustration. Additionally, Usabilla support told me they’re working on redesigning the back end to make it more intuitive, so hopefully these—relatively minor— problems will be quickly resolved.

The panel question remains an issue, however.

Miscellaneous: Having data on the average speed of task completion is very valuable in some cases. For example, once Strike Models gets its ecommerce functionality implemented (and hopefully a proper redesign), I’d retest the homepage to see how fast people can get to product details. I’d guess this ties back in to bounce rate as well…

Another thing that has its pros and cons is that in contrast to many tools currently out there, Usabilla is driven by click data, not screen sharing or recording.

The end result is that you can review certain aspects faster (since you don’t need to watch video) and have a more reliable idea (assuming you recruit enough users) of what works and what doesn’t on a statistical basis. Also in contrast to video—where you’d have to fast-forward or mark particular points (which some usability tools do enable, however)- you can review particular questions’ answers together as a group, which also speeds analysis up.

Conclusion: Usabilla provides additional, valuable insights to the usability professional’s toolkit. The qualitative insights one can get are valuable and practical, and it has the advantage of being slightly faster in terms of analysis.

The main disadvantages are that you need to recruit testers yourself and you get somewhat poorer qualitative data than the full thought stream provided with audio/video recording.

Interested in more reviews? I’m writing a series of reviews on these tools, which started with a review of UserTesting.com.

You’re seeking feedback from users on how well My Blog Guest is working for them. What other means can you use to get feedback? Are there analytics metrics you focus on or read in a particular way?

I haven’t used any software for usability testing so far because (1) I am unsure which one to use because I am a usability newbie and (2) the site is very new and I haven’t yet implemented all I had planned to, so there’s no much to test with the software so far.

How did you develop the design?

Let me share how the site started.

I’ve done a lot of guest posting both to promote my own business and from behalf of my clients and I saw the potential usefulness of such a project.

However I had no idea which form [to give the site.]

First, I thought it would be a marketplace (a la Text Link Ads), where a publisher can register, pick sites he would like to provide content for and “order” content spots with them.

That seemed easy to implement but when I started writing the project description for developers I came across a few essential questions and issues:

• Guest blogging is more about networking; will this “marketplace” prevent the bloggers from making contacts?
• Guest blogging is all about placing content for free, about providing quality and getting exposure and links in exchange; will the marketplace poison this nice concept and turn the place into the buy-sell thing?

So I decided not to make it as a marketplace. But what then?

A social community? I didn’t want too much clutter—a social community could degrade into too much talking and too little action.

So after a bit of thinking, I decided to make use of a good old forum platform. I made up my mind to start with only few basic features because I had no idea what the members might need, and then [to] add features moderately in the process of the site development. (For the same reason the site is not beautified at all.)

I started with:

• The forum itself
• User profiles that included some basic “Power” calculation algorithm, to help members predict how much help in promoting their content they will be able to each other
• Tags that could reflect members’ interests and/or site topics

What plans do you have to iterate?

The features suggested by the members (and already implemented) include a “quick-start guide” after signing up (to help them understand what it is all about and what they can do after signing up), and categories to better organize new sites (and help members track them more efficiently).

It is funny that when I was starting the project, I called it “The community of guest bloggers” and listed some benefits of guest blogging.

It had never occurred to me that it might be unclear to some people what the guest blogging actually is. I got plenty of messages asking me about that—people only stopped asking me that question after I added little blue link “What is it all about” to the home page.

This short example perfectly illustrates that not all people know what you seems perfectly natural and widely-known to you, and that every single little detail matters a lot.

This is why beta testing is so important and this is why I plan to do a lot of it.

Have you viewed any users trying to use the site? If not, what do you think you’ll see when you get around to it? If you have seen them, what’s on your to-do list to make the site more usable?

All I currently use is Google’s content drill down (and also goal conversion to track registration but that’s more about promotion than usability).

I did my best to create clear user paths to avoid any confusion:

1. Registration
2. Quick start guide
3. Add social media accounts
4. Add your tags
5. Add your sites
6. Enter forums

I do hope this all worked out well.

Did any other online communities serve as inspiration for creating the site? If so, what usability / member retention lessons did you learn from them?

I tried to use create something between a forum and a marketplace (with enhanced searching and tracking options), but I guess people view it rather as a forum and use the forum feature most of all.

This drives me to another important conclusion that when you are creating something new, try adding some (or many) features that are well-known and widely-used in the given sector. People will appreciate that and will feel more at home.

Do you see any overlap between what you’re doing to SEO the site and usability?

Currently the site is not SEOed at all. I don’t rely on Google for getting the word out. I don’t thing “guest blogging” is searched much anyway. What I rely on is word-of-mouth.

But to answer your question, I don’t think this overlap should occur [because I forced it] anyway. If it does need to be forced, I would be doing a poor job [since SEO and usability mesh naturally].

Thanks Ann. Note: You can read my review and tips on using predictive heatmapping tool Feng-GUI to help identify what users are looking at and iterate the layout and graphics.

I began my search at Shopping.com, the big affiliate datafeed aggregator and comparison shopping site. Following that, I typed “laptops” into Google to shop some Canadian laptop sites, assuming correctly that I’d get mostly big Canadian retail names. I ended up browsing the following sites.

How did I evaluate each of these sites?

I rated some of the sites on all the following usability metrics, while blitzing through a few because I was running low on patience by the end of this comparison.

• Onsite search quality: The relevance of onsite search results.
• Filters: The variety of filters available, their usefulness and whether they’re cumulative (e.g. can be used together) or exclusive (only one filter allowed at a time).
• Message matching: Whether the site’s claims on one page are fulfilled when you click through to the next. I don’t score this since it’s just a negative goodwill factor—it’s expected, not a “bonus” for customers.
• Other “goodwill” factors: Steve Krug’s “goodwill tank” paradigm says that visitors have a measure of goodwill that can be increased or depleted. Once depleted, visitors leave.
• Selection: How accurately needs are met—is it a “least worst” option or exactly right?

The metrics are listed in the order they affected my research and my perception of how others prioritize them, too. I’ve bolded factors that stood out.

Shopping.com

Onsite search: 2/10. Terrible. Searching for a laptop model-name yielded car parts, and looking for a “16′ HP laptop,” or “16 inch HP laptop” yielded mainly accessories.

Filters: 10/10. Excellent. Shopping.com provided me with useful options to filter by specs (screen size, hard drive space, price etc.) and the filters can be used together (e.g. 16 inch and 250 GB+ hard drive).

Additionally, these filters drive the site’s title tags, URLs and page heading, all of which boost SEO and message match.

Note: Having great refinement options does not replace quality search results because many visitors just want to search, not click.

Message matching: Shopping.com delivers on its promises. ‘Nuf said.

Other goodwill factors: 7/10. The site features reviews, and the calls-to-action are easily found when I’m ready to buy. However, it doesn’t show whether the merchant has it in stock, which data you’d hope a mega-affiliate like this would have access to.

Selection: 10/10. Even after drilling down through multiple refinement options, Shopping.com showed me dozens of products.

FutureShop.ca

Onsite search: 0/10. The same three queries as above returned no results. This is despite having the model listed on the site (it was the one my dad wanted) and a selection of 16 inch laptops.

Filters: 5/10. You can filter laptops by screen size, refurbished origin, or “tablet/specialty.” Those are useful options, albeit limited in variety. Additionally, the filters aren’t cumulative, so you can see either 16 inch laptops or refurbished laptops… but not refurbished 16 inchers.

Message matching: Covered.

Other goodwill factors: 6/10 In-stock vs. sold-out status: FutureShop.ca handles this poorly (they get a 6 instead of a 7 because they have direct access to inventory data, unlike affiliates). Category pages display out-of-stock products without telling you they’re out of stock, so you waste time going to detail pages where the information doesn’t stand out. Because it doesn’t stand out, you waste time looking for the info or for the “add to cart” button if you assume that the product is in stock.

On the other hand, FutureShop does feature breadcrumb navigation that helps you make sure you’re in the right place, and for in-stock products, the calls to action are prominent.

Selection: 7/10. If you’re content with browsing according to limited refinement options, FutureShop has a fair depth of laptops within the various sizes available. On the other hand, their refurbished and tablet laptop selection is negligible.

TigerDirect.ca

Onsite search: 11/10. TigerDirect takes extra steps to go the mile! Not only are their search results relevant, they also offer you filter links above the results that match popular filtering options such as brand and price. The only drawback is that the results are below the fold due to a large site-wide banner.

Filters: 10/10. The retailer offers a deep variety of useful filters, and they are cumulative.

Message match: If I scored this, I’d dock (just) a couple of points from TigerDirect for making me scroll to see my search results. Other users might not know where they are. The refinement options are above the fold, so it’s not the end of the world.

Other goodwill factors: N/A . I did notice a Hacker Safe logo, which typically helps sites convert better by reassuring visitors, but I’m personally indifferent.

Selection: 5/10. While TigerDirect has a great variety of sorting options, their inventory is low relative to Shopping.com. You can argue that’s normal since Shopping.com is an affiliate getting feeds from multiple merchants, but this argument cuts both ways since some advantages are inherent to being a merchant (e.g. greater likelihood to get reviews posted).

HP.com/Canada/

Onsite search: 0/10. The results page lacked any reference to 16 inch laptops, and the first three categories it sent me to (those visible above the fold) lacked any such models. Additionally, my eye was first drawn towards the green-highlighted “Recommend Links,” making me think I wasn’t looking at search results.

Filters: 0/10. What filters? HP’s idea of a filter is to let you pick between a variety of notebook PCs “helpfully” branded as Compaq Presario, HP Home, Pavilion, Touchsmart of HP HDX. As an outsider to HP, these are meaningless to me. The rest of the site doesn’t make things any easier.

Message match: As with the search and filtering functions, this isn’t intuitive. After deciding to try some of HP’s jargony category links, I thought they must have been broken because they don’t take you to a new page. It turned out that they changed the main rectangle above them—which they’re not related to by visual hierarchy—so I didn’t notice what happened until I looked around. My eyes were focused on the selector links below the main rectangle, which is why I didn’t notice and had to think, to paraphrase Steve Krug…

Similarly, the dark colored links that appear are virtually indistinguishable from ordinary text. So unless you rollover, it’s unclear where to click for the next step.

Other goodwill factors: If the above weren’t enough, HP commits two more sins. First, clicking one of the marketese category links takes me to a page featuring a laptop picture and two calls to action (compare these products and HP Pavilion Home Network PCs). It’s not at all clear why I’m not already viewing their selection of laptops, so I click HP Pavilion Home Network PCs again (the smaller green area, not the title/banner higher up):

To reward me for clicking, HP serves me an unholy avalanche of information overload. Notice the horizontal and vertical scrollbars?

Selection: 5/10. The selection is wide, but as we’ve just seen, it’s not filterable or accessible in a format that allows me to make my decision easily.

LaptopCloseout.ca

This site is remarkably fast. It took less than 60 seconds to show me the door! Why?

Calling the prominent support number drew an automated message that the inbox was full. Also, the warranty page refers to various conflicting periods such that it sounds like there’s no warranty despite the looks. Consider:

• “If you’re not satisfied, simply return the laptop in its original condition within 3 days of receipt…” I have 3 days to return it if I’m not satisfied? That’s all?
• “All laptops sold through Laptopcloseout.com are entitled to laptop closeout 1 to 6 months days full warranty…” Huh?

Lenovo.com/CA/

With the exception of a gibberish 156-character URL that would reduce CTR from search results, Lenovo scored very highly.

Onsite search: 7/10. Searching for a Lenovo model (the S10e) returns it #3 behind some support results for it, which isn’t perfect but still good enough. And a search for a 15″ laptop returned numerous models (they have no 16 inchers from what I could tell). The only letdown was a search for a “15 inch laptop” only returning 1 result.

Filters: 10/10. With numerous filters, including some particularly valuable ones for holiday shopping (ships within X# weeks), which are cumulative, Lenovo makes it easy to find what you want.

Message match: Taken care of.

Other goodwill factors: 8/10. Above the fold, the intro paragraph to each laptop line clarifies which type of customer each laptop line is intended for.

And the product-line comparison chart they offer below the fold is extremely valuable to methodical shoppers who want more details for comparing their choices. On a slightly odd note, their search results include many IBM.com pages, whose presence won’t make sense to anyone unaware that Lenovo bought out IBM’s Thinkpad laptop line.

Selection: 7/10. While it’s again narrower as compared to Shopping.com, Lenovo is a manufacturer, so this is to be expected.

MDG.ca

We regularly receive MDG’s flyers at home, so I was curious to see how their site would perform. The performance was laughable, since there was such poor message match across the site’s pages.

On one page, MDG offers to recommend the right laptop according to your lifestyle/usage.

On the next, instead of recommendations, I get a credit application. No thanks – that’s not what I asked for.

Conclusion

The user experience offered to a potential visitor can’t be determined solely on the basis of whether a site is an affiliate site or a merchant, regardless of the brand size. HP is a merchant and the best known brand of any of these companies, and their performance was the worst. Lenovo is also a manufacturer and retailer, but their performance was highly impressive, just as Shopping.com’s was.

LaptopCloseout, MDG and FutureShop showed that you can be small, medium or large retailer and still leave users scratching their heads.

Perhaps those trust metrics from Google’s Vince update reflected how quickly visitors click through from Google or other “user happiness” metrics (Google’s term), but they don’t necessarily correlate with user experience metrics post-click.

This is called predictive heatmapping. So how does it work and what good is it to you? The answers to that plus some real case studies, after the jump.

What is an attention heatmap?

An attention heatmap is a combination of two elements: eye-gazing, and predicted attention.

Eye-gazing simulates the sequence of extremely rapid and involuntary eye movements (“saccades”) that happen as your eye scans a scene. This is overlaid on a heatmap of the attention represented by different colors which predicts where the brain will focus. Hotter areas indicate a more intense focus, while cooler areas show a lower level of awareness and importance.

What factors are considered in generating a predictive heatmap?

As Feng-GUI’s neat “how we do it” page explains, their algorithms are based on recurring themes detected via eye-tracking. These are things like color contrast, orientation of an object, etc.

How can it be used?

The description continues:

“It can be used to refine landing page designs for existing or new pages, and to improve conversion rates. It can identify which page elements are being looked at and which are being ignored. This allows the designer to focus attention on the correct parts of the page and increase the likelihood of conversion. Attention heat maps can be created several times during the design or re-design process to ensure that the refinements are having the intended effect.

“A ‘busy’ eye-gaze path and scattered heatmap with many hotspots is usually an indication that visual priorities of the page are not clear, and will likely result in confusion and a higher bounce rate for your landing page. Relatively simple eye gaze paths and a small number of clear hot spots (centered on the desired conversion-related areas) are a good indication that the page will be more effective.”

Another way these can be used is in improving the effectiveness of print ads. My friends in the Community Law club at McGill Law are putting on an event dealing with depression in law students. Here’s their ad:

And here’s the heatmap:

Notice how little attention goes to the upper right call to action? When I first saw this posted on a bulletin board, I didn’t notice it at all. I thought it was strange someone would just post a random fact, so I looked around and only then did I notice the call to action.

A better ad would have probably made the call to action larger and placed it somewhere below the main copy.

Indeed, this bottom-left to forward-right layout, which I’ll call the forward slash layout ( / ), is very confusing to Western readers, because we’re accustomed to reading from left to right and top to bottom. As my fellow fans of Gord Hotchkiss and Enquiro B2b SEM and usability know, study after study has confirmed this pattern.

Now let’s see these heatmaps in practical web action. Let’s see what we can use them for!

Some sample heatmaps

Suppose you’re SpyFu. And you’re curious to find out your competitors’ conversion rates—for example, AdGooroo. Here’s AdGooroo’s landing page.

What jumps out at you? Personally, the toolbox in the middle grabbed and held my attention, which is bad news for the copy. Feng-GUI seems to agree:

So if you’re SpyFu, you might think that you’re laughing all the way to the bank. The problem is SpyFu’s landing page…

… isn’t so hot either.

The main call to action above the fold looks like some support link because of its placement near the login link, and the areas of the page that are emphasized are secondary (Live Help) and a fake button (Top Secret) that isn’t actually clickable!

For people considering media buys, and what creative to use, I have the [untested] belief that Feng-GUI can predict the best creative. How?

Have the graphic artists mock up a screenshot with your various banners in place. Then put each screenshot through Feng-GUI to see what draws the most eyes, and you might have found the ad that will draw the greatest CTR for you.

As an aside, I’d like to point out that the faces in the heatmap aren’t highlighted, yet eye-tracking reveals that faces tend to draw our eyes. Feng-GUI has an option to let it know there are faces in the image, and it does slightly better with this feature enabled. But nonetheless, the resulting heatmap isn’t great. So be aware that the tool isn’t 100% accurate; Feng-GUI itself only claims to accurately assess 75% of what actual eye-tracking would capture.

Ironically, Feng-GUI doesn’t seem to have used its predictive mapping on its own site. When I first tried buying credits for Feng-GUI, I went crazy trying to find the “buy now” button. Beside the fact that “buy now” is not in the main navigation, it’s not obvious on the products page either. Looks like I’m not alone in my frustration—I used the tool on the page and ended up with this result:

The order button is one of those small, unimposing white jumplinks at the top of the page. Exactly in that negligible row of links above the active window that gets no attention. And because Feng-GUI couldn’t be bothered to add a simple buy now button with Paypal, you need to email them to request to buy. Only then will they send you a Paypal money request.

By the way, there is a competing predictive heatmapping tool, Attention Wizard, currently in beta. I’m not spending time reviewing their tool because it currently only allows you to generate one heatmap a day.

My friend and occasional client Michael runs Greek For Me, an apparel store for Greek fraternities and sororities. He recently asked me for usability consulting to help him increase his conversion rate. I thought I’d share this mini-usability review to help Mike and other store owners who may be struggling with these issues.

Let’s check out Greek For Me’s Alpha Xi Delta Hoody detail page.

(Click to enlarge)

The page gets the general info down fine. It obviously matches the keywords likely to deliver visitors, and like the rest of the site, there’s shopping cart info in the top right and breadcrumb navigation to orient visitors.

Where this product detail page drops the ball is with the drop-down menus used to select size and colors. These overwhelm visitors and make them anxious.

Suppose you’re curious to see the sweatshirt in other colors than gray. If you pick any color, the product picture doesn’t update to reflect your choice. And you can’t click some secondary pictures of the item for different views, because there’s only the one picture.

The same problem repeats with the dropdown boxes for the foreground color and border color. You can’t tell what they will look like on the sweatshirt, much less what they will look like in combination with each other.

And this is compounded by using jargon color names, like vegas gold, old gold and light gold. What if I just want gold? Besides, who goes shopping, thinking “I want a sweatshirt with vegas gold lettering!”

Similarly, if you pick a size, you might have a general idea what that entails, but it’s by no means certain. For example, SEOmoz gave me a tshirt last year that was a size medium. I thought it would fit as I’m about a medium build, but it turned out that I needed a large, so I reluctantly gave the shirt to the girl I was dating at the time. Which might explain why we broke up and I still don’t have a girlfriend. But I digress…

To reiterate, product detail page dropdown menus make visitors anxious. They create a fear of getting the wrong size product with the wrong colors. We as humans fear loss more than we seek gains. So these fears of wasting money on the wrong product reduce purchase momentum (kudos to the Bryan Eisenberg for the momentum metaphor.)

Possible solutions to test:

• By far the easiest solution is to offer no alternative colors. By making the color question a simple yes-or-no decision, momentum is a lot easier to maintain.
• A better solution is to offer a very limited range of popular colors. You could probably copy The Gap and go with blue, pink, gray, red and black. This avoids leaving money on the table in the case of people thinking, “No, I don’t like the default color.”
• Use only one foreground-border pairing for each background color. If you get the blue background sweatshirt, your lettering is white-and-gray, period.
• Add pictures of the product in the alternative available colors.
• Have some sorority girls model the product, and explain what size they’re wearing. Tests typically show that actual-use pictures convert better.
• Create a customization tool that dynamically alters the product image as people select different options. This is probably the most expensive solution, and would likely need to have its own section on the site.

In fairness to Mike, I should point out that he has a clickable link to see the colors involved, but these just show a chart with various swatches of color. And while that might be enough for some people, I only barely noticed it after already writing most of this article.

So it’s skimpy information, presented in a muted part of the page.

Shipping questions for detail pages

Two common questions visitors have are:

• When will the product arrive? (Sometimes phrased as, “When will it ship”)
• What will the price of shipping be.

The product arrival date info is automatically estimated, which is a great piece of functionality. Unfortunately, this too is hidden in the discreet “Additional Information” box below the product image.

As to the price of shipping, this is nowhere to be found on this detail page or any others.

The site tries to solve the problem with a shipping price calculator that appears on the next page, below the cart information, when you click add to cart.

(Click to enlarge.)

The catch is that the visual design of the page emphasizes the checkout buttons. They strongly contrast with the rest of the page by their shape and color (see more on buttons and layout in my friend Sandra Niehaus’ great article on button balance and contrast).

Normally this emphasis on the checkout is good, but in this case it will create a lot of scenarios like this:

• Add to cart
• Check cart info
• Continue to checkout
• [Miss the shipping calculator.]

Then when people move on to the billing page, the ‘Standard’ and ‘Rush’ shipping options don’t provide any more info on price.

So what?

So the net effect of this lack of information on shipping times and rates creates anxiety. Again, this slows momentum towards conversion.

Possible solutions

• Embed a simpler calculator in a reasonably prominent part of the product detail page. For example, some of the whitespace on the right hand side could be used without affecting how clean the page looks. Of course, that’s just a hunch – you’d have to test that to know for sure.
• Since most products have a standard weight and size, Mike could use USPS’ “If it fits, it ships” product and just automatically list shipping rates on his product detail page according to product type.

The fundamental role of a product detail page is to decrease anxiety by spelling out clearly what the product offer is. It should offer enough information to answer visitors’ questions, without overwhelming them and making them bounce.

Summary sections

Wikipedia offers summaries at the top of its feature-length articles. While I haven’t got access to Wikipedia’s visitor data, I’d hypothesize that these summaries help reduce the bounce rate (bounce rate is the percentage of people who arrive and “bounce away” within 10 seconds or so). Let’s think about the factors that play in bounce rate to understand how summaries can help.

I just finished reading the book Honest Seduction: Using Post Click Marketing To Turn Landing Pages Into Game Changers, by my friends at Ion Interactive. One key takeaway of their book is that bounce rates are partly a function of visitors’ mental time commitment.

Someone who clicks an organic listing just made a five second commitment to click. If your page displays a mass of text, you are asking for a five minute commitment. Many visitors will bounce due to the disconnect between their initial five second commitment and your subsequent request for five minutes of their time.

By featuring a summary on your page, you invite users to make just an incremental five or 10 second commitment. By bridging the gap in this way, the summary prevents a commitment disconnect.

Table of contents sections

A table of contents section is another valuable item for users that are in a hurry. While I have to admit I haven’t tested it, I’m willing to bet that these tables reduce bounce rates by boosting message match.

Besides Honest Seduction, I’ve also been reading the Bryan and Jeff Eisenberg’s books on search marketing, and I recently also finished Tim Ash’s Landing Page Optimization. One common thread amongst all these conversion rate experts’ advice is that message match is essential.

Message match is the continuity and total coherence between a click source and the page(s) it leads to.

For example, an organic listing titled “Browse Used Toyotas Starting At $5,000″ should lead visitors to a page with that headline and a number of used Toyotas with prices starting at $5,000, not $6,000. If it’s not immediately clear that your page matches the organic listing/banner ad/PPC copy/email message, then people will leave in droves.

The key is to make things obvious

The problem is that with long textual articles you can’t have everything above the fold. So if a visitor came for something that is below the fold, there’s a good chance that they will bounce.

Now, summaries are a helpful way to address the problem, but they’re only a partial solution. Summaries typically follow the flow of an article, so a scanner with a 7 second attention span might miss the second to last line of the summary. So he might not know from the summary that the article contains what he wants.

Enter the table of contents. This hyperlinked group of keywords is an easily scannable information desk, like the ones you find in malls. It enables rushed visitors to get a better idea of what lies below the fold and deep in the heart of any article.

Thus a table of contents is a tool to reinforce message match on long tail keywords and content that lies below the fold.

In the absence of a table of contents, I’d venture a guess that people may start reading at the introduction (or summary). But if they’re not captivated quickly, they probably won’t dig down in the hopes of something more interesting showing up.

With a table of contents, even if the introduction fails to grab readers, the table of contents may lead them further into the text.

Copious inline navigation

This is another way of lifting time on site and pageviews, as I’ve seen in my own personal experience. In most cases, greater time on site and more pageviews correlate to greater engagement and likelihood to convert. Additionally, this obviously ends up being very valuable for SEO.

Wikipedia is well known for pioneering the linking of keywords throughout its articles to whatever Wikipedia page existed on the topic of that keyword. Its users appear to largely appreciate this navigation, as I learned from my friend Rachel.

While chatting one day at one of my school’s computer labs, Rachel told me that she loved reading Wikipedia, for entertainment. “I could spend hours and hours reading Wikipedia,” she told me.

Obviously, she and other users browse Wikipedia articles and click from one to another. Heck, I’ve done it myself while reading about famous privateers/pirates in colonial times.

While broad general navigation is good, Rachel-style use of Wikipedia is only possible due to the simple system of internal linking that Wikipedia has created. When editing a Wikipedia page, you just need to put double square brackets around a word (i.e. [[word]] ), and it will then automatically link to Wikipedia’s page on the topic.

Nowadays, major newspapers like the New York Times have adopted the practice, and it gains more currency amongst the WordPress community daily. WordPress bloggers have downloaded my Internal link building plugin, which automates Wikipedia-style inline navigation, over 7,300 times from my own site since it was released a year ago. And many more downloads have occurred of translated versions hosted by East European SEOs.

So consider using in-line navigation to engage users longer and make more money.

Competitive intelligence doesn’t just come from analyzing competitors’ backlinks. By analyzing some of the web’s most popular sites, it’s possible to discover usability techniques that can be easily copied over to lower bounce rates and lift time on site. The conversions should follow.