Intimidated by the slick moves of the new GA Motion Charts? Gaining insight was hard enough with a stable graph, you might be telling yourself – how can I find the trends in one that changes before my eyes? These quick tips will help you use the charts both within a larger analysis process, as well as offer insight on some of the trickier Motion Charts’ options.

The great thing about Motion Charts is that they allow for several distinct types of analysis: higher-level identification of trends, as well as exploratory analysis aimed at finding “buried treasure”: the less obvious trends, or the unexpected spikes in traffic or value for a dimension that lead to new marketing tactics.

We’ll start with some tips for using the Motion Charts at a higher level. Let’s say you are a Marketing Director, with no analyst in the department to make sense of the charts for you (a not-uncommon phenomena!) What essentials do you need before clicking the ‘Visualize’ button?

First off, its important to realize the distinction between the Motion Charts as compared to the more standard line, pie, and bar charts. In contrast to “bins” of data represented in a bar or pie chart segment, Motion Charts uses unconnected points to represent the values for various dimensions (Keywords, Traffic Sources), and metrics (Time on Page, Goal Conversion Rate) within the chart. Each point is one value for a particular time step. This means that some of the basic guidelines required of analyzing point pattern data become important.

First of these guidelines? Learning how to tune out random variation, or those spikes in the data that don’t suggest larger trends, is key. For example, here is one view of keyword data, followed by the view of the graph immediately following:

Notice how in the second shot, which represents just one day later, the bubble representing “internet marketing company” has shifted to an entirely new position? Focusing on this single jump takes you further from identifying the big trends at work. If it’s higher-level, comprehensive trends you seek, get used to tuning out these jumps, and focusing instead on bubbles that hover in the same general regions of the graph.

By default, the GA Dashboard segments the time period into days. But when looking for big trends its appropriate to change the time so that values are averaged over the week (Select the middle button next to the “Graph by” label. Doing this will make it easier to filter out the variation you don’t need to worry about.

Now for some analyst-level tips related to one of the features of the Motion Charts that might be less intuitive. Let’s start with an example. Here are two shots from the same graph at different points in time. Again, the data is keywords:

Notice how in the first shot, all of the values hover in a line above the X-axis? This formation continues for most of the period, except for a single day in which the bubble for one keyword, “pure visibility” moves to the very top. In the last example this was exactly the variation we wanted to tune out. This time, we’ll take notice of the spike (and perhaps look the traffic sources, visitor regions, or other dimensions for the date in question).

But what about all the other keywords? Should they be ignored? When doing a deep-dive analysis, there may be slighter, yet still noteworthy insight within their values. For instance, what if there is a slight trend toward higher or lower values for for variations containing a single keyterm? To a paid search analyst controlling a large budget, this sort of insight could suggest more (or less) use of the term in paid search marketing.

There are two ways to find this subtler differences, when the graph’s scale makes it difficult at first view. The first involves changing the scale from “Lin” (linear) to “Log” (logarithmic):

This above example is a time when changing the scale will allow you to focus on the more granular differences between the other words. A logarithmic scale is appropriate whenever the range of the values being graphed includes both very large and very small numbers – and those few large or small numbers make other values look virtually identical.

The second way to scale the data so that the big trends (those bubbles representing much higher values) is to use the search bar at the bottom of the grid to filter out the keywords or other items (Traffic Sources, etc) that are throwing the scale off. This example uses a regular expression to exclude the term “pure visibility”, both with and without the space between words:

Happy graphing with Google Analytics Motion Charts!

What is Google thinking with its new SearchWiki product?

Google launched SearchWiki on November 11th, stirring considerable controversy.  There have been complaints over the inability to opt out of SearchWiki and the appearance of the product and concerns over privacy and what influence random strangers may have over search results.

With SearchWiki, Google has delivered a perplexing user experience. How do you know whether a search result was relevant or not before visiting the site? While the ability to remove search results makes sense when a user is looking for information, what incentive do users have to go back to their search results and vote up or comment on a search result after they’ve found the site they want?
Read More

Have you ever wanted to be systematic about testing different versions of a popular page on your site, but felt overwhelmed by questions of how long a test should last, which traffic channels should be sent to the pages being tested, and how certain your results are? Landing Page Optimization by Tim Ash is the most comprehensive book available on test design, analysis, and the finer points of both.

The book’s clarity is its strongest feature. The risks, advantages, and disadvantages of A/B versus multivariate tests are explicitly defined, and examples that put success in terms of revenue are used throughout the book.

Another memorable aspect is Tim’s observation that both the visits to the site, and the element variations being tested on a page (for example, the font, or a graphic, or any other aspect of the page design) are are independent of one another in effects. This is something we’ve discussed here at Pure Visibility on multiple analytics projects. For example, you may have a page for which you are testing two background colors, and two very different button designs. It may be very well be the case that the color and the button design displayed effect each other, either positively (increasing the conversion rate) or negatively (canceling each other’s positive effects). In this simple scenario, it is relatively easy to add tests for each pairwise combination of elements, to determine the truly winning “recipe”. However, in a more complex test, separate tests of each combination increase the time and data needed for the test. Sometimes, independence must be assumed to some degree. Kudos to Tim for addressing this in a book anyone with basic stats knowledge can understand.

Backstory: On November 11th, the Wall Street Journal has an article that says Facebook still isn’t selling ads like MySpace, despite surpassing their market share. It’s interesting, and compares formats of online advertising that are now considered traditional with more novel ideas like Facebook’s engagement ads.

Wired and cnet were, of course, among those to respond to the the WSJ article. All three articles quoted industry & media research people, like eMarketer, Forrester, etc. But here’s the part I want to talk about…Wired takes a quote from the AdAge interview with Ted McConnell, who heads up Proctor & Gamble’s interactive marketing and innovation operations.

“What in heaven’s name made you think you could monetize the real estate in which somebody is breaking up with their girlfriend?”

The piece on McConnell is wonderful, read it; it shows that he’s very conscious of the potential value of demographic targeting on sites with user generated content. But he identifies as a consumer advocate, and suggests that the propriety of monetizing media is dependent on who creates it and why. It bears consideration; I’m really interested in how to do online marketing in ways that satisfy both consumers and advertisers. I feel that way because advertising is core to the business model of many free web-based apps that I’m crazy about.

So, take McConnell’s example. How, in heaven’s name could you think about monetizing the real estate where someone’s breaking up with his girlfriend.

A couple ideas:
Puffs (P&G tissue brand) a sad story contest. prize = a vacation for one.
iTunes or Amazon solicits user created breakup mixes

What do you think? Worst / most heartless ideas you’ve ever heard? Am I awful for thinking that would be funny? You see any fun opportunities given this difficult hypothetical?

WSJ : Facebook Tries to Woo Marketers
cnet news : MySpace beating Facebook on ads? Well, duh
Advertising Age : P&G Digital Guru Not Sure Marketers Belong on Facebook
Wired : Facebook Ads Not Dead Yet

We have always argued that Google Analytics is an Enterprise-level application, but with the latest release, the tool has a muscular look and feel that puts it solidly in the Enterprise space, along with long-term giants like Omniture. We recently returned from the Google Analytics Authorized Consultant (GAAC) conference in Mountain View, and it’s clear that Google is committed to this tool and its long-term possibilities.

Now that it’s arrived, how should it be used? Never has such a powerful analytics tool been available for free. But does that make it different? What challenges do companies face when trying to install it? How should it be marketed? In the next couple of weeks we’ll explore these issues and possibilities.

Some authors have already written at length about the new implementation, including Eric Peterson. We recommend you start there while considering the possibilities of Google Analytics for your company, whether you’re a marketing firm trying to improve your web marketing offerings or a company trying to make sense of what’s happening to their website online.

Google Insights for Search is an exciting tool and if you haven’t had the chance to check it out yet, you ought to. This tool lets you research what people search for and when they have done it. It’s the sort of tool that can provide hours of entertainment to people interested in Internet marketing.

Certain parts of the user interface, however, present usability problems. It all comes down to an relying too heavily on the dropdown menu when other controls would work better.
Read More

At Tuesday’s marketing roundtable at Spark Central in Ann Arbor the panel had a very interesting discussion about the future of mobile marketing. They talked a lot about geo-targeted ads that are sent to a potential customer’s cell phone who has opted in to receive such message.

This got me wondering how paid search will work on mobile devices. One option is Google Mobile ads which are designed to appear on mobile websites but what about the growing number of iPhone users that don’t access mobile websites rather they view standard HTML? We can assume they will see paid search ads the same way we view them on a desktop computer. The next question is how many mobile internet users use search engines. A study done by iCrossing showed that 75% of mobile web users conduct searches, which in theory means traditional paid search ads will display 75% of the time mobile web users get online. This is a big number so I think it’s important that paid ads are displayed accurately.

Is a 3” screen is really large enough to be effective with paid ads? First of all I would think you need to zoom in to read any of the search result listings on a cell phone. Regardless of legibility the first 3 paid ads above the organic listings are still be prime real estate but once zoomed in I would think the paid ads along the right hand side wouldn’t be visible. This got me to wondering if cell phones are truly going to be the only way we get on the internet are we really going to settle for that small screen?

At Pure Visibility we work off mobile laptops but when we get in to work we typically plug in out computer to bigger monitors to make it easier to read in addition to keyboards and mice. I am a believer that in the near future both desktop and laptop computers as we know them are going to be gone and replaced by powerful mobile sized computers. If you think about it we are already there. Apple’s iphone, Google’s android, and Blackberrys aren’t just cell phones, they are computers. Once a cell phone/computer is as powerful as my laptop why would I lug around my laptop to and from work and travel? I could just carry the mobile computer and while at work plug it into a bigger monitor with keyboard and mouse. I could have the same set up at home so that I could work quickly and efficiently there too. This would also solve the problem of the screen size and paid search advertisements when I’m working from a stationary place. Some people are speculating that our mobile computers will have screens in front of our eyes that track our eye movement and allow us to control the computer. Sounds too far out for me, but I really like the idea of my work computer being able to fit in my pocket because it means commuting by bicycle is that much easier. My prediction is that paid ads won’t be as effective for people searching on the go, but as far as a mobile internet and a standard internet I think we are heading to standard internet and this idea of a different mobile web will become irrelevant.

Can you still provide “high touch” service through an ecommerce site? Yes.

Touch still happens in ecommerce and can be an important factor in differentiating yourself from your competition. A well-crafted user experience is the touch. Anyone can throw together a website that sells widgets. Crafting an effective and helpful website to sell widgets and combining it with stellar customers service at every point of interaction takes effort and is going to appeal to those with the money to buy a better experience.
Read More

LinkedIn has added a great new feature to help people network!  Last week, LinkedIn added Events, a feature which lets users list events and other users to indicate whether they are going or if they are interested in going. This feature isn’t groundbreaking, but the value of having it in LinkedIn is that it is now easier to promote an event to your LinkedIn connections.

When I found out about this on Friday, I was excited. Within a couple of hours, I created a listing for the World Usability Day event I’m planning here in Ann Arbor. I thought I was exploring uncharted territory. You can imagine my surprise when I saw several listings for upcoming Pure Visibility classes. “What’s up with that?” I said.
Read More

Google Analytics provides powerful aggregate data about general groups or categories, but companies, particularly when targeted a key smaller market demographic, may want data at the level of the individual.

Unfortunately, Google Analytics tools is not designed with this function in mind, for privacy, technical, and philosophical reasons. While we generally agree with this setup, sometimes you just need to get data on individuals. In those situations there are some strategies that you can use to get the individualized data integrated into Google Analytics. Here are four we use a lot:

Read More