Google Analytics collects information about your website visitors that pertains to their experiences on the website. This information includes:

None of this information, with the glaring exception of the visitor's IP address, can be directly tied back to an individual user. Even the custom variables are collected and displayed anonymously and aggregately, so that any individual user's information cannot be found in Google Analytics. The IP address that's collected is merely used for processing and used against any filters that a profile owner may have set up — you won't find any IP addresses in Google Analytics.

Depending on the web-analytics platform, you may see slightly different information in the reports. For example, in Yahoo! Web Analytics you can see demographic data such as the visitor's age range and household income. Other platforms do show you the visitor's IP address, but have clear statements in their terms of service that prohibit users from taking those IP addresses and using them for their own efforts.

Google Analytics does not collect any personally identifiable information, and does not display any personally identifiable information in reports. Google Analytics also does not allow anyone to rig the Google Analytics data collection process in any way to collect information that may be submitted on a website via a form or another method. Since this is such a sensitive topic in web analytics, I encourage you to at least be aware of the terms and conditions, and even the privacy policy, of your web-analytics vendor. If that vendor is Google Analytics, you can visit http://www.google.com/intl/en/analytics/tos.html and be completely caught up.

Google Analytics does not collect or display in reports any of the following information:

Most if not all web-analytics vendors follow the same guidelines. When integra-tion with a customer relationship management (CRM) software like SalesForce is present, this type of data is still not available from Google Analytics — only in SalesForce or your CRM tool of choice.

When you read http://www.google.com/intl/en/analytics/tos.html, you'll come across point number 7, which deals with privacy. This single para-graph outlines everything I just talked about under this subheading, but it also includes an interesting final sentence: "You must post a privacy policy and that policy must provide notice of your use of a cookie that collected anonymous traffic data."

Some websites do this already, but most don't inform visitors about what web-analytics programs they're using or what tracking methodologies they employ. I recommend finding a way to inform your users that you are collecting their usage statistics (but not their personally identifiable information).

I mentioned in the beginning of this chapter that Google Analytics sets first-party cookies on a visitor's computer. It uses only first-party cookies, and no third-party cookies.

A first-party cookie is a cookie that can be read only by the website that sets it. If you visit metallica.com as I did back in Figure 8-1, and Google Analytics is installed there, then only metallica.com can open up those cookies, read them, and update them when necessary. First-party cookies are safe and secure, and cannot be sold or referred to other websites for any purpose. Most web-analytics vendors use them.

A third-party cookie is a cookie set by a website other than the one you're visit-ing. Let's say that you visit aol.com and there is an image ad running on that site for brownshoes.com. Let's also say that aol.com allows third-party cook-ies to be set. The brownshoes.com site is therefore able to set and read cookies on your site, even though you didn't visit brownshoes.com. This means that aol.com is allowing your sensitive web information to be read and accessed by another website. For security reasons alone, this is a dangerous practice that can open visitors up to all sorts of privacy problems. Sites can use third-party cookies to do things like use behavioral advertising (targeting), which allows sites to advertise to you based upon the history of your online activities. Some programs still allow third-party cookies. So be aware of all the facts before you decide to allow third-party cookies to be set on your computer.

Today there are also Flash cookies that can be set to track users, and in some cases to reset deleted cookies. A lot of Flash applications use cookies to remem-ber your audio preferences, video preferences, and other settings. However, some also use them for nefarious reasons. You can manage your Flash player settings by visiting http://www.macromedia.com/support/documentation/en/flashplayer/help/settings_manager.html, clicking Global Privacy Settings, and then clicking the Website Privacy Settings Panel link at the very bottom of that section. Figure 8-2 shows what a settings panel looks like, and you can manage your Flash cookie settings from here.

Google Analytics does not use Flash cookies either, only first-party cookies.

Figure 8-2. Adobe Flash cookie/privacy settings

As great as Google Analytics is and as much as I'm a fan of it, it's not perfect. In its defense, it's not designed to be perfect. In some respects it's only as good as the data that it can collect from its users, which is something that every user is in full control over.

Some of Google Analytics' limitations are:

Figure 8-3. Data sampling in Google Analytics

The limitations of other web-analytics solutions are as varied as you can imagine. Some limitations of Google Analytics are duplicated in Omniture SiteCatalyst, and some WebTrends limitations are default features of Google Analytics. For example, consider the "bounce rate" metric. This is not a supported metric in the latest version of WebTrends. But it's standard in Google Analytics. Lyris HQ (formerly ClickTracks; http://www.lyris.com/solutions/lyris-hq/web-analytics/) and TeaLeaf (http://www.tealeaf.com/) have features that no other platform offers. Only Google Analytics has native Facebook Markup Language (FBML) integration, while platforms like Klout and Twitalyzer are the only ones that measure your Twitter influence. As you can see, different platforms report and behave differently (which is a lead-in to one of the most common web-analytics misconceptions, discussed later in this chapter).

This not-for-profit organization is the de facto governing body of the web-analytics industry. It has published research documents and metric standards with the ultimate goal of influencing every web-analytics vendor to adopt consistent, standard naming conventions and metrics definitions across the board.

The group's 2008 Standards Definitions Volume I can be downloaded via this URL:

http://www.webanalyticsassociation.org/resource/resmgr/pdf_standards/webanalyticsdefinitionsvol1.pdf

When you have some time, read up on how this body classifies metrics into four categories and how it defines every one of the standard metrics.

Everything that Google Analytics collects is bucketed into one of two categories. It's either a metric or a dimension. A metric is a count of something, like a page view or a goal completion. A dimension is an informational text string, like a search engine, page title, or keyword.

Let's define the metrics first and then the dimensions.

Remember that Google Analytics isn't QuickBooks. Many, many folks feel that Google Analytics was created to collect 100 percent of all web visitors and match up to the exact decimal point the number of seconds spent by all users on the site. They feel that keyword searches, goal completions, and e-commerce trans-action and revenue data should be exactly what they see in their server logs, in their version of QuickBooks, and in their CRM database. Even the slightest deviation is often unacceptable and is used as a basis for reasoning that Google Analytics or any other web-analytics platform is broken.

The thing these folks should realize is that web-analytics programs aren't designed to function as hit counters or as accounting software that requires absolutely precise totals and tabulations. Web analytics has its limitations, and therefore it will never be able to provide such a level of completeness that you can compare it side by side with more precise tools. If a user has cookies or JavaScript or images or anything else blocked, the chances that Google Analytics or any other tool will be able to collect that user's info become zero. If a user has an extremely slow Internet connection and is making moves on the website before it fully loads (and before the JavaScript tracking codes load), you can kiss accuracy goodbye. If tags are not installed correctly, if URLs aren't tagged properly, or if other factors intrude, it's so long data.

These sound like simple hurdles to get over, right? Not necessarily. Many of you readers will have bosses to answer to, or key stakeholders to present to. Bringing up the topic of total accuracy (or its lack) may ruffle feathers and poke giant holes in conventional wisdom, and could cause you to commit unintentional career suicide. Totally perfect data is fool's gold: Even if you get there one day, what are you going to do with it? Is having 5,423 goal conversions in a perfect data-collection world going to change your rationale and make you do things differently than if you had 5,419 goal conversions instead?

Web analytics is designed to analyze trends, provide insights, and high-light key data that you can use on your site and in your marketing initiatives. The sooner your boss or your colleagues understand this concept, the easier everyone's jobs and lives will be, and the sooner you can make progress and improvements online.

In high school I gave a senior-class presentation on the JFK assassination. I remember talking about how it was impossible for Lee Harvey Oswald to have been the lone assassin. Yes, I'll believe a lot of things, but even I can't buy the theory that Google publicly shares and sells your information.

In the data-sharing section within your Google Analytics account, you have the option to anonymously share your data with Google services and other products. The data is used to create tools like those in the benchmarking section. Other than that, there's no evidence other than accusatory blog posts that Google is just waiting for the right time to sell everyone out and rake in huge profits. If there were evidence, it would probably trigger lawsuits and congres-sional investigation. There isn't any such evidence. Google isn't selling your data, period.

I think Omniture SiteCatalyst, WebTrends, and CoreMetrics are excellent web-analytics platforms. They have features beyond anything Google Analytics can currently do, and in my opinion they provide the type of competition required for Google to continue to improve its own product. However, these tools come with a pretty hefty price tag, going all the way up to the six-figure range, depend-ing on your package and service selections.

What's at work here is the idea that you get what you pay for. Google Analytics is free; therefore, it can't be all that good, right? However, big expensive programs should be able to answer all questions, solve all problems, and even iron your pants for you. Unfortunately, while they may be good, they aren't that good either, no matter how much money you paid.

The truth is that it still takes a human being to use the tool and to extract insights and data from that tool. The web-analytics platform is only as good as the person using it. Think about a first-time driver getting behind the wheel of a Formula One racing car — he'd crash it within seconds! But take Michael Schumacher and put him behind the wheel of a Fiat, and I'm sure he'll get the most out of that Italian compact.

Just because you pay a lot of money for a web-analytics tool doesn't mean that it will raise your conversion rates and bring you more revenue. You're still going to need a great web analyst and a great marketing team to help you along the way.

People who buy a big-dollar web-analytics platform also feel sometimes that it should be the only tool they'll need. That same feeling can extend to Google Analytics — those who use it can begin to see it as the only thing around, and be unaware of what other tools are out there.

Over time you'll want to develop a repertoire, a bag of tricks that includes diverse tools and services. Just as you'd want to diversify your 401(k) stock portfolio, you'll want to diversify your web-analytics strategy to cover the most possible ground.

You have Google Analytics/Yahoo! Web Analytics/Unica NetInsight? That's great. You also need a customer-feedback service like Kampyle or 4Q by iPerceptions. If you use social media — and who doesn't these days? — you'll need to make use of Facebook and YouTube Insights, Twitalyzer (http://www.twitalyzer.com/), Klout (http://www.klout.com/), or paid services like SAS (http://www.sas.com/) or Radian6 (http://www.radian6.com/).

Your arsenal won't be complete until you dive into tools like HitWise (http://www.hitwise.com/), comScore (http://www.comscore.com/), Google Insights for Search (http://www.google.com/insights/search/), or Google Trends for Websites (http://www.google.com/trends/).

That is, until you want to start doing mobile analytics with tools like PercentMobile (http://www.percentmobile.com/) or Bango Analytics (http://www.bango.com/), or you need to measure blog statistics and performance with FeedBurner (http://www.feedburner.com/) or Sentiment Metrics (http://www.sentimentmetrics.com/). Oh, and don't forget about tools like CrazyEgg (http://www.crazyegg.com/) and clickdensity (http://www.clickdensity.co.uk/), which allow you to view click data on top of your website's pages.

As you mature in web analytics, your current lone tool will not be able to answer every question or measure every statistic that you ask it to. It's not the tool's fault; it's just not designed for it. So you'll have to expand your horizons and get out of your comfort zone to find new data sources with interesting and useful information. I'll highlight some of these in Chapter 17, but for now, check out any of the tools that I named in the previous paragraph for a good head start in what will be discussed toward the end of the book.

I often come across a website with more than one web-analytics package. More often than you may think, a website may be using Google Analytics, Yahoo! Web Analytics, WebTrends, and Omniture SiteCatalyst all at the same time. (Okay, maybe not exactly that scenario, but it's quite common to find more than one web-analytics tool installed on a site — simply view the source code or install a tool like the Web-Analytics Solution Profiler (WASP) tool for Firefox, to view what web-analytics tools are setting cookies on your machine.)

People also tend to compare the statistics from one platform with those from another, and wonder why the two systems don't match. A common conversation can start out like this: "So I'm looking at WebTrends and I see 4,350 page views for the month of March. But then we installed Google Analytics and we see 3,780 page views for the same month. Why is this happening and which one is right?"

From reading up to this point you know why this happens — tools collect, process, and calculate differently from one another. Some tools (like WebTrends) count search engine spider activity, while other tools (like Google Analytics) do not. Also, some settings, like filters, are applied on one tool and are not applied (or aren't possible) on another. But you knew that already. The next question may be the tricky one to answer: "Which one is right?"

The answer is that both of them are right. This is a tough pill to swallow, especially if it's a client you're talking to. But assuming that both platforms are installed and configured correctly, the logical conclusion is that there are two correct answers, even though the answers themselves are different. The key is to not "use the bigger number" all the time (which is easy to do with things like e-commerce transactions or goal conversions). Remember that web-analytics platforms are designed to analyze trends and are focused on cultivating insights — they're not a replacement for server logs or accounting software. Train yourself and your clients to think this way and to stop compar-ing statistics from one tool against those from another.

To the contrary, Google Analytics can handle large volumes of data just fine. The root of this misconception is the five-million-page-view limit per month for non-AdWords advertisers. This is simply a limit for those web analytics accounts that are not synced to an active AdWords account in good standing. If you wished, you could open up an AdWords account, sync it to your Google Analytics account, and have as many page views as your website can handle.

At the time of this writing, both Facebook and Twitter — two of the Internet's largest-volume websites — use Google Analytics on their sites. They wouldn't be using it if it couldn't handle volume.

In Google Analytics you can add as many reports to your dashboard as you like, which you already know from Chapter 7. When you make an executive dash-board in Excel, there are also no limits to what you can do with it. Sometimes the dashboard is dictated by your CEO or VP, and you basically just have to follow orders.

But it's high time that folks started creating dashboards that made sense. Far too many executive dashboards, high-level overviews, or top-tier presentations include pages upon pages of stuff with charts, graphs, metrics, columns, rows, and data, until the view gets dizzy. The entire purpose of a dashboard is to have a high-level review of what's going on — not to provide every single detail. If detail is requested, then that can be provided in a separate tab, worksheet, or file, but don't pollute your dashboard with too much data. Save the good stuff for later. Unfortunately, far too often I see dashboards that remind me of an episode of Hoarders, with so much data piled in one small place that I wonder what I'm supposed to do first.

When I create a dashboard I try to think like a salesperson, whose objective is to make that sale or renew that contract. When the salesperson is putting together a presentation or sales pitch, it usually has at most a summary of bul-let points, some high-level information, and possibly one or two paragraphs of explanation. In essence the salesperson is creating a dashboard that will be reviewed and looked at by potential clients. Try this approach the next time you're creating or working on your dashboard — remove the clutter, cut the fat, and don't become a dashboard hoarder.

Another one of the most common web-analytics misconceptions concerns the Top Exit Pages report, found within the content section in Google Analytics.

Take a look at Figure 8-4, where I show the top half of the Top Exit Pages report, purposely without listing the page names. What insights, actions, or intelligence can you derive from this information?

Figure 8-4. The top half of the Top Exit Pages report in Google Analytics

As I discussed earlier, visitors have to leave a site eventually. They can't stay on your site forever. Looking at this report and trying to get anything out of it other than some fun facts will drive you insane. You will think your homepage is "leaking traffic," and will make some radical change that won't work and will wind up breaking your CMS System, while simultaneously taking your website offline. If that's going to happen to you, at least do it after looking at the Top Landing Pages report, which is far more useful than this report could ever be even on its best day.

If you have colleagues, customers, or clients who use this report, please attempt to discourage them from doing so. Offer metrics like bounce rate and reports like Top Landing Pages as alternatives.

When a year-over-year comparison is performed, I bite my tongue really hard so as not to say anything. It's even harder when I see comparisons ranging back a couple of years, or (as has happened) a comparison between log-file web stat online data from 2000 and 2010. I added this final piece of this section to cover just this topic.

Online data is like a jar of mayonnaise — it's perishable and it has a shelf life. Your data, after a certain period, is just not usable anymore. It could very well be rotten and ready to be thrown down the garbage disposal. On the Internet, things change really fast. In 2006 Twitter had just been born, and the topic of social media was new. Our offline and online worlds have completely changed in a relatively short time. Comparing web data from 2005 to web data from 2010 is like comparing Nielsen television ratings from 1950 and 2010.

There are a few reasons why you might compare year-over-year data, but only a few. If you know people who habitually compare visits from 2005 to visits from 2010 or page views from 2003 to page views from 2010, you might try to educate them.

If you haven't failed yet at doing online marketing or web analytics, then you probably haven't been doing it very long. If you have been doing it a long time without a failure, consider yourself one of the very few extremely fortunate ones. It's virtually inevitable that you'll fail sometime at measuring, implementing, or doing web analytics (or online marketing). You'll make a mistake, and it may even cost you time, resources, money, or, worse, your data.

Just as in sports, you learn more from your failures than from your successes. What I've learned from my failures is that I don't know everything, and I don't like to consider myself an "expert" or a "guru" (I actually don't like either term). But early on I thought I knew everything, until I made that inevitable screw-up that cost a client some important data. I learned the hard way that I don't walk on water, and it humbled me. This knowledge was reinforced soon afterward when I struggled to answer a client's web-analytics business questions, despite my very nice summary of web data and stats, which were of no help to the client.

You will fail in web analytics. Don't fear it — simply be prepared when it happens. Try not to get mad or frustrated — learn from the experience, store it away in your bag of tricks, and move on to the next assignment. And don't call yourself an "expert" or a "wizard" or any other self-promoting noun — there is always someone out there who knows more than you, or is going to know more than you someday.

The key to success in web analytics is segmenting your data. Data at an aggregate level can take you only so far and can answer only a few questions. You need to know which keywords came from Google, and which visitor types landed on your homepage. You need to be able to customize your data segments, as I'll show in Chapter 9. Without the ability to slice and dice your data, you won't be able to obtain the deep insights that are required in order to truly understand your website. Figure 8-5 shows you an example of a segmented report in Google Analytics.

Let's say that you start at the very top with all visits, and see that there were 50,000 visits to your site during August. This is the highest-level, most unseg-mented data possible, and it doesn't tell you anything at all other than how many visits you received. What action can you take on your website, AdWords campaign, or e-mail marketing blasts from knowing that number? You can't take any action, and that's the whole point of segmentation with web analytics.

So you start drilling into the Traffic Sources Overview report and find that 60 percent of your site's traffic was from search engines, 20 percent was direct traffic, and the other 20 percent was from referring sites. Your large and costly e-mail campaign isn't even a blip on the radar, but you see as you segment one level deeper that almost all your search engine traffic came from your Google AdWords campaign. Now, because you segmented, you know that your web-site's traffic is highly dependent upon AdWords, which means that you have no organic visibility and not very many people are typing in your website's URL directly. You have SEO and branding work to do.

Figure 8-5. Segmenting in Google Analytics

But then you go even deeper and see that there are only two campaigns that are converting out of the nine campaigns you have running in AdWords. You drill down into those two campaigns to find the ad groups containing the three converting keywords in each that are the sole contributors of business for your online initiatives. You then segment those keywords by geo-location to find that the conversion rates are the best in New York City, Chicago, and Los Angeles, and segment again to discover that almost all purchases happen in the afternoon Monday through Friday. These purchases from those very lucrative keywords are only by returning visitors with fast connection speeds using Firefox. They take six visits and three days to purchase on average.

You then backtrack to find out that the non-converting keywords are money-guzzlers in need of strong optimization efforts, because they are bouncing from your pay-per-click landing page. The keywords that do make it but don't convert are leaving on Step 3 of the goal conversion funnel, and visitors are spending less than four minutes on the site on average. They're also searching for product catalog IDs on your site's internal search function, which is not working properly. This could very well be another reason they're not converting.

Yes, I love segmenting. And so should you. Or you can continue to use aggre-gate metrics like 50,000 visits to make your decisions. I'll be seeing you at next year's holiday party with all your clients in my portfolio.

I don't like doing year-over-year comparisons, but to become a great web analyst you're going to have to perform at least some month-over-month ones. You need to have some frame of reference, like the most recent month, to be able to tell you how your efforts are progressing (or regressing) over a reasonable period.

With Google Analytics you can click the Compare to Past checkbox within the date-range tool to perform a comparison of data between one date range and a previous one. Figure 8-6 shows a 30-day date-range comparison by search engine, to see if my conversion rates have increased or decreased over time. In Figure 8-6, at least for Google, I am showing a very strong increase in the conversion rates across the board in this particular set of goals, which reaffirms to me that whatever my optimization efforts were, they are paying off big-time. The interesting thing to note is the very first column on the left-hand side. You can't see the column heading in Figure 8-6, but it's the number of visits within my specified date range. Notice that I'm getting many fewer visits in the most current time period than in the previous time period, yet my rates of conversion are much higher. This is another great indicator of success — I've refined my site and my efforts in such a way that the traffic I am now receiving is much more ready to convert. In essence I've been able to weed out the irrelevant traffic, which wasn't doing anything of value anyway, so it's not that important.

Figure 8-6. Comparisons in Google Analytics

Yes, your dashboard should never be a candidate to appear on the show Hoarders. It needs to be clean, crystal-clear, free of clutter, and easy to read, not just by your client or colleagues, but by you as well. After all, you are the keeper of the flame, the web-analytics practitioner who will give advice to your client, boss, or yourself if you own your own website.

Try to find a way to show what's really important on one page. If you abso-lutely need to have more data for a weekly meeting or client report, attach it in subpages, leaving you with one executive summary page. Use a font size that's easy to read and big enough that no one has to squint. Use simple, plain color schemes and don't get too creative with your background logos or Flash pieces. Organize your columns and rows neatly and with common sense. Of course, double-check the validity and accuracy of your data — nothing ruins a dashboard like data that's incorrect.

If you're planning on using the Google Analytics dashboard, make sure you add only the reports that you need. Feel free to reposition or remove the default dashboard reports. Limit the reports you add to your dashboard to six (any more than six will generate another page when downloaded). Remember that there's nothing you can do about the Site Usage window or the trending graph — those two will always be there on your dashboard and cannot be removed.

Take the feedback that you receive about your dashboard or executive sum-mary constructively, not as a personal attack. Remember that the data you're presenting has to satisfy your clients or stakeholders, so ultimately they're in charge. Temper their requests to fit the framework of what you know works best in a dashboard, and try your very best not to let it get out of control.

I mentioned this earlier in this chapter as well as in Chapters 3 and 6. You are not the only fish in the sea. There are competitors doing what you do and thinking of ways to do it better than you. So you're going to have to incorporate some competitive-intelligence insights into your framework to stay on top of your rivals, and keep your clients or customers right at home with you.

If you have the budget for it, try using HitWise's wide range of competitive-intelligence services (www.hitwise.com). Figure 8-7 shows HitWise's weekly retail intelligence report, which is freely available on its website. You can view industry rankings arranged by the percentage of visits, the fast-rising search terms, and the industry search terms. This is great data that you can have for your site if you can afford the paid version of HitWise.

If you don't have the budget for it, try Google Insights for Search and Google Trends for Websites, which I'll talk about in Chapter 17. You can view hot trends and fast-rising search terms, but unlike in HitWise's freely available report, you can modify date ranges and segment by country, state, or city without even logging in to your Google Account.

Figure 8-7. HitWise weekly retail re port

In Chapter 1, I talked about asking yourself the question "What is the purpose of my website?" To be a great web analyst, you're going to have to understand your website like no other person on the planet. You're going to need not only to answer the "purpose" question, but also to know what the secondary purposes of your website are, what functionality the site provides, and where all the nooks and crannies are. You need to understand your website's target audience, the site's content, and what actual visitors are viewing your website. This is aside from knowing how to perform analytics and calculate metrics. You're going to have to become that word that I hate — an expert — on your own website in order to truly be a great analyst.

Now, I understand that those reading this book may include consultants, marketers, and analysts in charge of looking at multiple websites. If you don't build in some time to learn as much as you can about the business, the web-site, and anything else you can get your hands on, you will eventually fail at doing web analytics. Take as much time as you possibly can to test-drive the site you're analyzing. Ask questions. Fill out test form submissions. Download files and play with the Flash applets. Make test purchases on the shopping cart and go through the order process. These experiences will help you become intimate with the website, which is what you'll need to do to effectively speak the language that your client or website owner is speaking, and to pro-vide the insight and recommendations necessary to succeed and be a great web analyst.

Flip over to the next chapter to learn about some advanced tools that Google Analytics has to offer to make you even better at web analytics than you already are.