Adam Howitt Consulting Blog - Google Analytics

0% Bounce Rate? Really?

Sat, 23 Jan 2010 10:17:00 -0400

You updated your website and your bounce rate dropped to zero percent or thereabouts - time to celebrate, right? Maybe.

Understanding Bounce Rates

Just a quick recap for the uninitiated - bounce rate represents the visitor who came to your site, looked at a single page and left without interacting with your website in any way. The goal of bounce rate is to count visitors who were so unimpressed that they didn't feel the need to find out any more about you or your company.

Reducing your bounce rate should be a good thing and generally is but be very suspicious of a bounce rate of zero or a drop so substantial to be practically zero if your pages never performed so well before.

Troubleshooting Bounce Rates

Note that I only said "be suspicious" and not reject them outright. All you need to do is to understand if there is anything artificially affecting your bounce rate. If you can prove that each page generates a single pageview in Google Analytics you can crack open the champagne and dance in the streets.

What are some of the factors that can lead to inflated bounce rates?

AJAX tracking code. I've seen AJAX and jQuery code extensions that load content and send a trackpageview on document ready. In some cases if someone interacts with a dynamic section of your site, you want that interaction to be recorded as an interaction and not a bounce but if the trackPageview call occurs when the document has finished loading you'll see your average pageviews per visit jump to 2 or close to 2 for this reason. The first "hit" is real - the visitor was tracked with the normal tracking code at the end of your page. The second "hit" happened without the user touching their mouse or keyboard so you should take steps to disable this trackPageview in this instance. By all means track it if the user did something to make the content load though.
iFrame content. If your page includes an iFrame showing content from another part of your site that also has the tracking code you will probably see similar symptoms to the AJAX description above of inflated pageviews per visit and decreased time on page to practically zero.
Duplicate tracking code. This is probably the most common culprit I've seen. View the source code of one of your 0% bounce rate pages and search for ga.js and urchin.js in the code. Make sure you're viewing the source as displayed in your browser instead of the server side code because you'll be getting the exact picture of what is being sent to google vs. forgetting a file that get's included on the server side. To view the source code right click in some white space on the page you're testing and select view source from the popup. If you have one of each on the page or two of either with the same user account "UA-xxxxxxx-x" string you have a problem. In certain circumstances you might have two user accounts tracking a page but if you're reading this post, my guess is that you aren't at that point in your analytics sophistication!
Third party vendor. Technically the same as a duplicate tracking code issue but slightly different. You might have a web page hosted on someone's platform like wordpress or site builder type setups. A common issue here is using built in Google Analytics tracking features provided by the vendor but using a designer's template that includes Google Analytics tracking too.

Tools of the Trade

If the tips above weren't enough to get you an answer there are a couple of tools you can use to see what is being sent to Google Analytics when your page loads

Packet sniffers. These tools are fairly technical but your developers may be familiar with these kind of tools. They capture the web requests leaving your computer so you can review them and look for any traffic to _utm.gif. Wireshark and Ethereal are the most technical tools to use and not specifically dedicated to browser only traffic
HTTP Proxies. Another technical tool you can change your browser settings to use an HTTP proxy that intercepts browser only traffic and will give you less background noise than a packet sniffer. I don't have any examples of these but someone in IT might be able to help
Proxy Browser extensions. The tools I use most are browser extensions that listen to the traffic from your browser but don't require an IT degree to install and use. Firefox has LiveHTTPHeaders which works more like a proxy and shows you the traffic it is capturing. For Safari you can turn on the develop menu from the advanced preferences screen and look at the resources panel for your page looking for _utm.gif requests.
Analytics Tools. If you are spending a lot of time on Analytics work for multiple clients or multiple internal sites, I recommend buying a copy of WASP, recently acquired by Avinash Kaushik's company (so it can only get better). As it is, it sits in Firefox in the corner and quietly reports the Analytics solutions deployed on your server including tracking mechanisms used by the advertisers if you have banners too. Click Google Analytics when you see it and it shows you the Google Analytics hits described in the most user friendly way to tell you what was sent to Google. It deconstructs the variables in a Google Analytics request to make it easier to read. If you see two tabs in the panel that pops up that both say Google Analytics you'll be able to see what is being sent. Looking at the trackPageview title should give you a clue as to what is being reported.

Still stuck?

If you're still stuck at this point, fear not! I've worked through many of these issues with my clients and I can help you get to the bottom of your issue and work with your developers to fix the problem. Send a request through our contact form and I will be in touch.

Google Analytics change to setVar to affect bounce rate and time on page

Thu, 29 Jan 2009 11:11:00 -0400

Google announced on Tuesday that as of Wednesday January 27th calls to the setVar method (used to create custom segments on your site) will no longer be counted as an interaction hit.

What is an interaction hit?

Prior to January 27th, a call to setVar would be logged as an interaction hit meaning that if it followed a call to trackPageView, the bounce rate for that page would be 0% and the time on page would be tiny, reflecting the difference between the first call and the second call.

This change is a positive move since bounce rate has long been considered a critical measure of success of marketing campaigns from all sources. If you drive someone to a page and they leave immediately it is called a bounce and indicates that the messaging used to lure them to your site conflicts with what they were able to see when they arrived on your site (in the 8 seconds they typically spend on a given page). Similarly if your time on page was underreported you might mistakenly waste time optimizing a problem that doesn't exist.

Who does this affect?

A significant number of people who use Google Analytics "out of the box" won't be affected by this change as setVar is only something you implement as a conscious decision to group visitors into custom segments. If in doubt, right click on a page on your site and click view source then look for setVar in the code (Control+F will pull up a search dialog).

Why use segmentation?

You might choose to group landing pages into custom segments so you can compare the effect of landing pages on visitor behaviour. Another common application is to use segments to label a visitor by their perceived intent.

I might track visitors landing on my site into segments based on source to differentiate the behaviour of visitors from social networking sites from that of other referral sources.

Why? The custom segments extend beyond the first visit and will only be changed if the visitor receives a subsequent segment, since the last referal before a sale is commonly accepted as the key influencer. That said, if social networking sites deliver a lot of traffic but convert poorly into sales, I can see that trend across the larger group rather than look at myspace, facebook and twitter separately. Similarly, the bounce rate and browsing habits might be very different so I would want to isolate this behavior from paid search to make sure I'm not optimizing to a biassed crowd.

Alternatives to setVar

Google recently launched a beta program for "Advanced Segments" located in the top right of the dashboard. It allows you to pick from pre-defined segments and create your own. It is a subset of the functionality provided by custom segments since it only tracks visits, not visitors, but is still incredibly useful.

For example, you can compare behaviour on the site between browsers. Looking for patterns here like drastically higher bounce rates could mean that your site renders poorly in one browser or actually appears broken to visitors for that browser.

I have two segments I created to identify visitors who started on my blog and another to track visitors who began their visit on my home page. Believe me, the insight this provides may help you divert the CEO's demands to put every new toy on the home page...

I encourage you to play with the advanced segments feature, create some segments and compare behavior against the site and other similar but distinct segments to look for trends. If you need help with this tool, of course you can contact me through the contact form on the site or leave me a question in the comments.

Double your conversion rate, quadruple your sales?

Mon, 12 Jan 2009 07:13:00 -0400

Sound unrealistic? It depends where you start from of course but one client I'm currently working with was able to turn up the heat through some simple changes to an estimate form pivotal to his business.

My client, Auto-Transporter.com, ships cars around the US for people relocating or just heading south for the Winter. The goal of his website is to get visitors to submit an estimate form to his sales team. He came to me looking for help developing a profitable Google AdWords campaign but I believe that is the last step of any optimization project, never a beginning. Here's why.

If your site converts 10% of your visitors and you get 1000 visitors a day you'll get 100 leads. To double your leads with Google AdWords you'll need double the traffic. Every day. One thousand clicks at a typical cost-per-click of $1.50 is going to cost $1500 per day, $45,000 per month. It's not sustainable for most businesses and depending on your experience with AdWords, you may end up attracting clicks from visitors outside your target audience, resulting in lower than expected conversions.

The best way to get an additional 100 leads is to double the conversion rate of your site from 10% to 20%. The same 1000 visitors a day now yields 200 leads and, since conversion rates tend to be consistent (for most businesses), you'll pay once for the work and reap the rewards day after day with no extra expenditure. If you run the same Google AdWords campaign after optimizing your site you'll have four times as many leads!

Just to be clear about conversion rates staying the same - it depends on what you do to market your business. A hugely succesful marketing campaign that provides more information than your website does is likely to drive a spike in conversion rates but it masks the issues on your website. Seasonal businesses are another exception but for the most part let's go with the consistency theory for the sake of argument.

Improving Conversion rates

Okay, so how did we impact conversion rates? We started by ensuring that every page in his site was tagged with Google Analytics to help get a clear picture of what was happening on his site. After confirming that the tracking codes were in place we waited a week to gather some meaningful data. After all, it's hard to know if you've improved something if you don't measure it!

In our next meeting we used the data to decide how to track the goal of estimate form completions as a goal in the Google Analytics reporting interface. We created a two step funnel to capture data about the number of visitors who saw the estimate page vs. those who submitted it successfully.

Another week passed as we collected the data to begin a meaningful discussion. The funnel showed us that 25% of visitors would submit it, or put another way 75% left without submitting the form. Think about this for a second. 75% of people didn't do what he wanted them to do.

The goals in Google Analytics are smart enough to know that a visitor who saw the form, clicked around on the site and then came back to submit it is still a succesful visit and a conversion.

Why do people leave?

There are many reasons why people leave sites and I'm sure you can think of others but some I have come across are:

Not ready to buy: Does your form alienate people who aren't ready to buy?
Form is buggy: Is it broken somehow in one or many web browsers?
Confusion: Is it confusing in the way it is laid out or are the questions confusing or hard to answer?
Restrictive: Do the options you offer as pre-selected answers to questions cover everything people might want to select?
Intrusive: Does the form require too much personal information too soon? If you need just one way to get in touch with the customer is it clear or does it appear that they should submit everything?
Intimidating: Is the form too long or does it look to complicated?
Lack of information: Did the visitors receive enough information prior to arriving at the form to be able to submit it? Is there anything they need to know to be able to answer the questions?

This is all well and good but how do you identify the issues? Google Analytics only tells you who came, who left and who submitted the form but provides no further insight into the underlying problems.

Form Analytics Tool

To dig deeper you need to use a form analytics tool and I can't recommend ClickTale enough. The script is installed on your form and the thank you page just like the Google Analytics tracking code. The full tutorial of Clicktale is beyond the scope of this article and the tool packs many more secret powers but I use it primarily for form analytics. Try to pick a plan matching the number of pageviews your form sees each day unless you're installing the script throughout your site, in which case you'll need a bigger plan. I like their model because even if you don't pick the right plan, it will record as many pageviews as you've paid for each day and no more but distributes these recordings throughout the day.

We waited a week for ClickTale to gather enough data about the form behavior then logged in to see what the tool could reveal. We selected the last week as the date range for the form analytics report and after a brief pause the form is shown with key statistics across the top of the frame. These include conversion rate, drop rate, completion time, blank field rate and refill rate.

Conversion rate is something we've seen before in Google Analytics but just to reiterate, it represents the percentage of visitors who turned into leads in this case. You can consider this an absolute measure of success for your form since increasing leads generates more sales! As the page loads you'll also see a funnel with more detail about how many people saw the form, started it vs. didn't, tried to submit the completed form vs. didn't and finally those who successfully submitted the completed form vs. those who were unable to submit it successfully. This last point means a visitor submitted the form but something was wrong with either their data, or there was a technical problem.

Drop rate is the number of visitors who started the form but didn't submit it. A high drop rate lets you know that visitors are getting stuck somewhere and if you click on the summary along the top it swaps out the conversion rate panel with a drop report panel highlighting the fields on your form where people most commonly stop filling out the form.

Completion time is the average time taken for visitors to complete the form over the period. It took an average visitor 8 minutes to complete the estimate form for the period we reviewed. Click the completion time summary to see the time taken to complete each field and find the fields where people are scratching their heads!

Blank field rate tells you what percentage of fields on the form are typically left empty. Click the blank field rate summary to see an analysis of which fields were skipped most commonly. If you really don't need the information that is being skipped you can reduce the complexity of the form and drive conversion rates by removing them. If you do need the information, maybe you're asking the question in a way that isn't immediately obvious.

Refill rate indicates what percentage of your visitors tried to answer a given field but were asked to try again. The summary covers the whole form but click that number to see the detail behind the failure rate of each field. A high refill rate suggests either a poorly worded question or a field that is allowing people to enter the wrong thing - like a date field that doesn't make the format obvious. While a web server can convert most date formats into the form it needs, people will spend less time thinking if you make it so obvious that they never get it wrong in the first place.

People are empowered by forms they can fill in easily but project their anger about a complex form at your company and are unlikely to come back, unless you have some kind of monopoly.

After reviewing my client's Clicktale data we removed some fields, renamed some and changed the options on others and just one week later he saw a 50% increase in number of leads submitted and conversion rates climbed from 25% to 36%. Drop rate fell from 35% to 27%, completion time fell to just two minutes from eight minutes, half as many fields were left blank (11 down to 6) and the refill rate dropped from 4.7% to 2.7%!

As you can see it's a big improvement from changes that took a couple of hours to implement. We've contracted a designer to provide some different form designs to test based on the feedback from the last review and we expect to continue to improve.

Fixing some of the issues on your forms is a cheap and effective way of increasing conversion rates which lead ultimately to a sustainable increase in revenue for the cost of less than a day of the Google AdWords campaign.

Seth Godin says my seminar is a lost cause

Wed, 03 Dec 2008 06:44:00 -0400

Well, close. Jeff just sent me an article from Seth's blog pointing out that marketing evolution is much harder than marketing gravity because gravity is something people already believe in.

I'm banking on people believing that evolving your website is cheaper and more productive than starting from scratch or paying for AdWords campaigns.

The seminar follows a logical flow from fixing the problems for the visitors you get before chasing new visitors with SEO and AdWords campaigns.

The morning shows you how to use Google Analytics to analyze the traffic you get to find the problems on your website. Next I'll cover Google Website Optimizer to help you split test a theory without fighting with the CEO over what goes on the home page.

The afternoon starts with Search Engine Optimization basics to make sure you're getting the best free traffic possible before you invest in pay-per-click, the focus of the last session of the day. Google AdWords can be expensive if the material covered in the first 3 sessions isn't addressed and I'll teach you how to change the way you buy your campaigns to get the most for your dollar.

The first one day seminar is December 17th in Chicago and space is restricted to a cozy crowd of 10 to promote interaction and make sure everyone goes home with a personal action plan. If you can't make it to Chicago for the day, let me know if you think there is a demand for the seminar in your city.

Website Evolution Seminar

Tue, 25 Nov 2008 11:39:00 -0400

Up the expertise you bring to the table. One-day intensive covers designing for search engine results, split testing of Web pages, using Analytics to arrive at a best design. Google tools provide you with a slam-dunk answers to every client design request.

I'm running a one-day four session seminar from the offices of Adam Howitt Consulting in Chicago on Wednesday, December 17th 2008.

Find out more about Website Evolution!

Cross Domain Funnels in Google Analytics

Tue, 02 Sep 2008 21:08:00 -0400

I've been battling Google Analtyics for the last week for two customers with a similar issue - each has a multi-step goal configured starting on one domain and ending on another.

As I understood it you add the tracking code to your store site and your third party shopping cart pages. What I didn't realize and have never witnessed before, is that you must add the code to every page in your store and every page in your third party cart. I found this out when I found a different Google Support page on the topic than the one I had read earlier.

The first symptom that something was wrong was a funnel with a required step only reporting the first page despite a simple series of steps defined. I just couldn't understand why the second page which was part of the same domain was no longer tracking since I added the cross-domain linker updates.

Next I re-read the page I referenced first above and realized that the cross-domain script goes on the destination as well as the source page but it didn't fix the funnel. However, I was able to use the GA Navigation Summary (Content > Top Content > Page detail > Navigation Summary) to find each page of the flow. Everything was being tracked here but not in my funnel. Hmm. If you are on the Navigation Summary you should see step 2 of your funnel as the exit to step 1 but that wasn't the case.

I searched for step 2 in the content dropdown and found it so I clicked that to see the Navigation summary for step 2. Here I saw entrances but no step 1 as previous pages. Stranger still, I could see step 3 (on the shopping cart domain) as a next page! For step 3 as the navigation summary though - I couldn't see step 4 so I was missing 1 & 4.

It was clear GA wasn't tracking the progression from step 1 > step 2 and starting a visit with step 2 somehow, recording step 3 next then ignoring the progression to step 4.

The light bulb moment came when I guessed correctly that the use of the linker script on step 2 and 3 meant step 1 was tracked with one cookie before step 2 announced the need to mark pages as part of the domain "none" so it created a second cookie.

The step 3 page also uses the "none" domain and so continues with the second cookie since the linker has passed enough info along to set a matching cookie on the new domain.

Step 4 (thank you on shopping cart) didn't have the linker script so GA created a third cookie because the shopping cart domain didn't match the "none" domain set by the linker on step 3.

I ran a test and, sure enough, my funnel started working again. Pah.

The bottom line is that when I re-read the wording on the reference pages, I'd missed the subtle fact that you don't just tag the pages involved in the link but every page in your domain that are to be tracked when you do anything with third party domains to make sure your Google Analytics setuprecords every click.

Google Adds Site Search to Google Analytics

Fri, 02 Nov 2007 10:00:00 -0400

One of the most important but least tracked aspects of your website is how people use search to find things on your site. Google Analytics has been updated to include a new section for tracking site searches by category.

I'll explain why it's important to track search, describe each of the new reports Google Analytics provides, offer some conclusions you may be able to draw from the reports and then explain how to configure Google Analytics to track searches.

Why track searches?
People typically resort to searching your site when they are either in a hurry or can't find what they are looking for. It can also be argued that if they are in a hurry and can't find what they are looking for that you still have an issue in that your site doesn't make the important information obvious enough. Given that the average person spends around 8 seconds on each page of your site you might notice that it seems like everyone is in a hurry!

What reports will using Google Analytics provide to help me?

Visits: Who searched and when?
- When did visitors use site search? - Trends over time to let you know when people are searching. The drop down box on this report allows you to dig deeper to see how many visitors searched, total unique searches, how many result pages were viewed per search, percentage of people leaving after seeing your search results page (hint: they didn't find what they were looking for), percentage of people refining their search to be more specific (hint: your search results were not very helpful), time spent on your site after they searched and lastly how many results did they user click on to find what they needed. All of these sub-reports should help you be more critical of your current embedded search. If people are constantly refining searches or leaving but you know the content they needed was there, maybe it's time to consider Google's site search or another service.
- How do visitors who searched compare to those who didn't? - This report has the standard 3 tabs you may be familiar with from elsewhere in Google Analytics "Site usage", "Goal Conversion" and "Ecommerce". The site usage tab allows you to use the drop down boxes to compare visitors that searched against those who did not in terms of how many pages were viewed, how many people after just one page (a "bounce"). If you have configured goals for your site you can switch to the goal conversion tab and, for example, compare visitors who added an item to their cart who used search versus those who did not to see if searches are more or less likely to purchase from you. If you have e-commerce configured for analytics you can go one step further and the e-commerce tab will tell you if you get more revenue from people who search than those who don't.
Search: What did visitors search for?
- Which search terms did visitors use? This report can tell you what the top terms used to search your site were and gives you the same 3 tabs as the previous report. This means that you can see whether people were more likely to leave your site or buy something based on the terms they used to search. Clearly if you do track revenue based on visitors to your site and you see that a popular search term results in substantial revenue, it is a strong indicator that this item or product isn't prominent enough in your navigation but has good demand.
- Which categories did visitors search? This report tells you how many people are using the more advanced category based searches. You can investigate whether one category of search causes more people to leave the site or purchase something than the other categories of search.
Content: Where did visitors search?
- Where did visitors start their searches? Searchers always have a starting page and this report can tell you which page was the last page they visited before they gave up clicking your navigation and tried to search instead. This doesn't necessarily mean that the starting page was the dud because they may have bounced around your site with the navigation first. People want to be successful and feel empowered when they use your site and you can improve this feeling by making visitors feel like they were able to find what they needed by clicking. It's a small victory but it makes people feel a little happier to use your site so if they resort to searching it can be very disappointing.
- Which pages did visitors find? Once they searched, where did they end up? This report will tell you where people went for each type of search.

What should you look for?
Take a look at the top 10 search terms on the site and think about the role of your website. Are any of the search terms valid terms for someone interested in reaching the ultimate goal of your website? The ultimate goal might be signing up for a newsletter, filling out an application, buying something or submitting a contact form. If the terms are for this type of content then you may have an information architecture problem. The solution to this problem is broader than the scope of this post but you should probably gather some of your team together to brainstorm some ideas. Here are some general pointers but for more information checkout Steve Krug's "Don't make me think".

Emphasis is an accent. If everything on a page is emphasized, nothing on a page is emphasized.
Prioritize. Your web page reads from top to bottom so the most important messages should appear at the top. There is only one number one slot on the page.
Keep it above the fold. Not everything deserves to appear in the top part of the page, the trick is deciding which content doesn't belong there.
Think like your visitor. Try to get some people who are unfamiliar with your website to review any different concepts for reorganizing your pages to make sure they can find the new elements you introduce. An alternative is to try the Google Website Optimizer to test the new designs side by side, otherwise known as an A/B test. Google serves your new and the old design to a fraction of your existing traffic so you can monitor changes in search behavior and navigation behavior. Ultimately you can track how many more people converted/ purchased/ signed up because of the new design and make the switch if you have positive results.

To setup search analytics for your site in Google Analytics

Go to Analytics Settings for the site where you want to monitor searches
Click the edit link on the far right of the "Main Website Profile Information" section
You should see a site search section with "don't track site search" selected
Change this to "Do track site search" and you should see a new field appear for query parameter and categories
The query parameter is the part of your URL which is used to tell your web server which search term was used. To work this out, go to the search box on your site, enter a search term and hit submit. Look at the URL on the search results page and you should see something like q=my+search+term or search=my+search+term or something similar. "q" and "search" are query string parameters and "my+search_term" is the search term. You may have more than one pair of parameters and search terms, especially if your search has options like cat=my+category where you filter a search based on a category of your website.
Enter a list of each query string parameter found in the previous step, separated by commas. So we would use q or search for our example above.
If you found that you do have different categories of search like "search people", "search offices", "search products" and there is a query string parameter used to differentiate each search type, select yes where it asks "Do you use categories for site search?" and enter the list of category parameter names.

Tracking customers beyond domain boundaries

Thu, 13 Sep 2007 14:44:00 -0400

When a user clicks a link on your site, taking them to a new domain or sub-domain, your Google Analytics reports show an abandonment from the first domain and a referral on the second. This poses problems if you are trying to track things like a sales funnel because you get artificially high abandonment rates.

You can install the Google script on both domains but because the cookie is tied to the first domain it is recorded as an abandonment for one and a referral for the second site you go to.

To get around this Google has instructions about tracking domain changes as your customers navigate from one domain to another and back again.

The intent of this approach is for things like third party shopping carts where you may have control over the layout to be able to show your script but the domain name is that of the provider.

Profitable Online Marketing with Google Analytics

Mon, 10 Sep 2007 20:51:00 -0400

After 9 years of working for other people I've decided to go solo and start my own consulting business. The site is rudimentary right now as I only handed in my notice today. As regular readers may be aware, I've evolved from an entry level programmer to a Senior Application Architect in my current role since getting my Masters in Software Engineering and developed an interest in implementing Google technologies. In my spare time I've been fostering a worldwide smash-hit, WalkJogRun.net, with over 4,000 visitors per day and 100,000 routes worldwide. I'm addicted to Google Analytics and have a white paper due out under my current employers, Duo Consulting, explaining how to monitor the effectiveness of your marketing campaigns and tune the content on your website.

The last three years at Duo has been a fantastic experience for me giving me the freedom to explore exciting new technologies and all under billable hours because our clients are willing to try new things. I've load tested and refined the Chicago Park District programs website to support as many as 3,000 transactions in the first 3 minutes of registration every quarter, debugged and load tested third-party ColdFusion CMS tools and built a Cicada Emergence Google Mashup in the last year alone. I've implemented Google Analytics for many of our clients resulting in important business changing results. Clients have paused underperforming ad campaigns, monitored the performance of redesign work, reallocated advertising dollars and even redesigned their shopping cart checkout process on the strength of tell-tale numbers through my analysis.

The primary focus of my new business will be helping clients to install Google Analytics, establish their goals for the websites and then monitor the effectiveness of their sites and the advertising campaigns they create. I'll help them identify areas of their sites underperforming and evaluate their ad campaigns to get the most out of every dollar. I have colleagues who can help with content development, Duo will be my first referral as far as deeper restructuring work and I can do development work too.

The secondary focus for my business will be to load test web applications and either help developers benchmark the sites or discover the bottlenecks and resolve them. I see this as a huge opportunity as the web continues to grow and web applications surpass the lowly traffic expectations they were initially conceived to handle. In some cases the trouble will be at the database level, sometimes it is the JVM, sometimes it's an innocuous ColdFusion funciton like createUUID that hit the OS behind the scenes and crash JRun every 10 concurrent visitors.
If you have read any of my posts in the past and value the expertise you have found, or if you have read this description and think you or someone you know would value from this kind of service, please pass on the link or get in touch. I can be reached by email at adamhowitt@gmail.com or by phone at +1 (312) 714 9229. Thanks for reading and stay tuned for more of the same tips, tricks and commentary.

Google Technology Presentation selling out fast

Tue, 12 Jun 2007 16:09:00 -0400

I've just had some exciting news that my "Using Google Tools to optimize content for business results" presentation at Web Content 2007 is filling up fast. The presentation explains how to manage the risk assoicated with making changes to your website through Google's A/B testing tool "Website Optimizer" and how to use Google Analytics to measure the effectiveness of your changes. It wraps up with an explanation about how to run a profitable Google AdWords campaign.

If you are interested in attending, the two day seminar is in Chicago on June 18th and 19th and includes Jason Fried of 37signals amongst other big names.

MVC - Most Valuable Customer

Wed, 06 Jun 2007 07:01:00 -0400

There was a great question about my Google Analytics e-commerce post this morning so I wanted to repost the solution I offered here. Greg wanted to know how to work out which referers and keywords drive the most revenue for his site when customers return over a period of a few months.

First I'll show you how to find high revenue customers generally then address your issue where people spend money over several months.

In the new reporting interface if you go to the "traffic sources" menu item on the left of your reporting interface and select "all traffic sources" you are nearly there. If you have correctly configured e-commerce tracking you should see three tabs above the table of results on the page "Site usage", "Goal Conversion" and "Ecommerce". Select "Ecommerce" and the table shows different columns. Left-click your mouse on the column "revenue" and your report is now a sorted list of the referring sites generating the highest revenue for you.

The other sort order you can use would be to click the "average value" comment to see which site refers the customers with the highest average order value which sounds like it might be what you are looking for.

The last report you may want to look at is under "Traffic Sources" > "Keywords" then select the "Ecommerce" tab and sort the table by "revenue" or "average value".

Google's cookie typically will keep track of a returning customer to your site so the original referer is maintained. This means the reports I describe above will still help you identify the best keywords and sites for visitors who keep coming back - just look for the highest average revenue referer and keyword.

If you need to track the average number of transactions per referer you can either set a cookie on the visitors browser incremented each time they place an order and pass that as a custom parameter on the checkout page or a more robust solution would be to query your customer order database on the checkout page to count how many orders they have placed and pass it in the same way. This is more robust because it doesn't rely on the client not deleting cookies periodically. This solution would also allow you to track a different stat too - you could pass the lifetime revenue as a custom parameter on the checkout page. A filter would allow you to translate the values into revenue bands and then under "Visitors" > "User Defined" you could review each revenue band. You have the option of looking at each user defined value cross-referenced by segment (the dropdown) so you can see which referers and keywords are responsible for the highest lifetime value customers.