If you ever visit the web analytics vendors's roadshow regarding optimization tools, it is common for them to publish case studies showing huge lift in conversions. Do not expect your site's test result to see similar lift by just implementing the tool. You need to understand what you'll be testing. Not only to plan what to test, you need make that testing into a process rather than ending at a project based testing.

Think about the resources, time, budget, and efforts that'll need to go into executing the A/B test or Multivariate tests.

• In order to create creative assets for testing, you'll need a designer to pump those graphical elements.
• You'll need to set up meetings to bring in people you'll need to share learnings, and key stakeholders who'll create a budget to perform your tests. Tests are never free as long as someone's time is involved. Including yours.
• Web programmer will need to implement that javascript code into the test pages.
• Depending on the landing pages you'll be testing, you may need tons of traffic. So you may need your SEM expert to pump some traffic after that budget is approved.

Once you hammer out these issues, and acquire key stakeholders buy-in, its time to go at it.

Start simple and start breaking it down. Consider testing these basic test cases.

Layout A vs. Layout B: When the call to action (CTA) are hidden below the fold due to layout constrains, or simply want to avoid long meetings and argue about what looks nice, layout tests are great way to start testing using A/B/n test.

Colors to Call to Action (CTA): Colors do matter when it comes to click-thrus by consumers. It may go against company's branding guide line, but it could be a great test to raise your voice about what works and don't work.

CTA and message: Messaging and CTA of them key URL links do matter, and you could be surprised how little differences that you didn't think of could lift your conversion rate significantly. 1% lift in conversion rate could mean X more million dollars in sales. Test it.

Size: Font size, image size, product size, you name it. Prominence of key elements fonts or images could mean significant lift in conversions because consumers could value something more than the other.

Traffic Sources vs. Landing page: A/B test or MVT may sound like you're only testing landing page only, but your test may yield different results by different traffic sources. Different acquisition channel may bring different persona, so results could look different when serving a content to banner traffic vs. SEM traffic.
Here is a great article that speaks to it: Maximize Your Google Website Optimizer Wins with Traffic Segmentation

There are so many data elements to test, it gets pretty overwhelming. That is why you need committed resources for testing and make sure that you have testing/optimization part of the process rather than a project.

In preparation of website optimization, the success metrics are not always increase in conversion rate or lift in sales. Perhaps, if you company spends a lot of hours in deciding the creative (e.g. home page banner), then doing the test could prove that testing helps reducing the time to decision. Those reduction in hours spent to decide could be a key indicator to a success in website optimization.

Enjoy testing!!

Typically these redirects (e.g. example.com/tv) would take users to a landing page without having to depend on the users to memorize the entire URL string.

You'll typically find them redirects used on TV ads, prints, radio, etc.

The objective of this article to bring the awareness that redirects could cause various issues including hidden actionable data that you could be missing out. This is a common culprit for big companies where they're opening redirects left and right, but never manage to close it or redirect it to new page.

What are the common implications for not managing the redirects correctly.

• If there is no campaign ID assigned to the redirect, you won't be able to properly assess how much that redirect is driving traffic to your site. It could be tracked under "direct" traffic in the traffic source report.
• It could be leading traffic to 404 error page because landing page has been killed, but redirect continues to drive traffic.
• You could be loosing a lot of opportunity to convert the visitors because the landing page is not optimal.
• Multiple redirects to same page with different campaign IDs attached to each URL. That could cause some issue with SEO, because same content could potentially be indexed or URL with camapign ID will be indexed and organic search traffic will counted as a campaign traffic.

To understand how much traffic is driven from a particular redirect, make sure have the campaign ID assigned to it. So in your 301 redirect file, make sure the destination page contains the ID tag. For google analytics, it would be the utm_ tracking codes. Here is a URL builder page you can leverage: Tool: URL Builder for Google Analytics

Make sure you have your IT department or engineer in charge could provide you with the full list, so you can understand what redirects are driving how much traffic and conversions.

Check the bounce rates and see if the landing pages are still valid. Is it a redirect from old campaign to a landing page that no longer exists? Take to some where relevant, and test it if you can if it is driving significant amount of traffic.

In this redirect optimization plan, landing page is the key. Since the redirects contributing traffic are contributing traffic for reasons. For example, It could be placed on another sites or blogs, book marked, etc. Therefore, direct them traffic to some where relevant; if it was directing traffic to a Product A under Category A, you could potentially redirect that traffic to Product B or landing page of Category A, only if that original landing page Product A does not exist anymore.

Another point is, make sure the landing page is specifying canonical link so that the page will not be indexed including the campaign ID. Here is a great article from Google's blog that describes this; Specify your canonical - by Google Webmaster Central Blog

Optimization does not only happen on creative, landing page doing A/B or MVT tests, but it could be done with significant impact on your bottom-line; by reviewing and taking action on fixing your redirect strategy.

I thought I would give a little different perspective on how we typically look at traffic data, using monthly visitors as an example on Z-Chart.

Before diving into looking at Z-Chart, let's quickly go over what Z-Chart is.
Z-Chart is a method of short/long term forecasting by plotting cumulative and moving total figures for any measured value across time on one chart.
There are three lines represented in the Z-Chart:
(1) Current line: It shows the variation for each period
(2) Moving total: Smooths out any seasonal variations
(3) Cumulative line: Joins these and may show progress towards a goal

The name Z-Chart arises from the fact that the pattern on such a graph forms a rough letter Z. Typical application of Z-Chart is on sales figure, but let's see this from web site traffic perspective.

Here is a rough example monthly visitors data from a site with positive growth, and another one with not so exciting traffic growth. Evidently, Z-Chart will tell you if your trend is sucking or not. Sharp positively increased Z (left image) shows some significant growth trend, while sagging Z (right image) shows the site traffic is not doing well as the other.

Note that the Z-Chart could be dissected to remove some of distraction like the cumulative total that gives an impression that the site traffic is growing. (Obviously 2009 data on the right chart is much lower than 2008). The idea of having these two Z-Charts side by side is to show how it would look when two different data are trending differently, and to apply general web analytics approach comparing one segmented data to another. (e.g. overall traffic vs. product page view).

Here is an idea for web analytics. If you can apply overall traffic and compare it against segmented traffic. For example, overall monthly site visitors for one Z-Chart and monthly visits to product pages on another Z-Chart. What does the difference look like? If it looks like the example on this post, then there might be something broken with accessibility or acquisition strategy on product visibility. That raises a question like "why is my product pages traffic are not seeing the same Z shape as overall site traffic?". Then you need to further dissect the chart and data.

There are many charts related postings out there and it is definitely worth reviewing them. You could be dissecting these type of charts in so many ways, and find that detailed insights are not immediately available form this one single chart. Going to leave that expertise to the chart gurus. :-)

What does your Z-Chart show you? Feel free to give me a holla back on twitter @k_irizawa.

Some links on Z-Chart related topics:
http://home.ubalt.edu/ntsbarsh/Business-stat/stat-data/Forecast.htm#rzchartforcat
http://syque.com/quality_tools/toolbook/Line/vary.htm

Very nice tutorial on excel and charting techniques.
http://www.peltiertech.com

For example, Google's browser size will show you a line with "90%" contour, which means that 90% of people visiting Google have their browser window open to at least this size or larger.

With more internet users' attention span and focus getting narrower, these tools are very useful to optimize your site to improve the outcome.

Here is what it looks like when you use it on Amazon's Kindle product page.

My screen resolution was set to 1280x800, and the page with Google Browser Size resolution marks account for browser and my Windows vertical toolbars.

What's interesting is that according to this overlay, it tells us that with this browser size setting, 50% of the Google visitors who visit this page will see partial "Add to Cart", "Add to Wish List", and "Add to Shopping List" buttons.

Using your web analytics tool (Google Analyitcs, WebTrends, Omniture, etc.), you can gain additional perspective on what your visitors will see for your site, including visitors who come from Google. Your web analytics data will allow you to understand where your visitors are likely to fall under in these marks accounting for Google's visitors who are likely to be more a major contributor of site traffic.

Taking this page as an example; hypothetically, let's say the majority of Amazon's visitors set their browser size to 950x500, that means, according to Google's data, 90% of the Google's visitors set their browser at least in this resolution.

Another example, if the majority of Amazon's visitors set their browser size to 1280x800, then 50% of Google's visitors will at least have this browser resolution.

At the end of the day, your site visitors and their screen resolutions are what matters the most. However, considering that Google is one of the most visited site in the world online, that is also telling you what resolutions are likely to be optimal. And that should matter to you if you are trying to acquire more visitors to your site.

Segmenting your resolution size data by traffic source could give you different perspective as well. Reason why we should take these different perspective into consideration is; consumers could and most likely start their search from Google (or other search engines), go directly to your site by typing the URL directly into the browser, to research before purchasing your product on your site, or follow the links thru various sites starting from search engines.

This Google tool can give you additional view on what general audience have their browser resolution set, and give you a good sense on how optimal your key call-to-actions are positioned within the site to search general mass online audience.

With more and more consumers going online using browsers thru mobile or iPad like devices, the resolution size is likely to get smaller. Taking screen resolution into your content layout strategy will be even more important than before.

Here is the link to Google Labs' Browser Size: http://browsersize.googlelabs.com/

For example, your web analytics might be integrated to your Microsoft SQL server, Oracle, SalesForce, or other enterprise solution; you could be feeding site visitors from your web analytics tool and run it against sales or market data. (CRM and supply chain data integration with web analytics are pretty common in ecommerce.)

When your BI solution goes through different changes or version releases you got to make sure the joins and data integrations aren't affected by it. In my experience, common challenges are filter requirements or scalability of data across multiple sources due to difference in definition of data and its application within the original source.

Good example of this is reporting different sales regions. You can apply different data attributes to identify regions. Here are some examples:
Page tagging values like Country code, Cities, Continent, Regions (AMR, EMEA, APAC), web analytics visitors data based on IP address, language ID, etc.

So when you mesh two data from different sources, using CRM and web analytics data as an example; CRM data may have user profile for where they reside (e.g. California), and web analytics data may feed traffic data based on country/language value tagged on the page (e.g. us-en). What that means from BI integration perspective, you need to have mapping table that maps California with us-en, so BI reporting will properly scale to recognize two data sources when users select "USA" from the BI reporting.

This example was a simple one, but it gets more tricky when different sources are involved (imagine when you have different subdomains or data managed by 3rd party, that you want to integrate...).

My point is, depending what the business goals and reporting needs are, you need to be careful, well planned, document the requirements, caveat the data, and do what ever it takes to not waste resources so at the end of the day you're not spending more resource than the value that will be expected out of the BI solution. Why I am saying this when it should be a project manager's work and not analytics analyst? At the end of the day you're likely to work with project manager to feed web analytics data (possibly other sources of data, too), and web analyst will be doing the analysis out of that BI solution and be accountable for that reporting/analysis.

All of that careful planning and execution makes web analyst valuable when involved in such BI projects, and got to be accountable for the reporting end when you use that data as an analyst. You definitely don't want ad hoc changes and poorly managed change management process to affect your analysis work. Good data coming into Web Analytics tool and feeding good data to BI solution can still yield bad data going out.

"Great data in Garbage out" is possible when you don't have the right change management or poorly planned BI solution.

When I talk to various experts in the industry, it is common to find few companies that do all of these integration with ease. If there is a company that did that well, a lot of props go out to them, and I bet they went through many challenges with many sweat and tears.

As web analytics get more sophisticated and integrated with many business data, web analysts will need to be more wise and educated in this field of data management.

Definition of change management (from Wikipedia):
Change Management is an IT Service Management discipline. The objective of Change Management in this context is to ensure that standardized methods and procedures are used for efficient and prompt handling of all changes to controlled IT infrastructure, in order to minimize the number and impact of any related incidents upon service. Changes in the IT infrastructure may arise reactively in response to problems or externally imposed requirements, e.g. legislative changes, or proactively from seeking improved efficiency and effectiveness or to enable or reflect business initiatives, or from programs, projects or service improvement initiatives. Change Management can ensure standardized methods, processes and procedures are used for all changes, facilitate efficient and prompt handling of all changes, and maintain the proper balance between the need for change and the potential detrimental impact of changes.

In this article, I would like to focus on tactics to increase KPIs relating to return visitors or loyalty customers. The common metrics in web analytics tools to measure loyal visitors are return visitors, loyalty metrics like frequency and recency, or number of advocates.

It is very common that conversions for eCommerce are likely to be higher for returning visitors to the site. That is because consumers are researching online and may visit the site again later to finalize their shopping. According to Foresee's research, "Compared to shoppers who are dissatisfied with an online experience, shoppers who are highly satisfied with a retail website are 65% more likely to purchase online and 44% more likely to return to the website than those who are dissatisfied."

That is a very convincing statement to support why you should focus on converting your customers from new to return and make them loyal to your service and brand. Obviously, your customer has to be satisfied to be loyal, but let's focus on the web analytics and the tactics and strategies to improve the KPIs.

Let's layout some KPIs, and discuss some example tactics and strategies to improve it. Return visitors, frequency, recency, conversion rate, revenue distribution, new and return visitors distribution.

These are common measures you can acquire from the web analytics tool out there in 2010. It is typically based on site visitors who were cookied or not. Caveat is that if users delete their cookie or switch their browser, they won't be recognized as return visitors; instead they'll be tracked as new visitors.

What to look for in returning visitors are the number and growth in returning site visitors, conversion rate, and revenue contribution and distribution. You don't want to take action before knowing where you stand. The number of returning visitors doesn't mean anything, but if you add some context, then it can tell you a lot. Here is an example of what you can start off with:

I just randomly made this up, so don't worry so much about the actual values. The idea is understand your current numbers, and where you stand. You might want to start from here and understand the growth and distribution of new and return visitors, and their revenue attribution. Ask following questions on how well your site is doing on:

1. New and return visitors. Are they both growing or not? In above example, your overall traffic is growing, but return visitors are not growing. Sad...
2. Is your returning visitor's distribution relative to overall visitors low or high? 20% of the site traffic is return visitors, and about 80% new visitors.
3. How much is the revenue distribution between new and returning visitors? 20% of your site visitors are returning traffic and they are contributing 55% of the revenue!!
4. Conversion rate for return and new visitors. What is the difference? Typically, return visitors are much higher than new visitors.

So from this scenario some the actions you can plan and take are:

• Grow that number of return visitors. You need to look at recency metrics and continue to improve that number so return visitors are coming to your site in shorter time span. Market to them, and continue to offer fresh ideas, content, products, price, features, and services.


• Improve conversion rate for new visitors as much as you can. If you set up your site so users can opt-in to receive future message from you, you can remarket to the converted customers.


• Improve conversion rate for both new and returning visitors, and do it through testing. Always test, test, and test. Testing in this case is web site optimization through A/B, multivariate testing, etc.


• Don't worry so much about new and returning visitor's revenue distribution. Think about the growth of retuning visitors, because that means your strategy to bring in more loyal customers is working.

Few tactics I've mentioned here to move that KPI's are:

• Fresh content, services, products, ideas, right price, features, etc. You have to have something to spark the interests of your consumers so they come back.
• Re-marketing. Re-market to current customers because you know they are likely to buy from you again. Re-marketing could be done through newsletters, emails, calling, sending out coupons to current customers, DM, etc.

After quickly searching on Google, Yahoo! and Bing, I could not find relevant content relating to Web Analytics for Customer Support and Care. Some may blame me that I didn't do a great job searching for such studies and writings, but I am definitely not happy with the results I found on search as of the date of this writing. So I am going to brainstorm an initial roadmap on web analytics for customer support and care sites.

Outcome

Obviously, companies are trying to increase customer loyalty by providing great services, support, and care. From a corporate web site's perspective, there are 3 main goals to achieve in increasing customer loyalty.

1. Reduce support calls & cost.
2. Customers finding the solutions they need help for.
3. Serve relevant solution and care through the site.

KPI - Key Performance Indicators

Support Calls & Cost

• Number of support calls for specific reporting period
• Total support cost for specific reporting period

Customers finding the solutions they need help for

• Number of entry visits to support section by "Support Related Redirect" and "Search Engines"; Increase direct entry to support pages from sources and channels where customers with problems would likely to start from.
• Bounce or Exit Rate on On-Site Search and other initial point of interaction that takes users to list of potential solutions (like FAQ page); Customers will leave (bounce) if the site serve suboptimal results, so it is imperative to measure if your consumers are findings the relevant solutions.

Serve relevant solution and care through the site

• Customer satisfaction score (thru surveys)
• Number of Task Completion and Rate per key interaction pages/points (micro conversions); Identify pages and customer interaction points that tells us the customer experience -- Example: Feeback on Helpful Yes/No, Visits to Email Support Confirmation Page, Reduction in click-thru to contact us (page with off line number), etc.

Strategies and Tactics

I feel like most of the companies are just throwing FAQ pages up and not optimizing the results. However, there are companies out there doing great job in executing the customer support site, and by looking at the analytics tracking code, you can get some sense of understanding on how they're positioning their web analytics tool.

Also I feel like strategies and tactics are executed without having a plan or roadmap. Maybe many major firms do have a clear roadmap, but I don't find many discussion about it. I hear more about measuring conversions, engagements, SEO, PPC, site optimizations, etc.

The critical point for customer care and support from web site perspective other than lowering cost and increasing user satisfaction is; "find-ability" of the solution. Theoretically speaking, customers should be able to find the answer and solution to their questions with ease. Perhaps finding the solutions directly from search, and from both Search Engines and On-Site Search. I think Apple has done a great job at least from the outside (don't know anything about their data). Check out the forums and FAQ pages' title and how they are indexed on search.

Other major firms' site are great source to check and see how you your site's usability compare. In order to avoid subjective views, "Search Engines" are great tool to measure it. Example -- looking at what contents are index using "site:forums.xyz.com" on Google, and checking out their robots.txt files, using google search insights, competitors' twitter account using twitter measurement tools, etc.

Customer Support and Social Media

I feel like I read more and actually find interesting in this Web 2.0 era is Customer Support + Social Media rather than using web analytics to measure and optimize the web site's property.

Here are some articles on Customer Care/Support & Social Media (mostly Twitter)
Comcast: Twitter Has Changed The Culture Of Our Company
HOW TO: Use Twitter for Customer Service
Customer Support via Twitter?

From web analytics point of view, there would be additional KPI assigned to the road map if customer care and support becomes more integrated and practiced via Social Media, but I will hold that conversation for another time.

A conversion could be anything, it could be an instant that defines the moment of success.
Here are some examples:

• Orders
• Subscribed for blog feed
• Email subscription
• Clicked on twitter, linkedin, about us link
• Generated product review/testimonial
• Shared content (advocated)
• Clicked on on-site ads
• Acquired a lead

When you segment your data based on a specific conversion, you will find various segmented data sets yielding different results in KPIs. Assessing landing pages based on conversions are great because it allows you to focus on the first page that convinced your site visitor to complete a desired action/outcome.

Looking at this example chart, you can definitely see the head and the tail, which is based on the cut off you define. Let's say 50% of the revenue was contributed by top 5 pages. Given that there are total of 35 landing pages that contributed to conversions, it is insightful to know that these 5 pages helped you generating 50% of your revenue!!

Biggest question is, what strategy are you going to introduce to optimize the other 30 pages? Even if you improve (theoretically) all of these 30 pages to convert better, the likely hood for you to see this curve from head to tail will always exist. However, it is very important to continue and set goals > measure > report > analyze > optimize > set goals (repeat).

What you need to look at when you optimize the long tail pages are:

1. How much did your optimization plan and action supported improving the number of orders in the tail segment, and revenue growth from these long tail landing pages.


2. Engagement and visitor type difference in head vs. tail. How did your optimization impacted these visitors.


3. Bounce rate improvements based on your optimization to these critical and key pages.

You may ask, what about the pages that aren't showing up here in the landing page's long tail chart? If those pages are so important that you think it should contribute to your site's success, then it is a clear indication that the page is very ineffective, and it is a call-to-action to do something about it.

It is imperative to take these number of entry pages (segmented by orders) and divide it by overall number of entry pages served. For example, if you had 100 pages on your site and 90 of them were your entry pages, then we can say 39% (35/90) of the entry pages are contributing to conversion or your desirable outcome. This is a good metrics to see the allocation of effective entry pages that are supporting your site's goal. A great bench mark to gauge after executing your optimization strategy.

The most important thing is to see how your campaign and optimization strategy impacted the landing pages, and I am sure you'll see some difference in this data matrix.

Good luck analyzing your landing pages, and I'd love to hear your thoughts. If not on my blog, maybe through Twitter.

Recently, I have been working with web developer for one of our new site. It is always nice to be working with a developer who is sharp and with full of interest for web analytics. These developers work so hard to help us web analysts get proper tracking into the site. This article is a bit of a dedication to those who help us analyst get the data that puts bread and butter on our table.

Here are some scenarios I would like to highlight where web analytics will work very closely with IT/programmers and integrate to drive better web site experience for your site visitors. (I only highlighted the things that came up in my mind at the time of this writing, and skipped the obvious part, implementing analytics tag.)

On Site Search

It is common to find more major enterprise to invest in on-site search (or internal search) technology. An example is Google Search Appliance. Obviously, the IT team in your company would be installing this to the server.

Web Analytics analyst should get involved in this, because even if the appliance has some nice data around what people are searching on the site, web analytics application should have standard metrics relating to the search results, and eventually assess behavior data and impact to the outcome (bounces, orders, revenue, and many other interactions generated from on site search).

Depending on the search results and the outcome, web analytics analyst should work closely with the IT in optimizing the appliance, so relevant search results are given to your site visitors. Potentially what you do in this optimization can impact the outcome as well.

Error rate

Launching a new site, replaced shopping cart, replaced old products with new, ending a promo, etc. For whatever reason, there are chances that your site visitors are landing on error page or page showing that the content is not available. Typically the server will send back a 404 error message. That is why it is usually error pages are referred to as 404 error messages.

The majority of Web Analytics applications have the capabilities to track such visits to error pages. Web Analytics analyst should know what is the % of error page served. Hopefully you are aiming for 0%. Understand where visitors are coming from to arrive at the error page, and with what URL.

All this information will help your IT folks to set up a proper redirect, or even get communication started with product/content managers to serve or activate relevant content to replace that error.

This is a great article on 7 Ways Of Handling 404 Error Messages. Checkout what some companies have done to address 404 error messages.

Redirects

Major corporations investing heavily on online/offline campaigns are likely to be setting up redirects to direct site visitors to relevant landing pages. Redirects would most likely be set up by IT, and web analytics would be working with the marketing and IT folks to set up appropriate campaign tags to assess marketing efforts.

It is not only important to assess the campaigns/redirects, but to understand if there were any impacts when redirects were taken down or if redirects' usage has changed (re-using old redirects for new campaign, etc.). Having a transparent communication process in place for managing redirects would be crucial to a success in web analytics practices.

Other Area of Technical Web Anlaytics

There are many other web analytics data that will help programmers and IT folks to understand what system environment the site visitors are in. Some of the metrics are Browser Versions, Operating Systems, Connection Speeds, Flash Versions, JavaScript Versions, etc.

It might be a good idea to sit down with the IT and Programmers before any major releases take place, and review if any upcoming programming or changes in infrastructure could impact site visitors or customers' site experience.

Sharing Results with the Programmers and IT

Programmers and IT folks deserve to know the results, too. When web analytics analyst creates report for managers and key business groups, it would be great to share the results with the technical people as well. I had the privilege to work with smart IT and programmers in the past, and sharp technical folks ask brilliant questions which in many occasions should have come out of marketing managers.

Having a conversation about optimizations and exchanging ideas with technical folks can benefit web analytics analyst in many ways. For example, having their resource to tweak on-site search and increase revenue can definitely make our contribution to business shine.

Considering that there are so many metrics and dimensions to deal with, just by tagging Google Analytics tags on every page on the site. Analysis and reporting could get extremely complicated.

What I recommend is to relax and look at the basic metrics, but applying smart segmentations. If you're going to puke before looking at the data, might as well puke after looking at bad data.

I've asked my family's friend to see if I can check out e-commerce data and use it anonymously. So here you go.

Very high level data, including overall site visits (which something we may not care at all). The important thing is to leverage available segments and apply few basic custom segments to better position yourself in a pool of data.

What I did here is applied "Visits with Transactions", "Non-bounce Visits", and "Users who visited the shopping cart, but did not purchase". It is amazing that even a site with average bounce rate of 32%, can have a significantly bad data that is puke-able.

First, visitors adding item to cart was only 5% of all site visitors. I think that's pretty bad, but let's assume that is good. Now, what about only 2% of the shopping cart visitors completing the order (0.1% of all site visitors)?

This pretty much paints the big picture of different level of opportunity pie, and how much qualified visitors you're brining into the site. Is this still good or bad? I don't know, maybe the average order value is super high ($1,000 or more..) and this site is making huge money... but that's something you and your business need to decide and set goals.

Looking at average pageviews with those segmentations (same as pages per visit), it is amazing that visits with transactions are going through roughly 31 pages just to convert. Notice that visitors who hit the cart page, but didn't ordered was 17 page views. That means people who hit the cart and ordered are going through 14 additional pages on average. Definitely some opportunity to optimize the conversion process. What does your data show?

Average time on site shows a similar story, and it is quite obvious that customers are going through some rough experience just to convert.

Just applying simple segmentation to high level data like visits, page views, and average time on site, can give you some very high level, but very insightful data to initiate of some more data diving.

One of the ultimate outcomes for a website is customer satisfaction. First site I think of e-commerce site with great customer satisfaction is Zappos.com. It is one of the ultimate models to service and satisfy customers. Zappos definitely made me think about customer satisfaction more than before. Before we think things in complicated manner, I think we need start somewhere and understand the measurement points for web analytics, because web analytics is the bread and butter of this blog.

Check out this diagram I made. Apologies if it is not pretty...

When a business owns an e-commerce site and deals with web analytics, we (web analysts) need be mindful of the key points that support and complete the e-commerce service. Allow me to share you my idea around three web analytics action and measurement points:

Point 1 (Website and Customers)

This point shows the interaction between customers and the website. In other words, it is the "experience" part for the customers going through your store. From web analytics point of view, we would measure and assess:

• Pages that the customers viewed that contributed to completing a key event.
• Entry pages with high bounce rate to understand what to optimize.
• Conversions.
• Visitor types and their behavior differences, and outcome analysis.
• Advanced segmentation analysis on various dimensions of data against outcome.
• And many other fantastic data we get out of web analytics applications.

These are typical metrics and KPIs that are available straight out of the web analytics applications, which requires customized tracking and set up for key events on goals. The business should know what KPIs should be gauged to understand how their site is doing against their goals.

Optimizations and testing will impact this area (point 1), so customers' interactions with the site will give us actionable data that supports decisions to improve customer experience. Measured and assessed data should tie with customer satisfaction because your website has to serve your customer with products/services/content they are looking for, and have your site optimal so that customers can convert with ease.

Point2 (Business and Website)

You are wondering what does business or website owner got to do with the actual site that relates to customer satisfaction. To be honest, after going through many websites, I've seen many situations where I doubt site owner's care and ability to maintain the site to serve its customers. (Maybe I am shooting my own foot here... lol)

Some of the actions taken by business owners could be adding/updating content, executing promotions, supply inventory info in e-commerce database, etc. All these actions could lead to customer satisfaction. Taking the supplying inventories as an example, I've came across sites where when I put an item into cart, it'll show "out of stock". Leaving me with no choices, but to leave the site.

From web analytics stand point, we would love to measure everything that impacts the bottom line, but here are examples of metrics and actions pertaining to customer satisfactions:
Inventory availability vs. demand orders, Promo response and application/redemption, site optimization A/B or multivariate test, fresh content uploads rate vs. avg. visitor frequency, applying SEO best practices, etc.

Point3 (Customers and Business)

I believe this is an important point as well as other common web analytics involvement affiliating with customer satisfaction. After customer purchase items from the website, orders have to be fulfilled, issues have to be solved; direct marketing may take place online or offline, etc. The speed and timing of fulfillments, customer supports, and customer communications are pretty critical to customer satisfaction.

From web analytics stand point there quite a few measurement points. Here are some examples.

• Direct Marketing (online/offline): Delivery, Response Rate, Conversions, etc.
• Fulfillment: Ship to order ratio, average order to delivery time
• Support: Customer response, incident reduction, support forum usage
• Direct testimonials to business, survey request to customer and analysis

Hopefully, this article gave you different take or angle in viewing customer satisfaction relating to e-commerce. Some of this concept may apply to non-commerce site, too. I guess having a creative mindset in web analytics field does not hurt. That's what I believe in.

In this writing, I would like to focus on remarketing with email. Keep in mind that in order to remarket via email, people have to have registered on your site or who have opted into your email program.

The best starting point is to review the type of site you are running the remarketing program. Your site could be an e-commerce, content, lead generation, etc.

Second point, what interactions by these identifiable visitors would you consider as abandonment? Abandonment in this case could be anything that you consider people who didn't execute the desired action, after subscribing or registering for communication.

Examples based on an assumption that the site visitors are logged into the site and registered for email communication:

• People who added items into shopping cart, but did not completed checkout.
• Visited a specific page (like promotional page), but didn't complete a desired action like subscribing/registering/applying for promo.
• Didn't sign up for magazine subscription.
• Didn't bounce, saw old product, but missed viewing the new product pages.
• Registered users with twitter account (based on profile info), but didn't click or attempted to show interest in following the site owner or company's twitter account.

You get the idea. Remarketing could yield positive results. According to a research found on a study for ExactTarget and TicketsNow, researchers found that transactional emails sent after a shopper had abandoned items in a shopping cart actually worked to improve overall company revenue by 30%. Therefore, remarketing could be an effective marketing tactic based on people's interaction with the site. So far, re-marketing sounds great and very promising.

A lot of the findings based on research talks about X% improvements in revenue or increase in conversion by Z% after doing XYZ. That makes me wonder about the kinds of web analytics data, which should be reviewed prior to investing your time in executing remarketing program.

Here are some ideas that I think web analytics metrics can help you decide rather to tackle remarketing with email or not. Idea is to understand where your site stands in terms of data, and gain a better sense of expectations from remarketing.

Number of registered users who can be emailed:
You site may be selling your products pretty well, but if a lot of them shoppers didn't opt-in to your newsletter (for whatever reason) then the overall volume of re-marketable people may not be significant.

Imagine an e-commerce site where the owner decided to give customer an option to opt-in for email communication at the thank you page, which caused only 1% of the overall customers to register. Let's assume there were 10,000 absolute customers to date. That means there are only 100 remarketable customers. Is that a good number for you to invest in re-marketing? You decide.

Email click-through rate (unique):
If your average click-through rate (unique clicks per email delivery) was around 1%, then only 1 response would be generated from 100 emails that you sent out (based on the above example).

If your site is awesome, and have 1,000,000 absolute customers, with 1% of them registered for email, there will be 10,000 remarketable customers. With 1% click-through rate, you have about 100 responders to your remarketed email.

Conversion rate:
Self explanatory, but let's assume your re-marketing tactics are so great (the best in the world), and 10% of the email responders actual complete the desired action. With the case with 1,000,000 absolute customers, you'll get 10 conversions.

Average order value (AOV) or Average Conversion Value (ACV):
Regardless of your site being e-commerce or not, it would be important to understand how much your conversion are valued. Say your site sells an item that worth $100 on average, then that will be your ACV or AOV. With the 1,000,000 customers example, that will mean your remarketing efforts may generate about $1,000 in value. Not sure if that is good yet.

Return on Advertising Spending (ROAS):
So you may be happy that $1,000 is a great outcome, and now you can go buy your team some drinks... but wait. Ask you self, about how much money are you planning or expected to go out of your pocket for this remarketing program?

Let's say you paid this one freelancer that use to work at a top marketing agency for about $1,500 (I'm sure it is more in reality...). Using above scenario, your ROAS is 67% (1,000/1,500). You basically lost some money just running the re-marketing program. Nice... maybe that's why you hear a lot of fancy agencies having trouble making their clients happy because clients might be seeing this ROAS below 100%.

Additional/potential value generating from re-marketing outcome:
This is where your segmentation in web analytics tool come into play. If your remarketing efforts are to drive certain outcome or interactions, what does your web analytics segmentation tells you about those visitors after they converted or show potential additional value?

Say you segmented those 10 converted customers (visitors coming from re-marketing email), and found out that they advocate the site's product through tell-a-friend feature. You happen to know that tell-a-friend or advocacy by customers generate about $50 in value (this is just an assumption for example).

Therefore, 10 customers telling 10 of their friends generated additional $500, making your ROAS 100%. The idea is to look at the segment of different interaction and outcomes so your remarketing efforts are based on interactions that drive the biggest return.

These interactions could be existing even before you run the re-marketing program, so understand the values from different "micro conversions" using segmentations. Here is a nice article that talks about "Macro and Micro conversions".

I guess the challenging part is to assign some kind of $ value on to those micro conversions, or those segmented interactions that leads to additional returns for your site/business.

A visual to support my writing...

When a dashboard is shared across different teams, then a specific group with special interests will start to ask more than what is in the dashboard, because their job duty is to support specific tasks within the business life cycle of the online business. In some occasions (probably most of the time), for example, when you create a web analytics dashboard for CEO, you'll always hear from some other team manager saying that that dashboard doesn't help "me" in anyway, and start to go off about what would be a great dashboard for him/her.

In a typical scenario, honest and hard working web analyst will try to make everyone happy, so he/she ends up making either an obese dashboard containing bunch of colorful charts, or bunch of redundant reports catered to every single managers in the company. You think a web site with one goal/objective would have a clear and simple dashboard for all to accept easily created, but it could get tricky. I believe it is important to think of dashboard as a starting point to dive into a pool of data.

The best way to create an effective dashboard, following points should be taken into consideration:

1. Understand the bottom line and the desired outcome of the web site.


2. Ask, why do each key business units exist in your firm and what is important for them to execute their job?


3. Know the kind of decisions your end users or business partners can make based on your dashboard.

Effective dashboards usually include:

• An intuitive graphical display that is thoughtfully laid-out, and easy to navigate.


• Logical structure so information is easy to consume.


• Regular and frequent updates of dashboard for relevance to current conditions.


• Answer fundamental questions about the web site's goal or overall business unit.


• It alerts issues or problems in such areas (error pages, ROAS, revenue, conversion rate, etc.)


• Supports decisions that impact the business or online strategies.

Killing multiple data needs with one dashboard

I am just using this phrase to push the minds to think in an effective way, so that KPIs are reported effectively to reduce redundancy. Check out this image of a scenario where you have multiple business units involved in a web site, and how they could have more data needs than that one Executive Dashboard which was created to make CEO smile and move.

The images may imply that web analyst will need to make 5 dashboards at the end of day. However, if you craft the dashboards effectively, you could potentially consolidate them to fewer reports. How you do that will depend on web analyst's skill to work with those key players and identify them KPIs for an actionable insights.

Here is a basic framework that I crafted to support approaching in creating that awesome web analytics dashboard (Tier 3 is not necessarily a dashboard, but most likely a report or an analysis.). It is intended to demonstrate that your dashboards or reports should be positioned and focused on catering the KPIs to appropriate audience/groups, and show different tiers of audiences which could reflect organizational structure with different focus on reporting needs.

Side note: If you give a web analytics access to each business units and allow one smart and motivated person to do their team's web analytics support, you can build allies of web analytics specialist. This could be a long term strategy to build web analytics culture within your company.

I hope you don't underestimate the impact of screen resolutions to the conversion rate. I have seen many sites where the call to actions or the important messages are under the fold.

Notice that height with 768 ranks second for this one site. (I have seen 1024x768 rank first for some sites)

Here is an example of a CPG site. Perhaps selling is not their main goal, but they have a call to action "Buy Online", so let's assume that is their main site objective for this case.

Notice that the red box (call to action) goes away when I changed my resolution to 1024x768. They are increasing the chance for visitors to bounce, because it makes the gap between the customer intent and that landing page's objective to widen.

1280x800

1024x768

I have seen cart add ratio decrease tremendously when call to action to buy is under the fold for specific screen resolution. I guess it's time to review your screen resolution and make sure that the drop off is minimize by a simple redesign.

I would also have to mention that not only people adjust their screen resolution differently, but browsers can also allow users to set different zoom percentage. That also affects your site's performance, but at this point, it is probably wise to test and find out. Good luck!!

I am not sure if there are a lot of companies out there that relies on the web analytics applications to integrate external data and report it out of the web analytics application other than the Business Intelligence tools.

The idea of integrating and assessing offline metrics to web analytics data is nothing new, but I feel like I don't hear much about offline metrics that impacts online business and strategies.

In this post, I would like to talk about some of the key operational metrics that web analyst or eCommerce managers should look at when assessing the success of your eCommerce business.

In web analytics, we focus a lot on marketing, traffic source attributions, conversion rates, etc. It is vital to also think about your customers' experience beyond the website or after they completed the transaction. In eCommerce, there are a lot of things happening even after customers completed purchasing. Orders have to be processed, transactions will need to be settled, products need to be wrapped, shipped, confirm delivery, subtract inventory, process returns, etc.

All of these required actions or processes correlate to customer satisfaction. Sure, you can improve your traffic by 200%, and increase conversion rate by 300%, but if your customers aren't receiving the products within promised timeframe, then they will not become a repeat buyer or even advocate your service. Here are some of the metrics you will need to take into consideration.

Shipped to Order
For a given time range of data, it is the percentage of shipped orders per total orders. If one of the weeks in previous months show less than 100%, then maybe something is wrong with the fulfillments or the orders aren't getting processed.

Duration between Orders to Delivery
If your slowest shipment option in your eCommerce service is 5 to 7days, you definitely don't want to see seven plus days for this metric. Your customers do expect to have their product delivered on time.

Number of return orders and percentage of returns
So you have increased your conversion rate and successfully optimized the campaigns to drive higher revenue, but you didn't realized that more products are getting returned due to whatever reason. When you're working with a great eCommerce solution, you should be getting back the number of returns. Make sure to gauge this key data.

Percentage of in-stock per order
Maybe your site is selling hot out items quickly, but the orders surpassed the amount of units available in inventory. You want to know what percentage of orders are actually available for delivery. If its below 100%, you might want to double check which products are short on inventory, and for what reason is it short. (sky rocketing number of orders for particular item?)

Remember, that one of the key desired outcome of websites is to "increase customer satisfaction and increase loyalty". Make sure to close the loop in your supply chain/logistics/fulfillments by taking action upon these key metrics.