eMetrics appreciation

Back from a fantastic trip in San Francisco for eMetrics. Catching up on emails and work. But I thought I should share my impressions while things are fresh in my mind.

Ok, I must admit I might be biased... as I presented on the Industry Insight day and was moderating one of the track on Tuesday.

BUT... It was an amazingly productive conference.

Way back then...

Some years ago (well... many years ago... I'm getting old!), each Internet World conference brought something new to the forefront of the industry. Remember "push technology", VRML, the early days of streaming? Without any coordination, "the industry" was moving in a direction. Although we can now laugh at some of those concepts, they nevertheless changed the face of the industry and how we use the web today.

The same pattern is happening at eMetrics. Without any preliminary coordination, there are some things setting the path for the future.

Testing & beyond

I noticed this year eMetrics was a lot about the value of "testing". Bringing the "testing" culture and the right tools to do it in order to optimize and achieve success. Contrary to most other field of expertise, the Web allows us to deploy quickly and continually improve. You don't want to do that with your car, your house or the space shuttle... but with the web it's how it should be. Most companies don't understand that and still impose strict project cycles, those who understand are not only demonstrating huge benefits.

From IT, to marketing, to business

The other outcome, highlighted during the Industry Insight afternoon round table and in Thomas Davenport's great keynote is the transition from web analytics to business analytics. Just like the web, web analytics started in IT, then marketing found out about it and took control. We are there now. But winning businesses understand the value of the web and have optimized some of their most important business processes around it. We are maturing to a level where we won't only talk about using web analytics for marketing optimization, but we will be talking about analytics for business processes optimization and strategic level changes.

Being a tutor of both web analytics and business process analysis classes, it's obvious to me there are very strong benefits in leveraging analytics to optimize business processes.

What will be this fall highlight? Next year?

Industry Insight

Jim Sterne told me the experience of the Industry Insight was very positive and will be renewed this fall in Washington: leading experts sharing their view of the current and future state of the industry. I will be there to bring some hard core data about the vendor market shares and exchange with fellow analysts.

WASP: Listening to VOC

One of the challenge with a R&D project like WASP and listening to the Voice of Customer is the number of ideas that comes in and how to sort through them while keeping a vision. Every week I get great feedback, suggestions and comments. Of course, sometimes people will send in a few critics, but that's part of the game isn't it?

Please allow me to share some of those ideas and what I intend to do with them.

Browser extension: yes

From the start I envisioned WASP as a browser extension. In my opinion, it doesn't only makes it very easy to constantly be "in context" of a real user session but it also allows for easy ad hoc testing of transactions, secured areas and development environments.

Actually WASP is only available as a Firefox extension, but you can expect an Internet Explorer one in the future.

Site crawling: yes

I've been running a quick poll for a while and that was one of the question. 43% of those who answered asked for the possibility to crawl a site and report on the tags and implementation.

That's coming up in WASP v0.40, currently awaiting approval from the Mozilla Foundation.

1. "Crawl from here" recursively spider a site
   
2. "Load from file" allows you to pick a file containing a list of URLs to check

Reporting: yes

As you browse (or crawl), WASP creates a session log. The resulting CSV report contains the URL, the type and name of the solutions found on any given page, the page title, the exact tags and some other information. This report is available anytime from the "Save log" feature.

Web service or hosted solution: not for now

I'm not planning to create a remote crawling or web service solution for WASP, at least not for v1.0. There are some solutions like Browsercam.com that will allow you to test remotely and simulate different browser/OS environments. Or the great DejaClick.com Firefox extension from AlertSite.com that allows you to build a test script that you can run automatically (and WASP will record correctly). More advanced tools like HP's LoadRunner or MS Visual Studio Team System Test Edition brings the full range of load testing, unit testing and the tools to manage the QA process and track bugs.

Transaction simulation: not for now

Once the site crawling feature is there, why not simply stick in a transaction simulation capability like what DejaClick offers? Two reasons:

1. there are already great tools to do that and WASP will correctly log the details
2. WASP is currently quite simple to use, adding such a feature would divert too much from the original intent

Just use the DejaClick extension!

Flash & videos: future

As the industry evolves, tags are appearing more and more in Flash animations and videos. WASP doesn't currently detect those tags but this should come up in a post v1.0 version.

More tools detection: yes

The upcoming version now includes over 100 tag-based solutions in various categories: analytics, testing (A/B or multivariate), behavioral targeting, surveys, session recorders, click maps, etc. In fact, some sites have so many tools in place that the status bar and the number of tabs in the sidebar is becoming an issue (I will fix that in an upcoming release).

Please continue to send in your feedback when:

• you know a site as a specific solution in place but WASP doesn't detect it: send me the site URL and the tools that is supposed to be there
  
• WASP doesn't detect a tool and it's not in the list of known tools: send me a sample site the URL and the vendor URL
  

Market research: controlled

As is often the case with R&D projects, one of the unexpected side effect of WASP is the ability to do market research. As per the license agreement and the opt-out option, WASP is sending anonymous information about the domain and the tools found. The initial goal was strictly for debugging and enhancing WASP itself.

In fact, I'm receiving information about 20,000 domains every day, over 350,000 monthly. That provides a pretty good picture of the vendors market shares for various verticals and regions. Of course I can also do specific research from any given list.

This is highly valuable information for market and financial analysts covering the web analytics space, for vendors and agencies looking for prospects, or even companies looking at who is using which tools in their specific vertical. For the most part, this is how WASP will be able to survive as a commercially viable product. I'm sure you will understand I will keep a close control over this information and how it is being collected.

If you want more information about WASP market research please contact me.

Send in your feedback and suggestions

I have created a customer support collaborative environment to ease the gathering of suggestions and comments using a great new service called Get Satisfaction.

I've got some offers for help, people asking to open source the code (which I won't), even some investment proposals. I still need some time to continue the "incubation" phase and the planning (hey! I've been freelance only since December!).

But I'm listening: if you have any ideas for WASP or any business proposals, don't hesitate to contact me.

Winning content, spring break and analytics

Winning content

A while ago I wrote an article on box plots and whisker plots using Excel 2007. When I wrote it, I had in mind the use of this concept in the web analytics field, but it turns out a lot of people are searching for this information, especially students enrolled in statistical classes. Soon this article got referenced by de.Wikipedia.org and en.Wikipedia.org.

Hypothesis #1: Reporting on spring break

The thing is... I'm getting a lot of traffic, good for my ego, but this traffic is for the most part, "not qualified" for my blog goals: talking about web analytics. So here's what happen when students go in spring break: a huge drop in traffic in the last week!

That's reporting: making up an hypothesis and comforting our conclusion with some data extracted from a report.

Hypothesis #2: using analytics

Is this really the cause? After all, I know it's spring break in most regions...

Then I did this:

• Entry page: identify which pages got the highest drop in traffic: yes, the Box Plot page is the culprit
• Search terms: double-checked to confirm this was the root cause: yes, there's about a 50% drop from searches containing the word "plot"
• Referrers: referrers from Wikipedia, is there a drop: no?! Hmmm... why?
  
• Segmented search terms by continent: traffic is down by about 50% regardless of the continent... it's not spring break everywhere at the same time!
  

The real culprit: a change in my blog setup to shorten the page titles using only "immeria::"+post title has got my ranking to fall dramatically in Google!

Conclusion

Doing web analytics often implies starting up with an hypothesis. Then we fall short of digging for a complete explanation and we jump to the conclusion as soon as we have some data to prove us right. However, pushing web analytics a bit further implies using different tools and different angles to observe the same situation. Just like a scientist would describe something with different attributes and different qualities and a team of scientists would bring their expertise from various fields to provide an even more thorough analysis.

Troublesome targets: where analytics and audience meet

Every once in a while I'm chatting with my friend Joseph Carrabis. In just about a year since we've met at eMetrics San-Francisco we've shared thoughts and ideas, learned a lot from each others, even had fun flying kites in the nice S-F weather and had dinner in Quebec city just after a snow storm.

What began as one of our usual chit chat turned out into a great article Joseph wrote on AllBusiness.com: "Troublesome targets: Where Analytics and Audiences Meet".

The storytelling skills of Joseph provide a great example of understanding the role of web analytics as a tool to measure business objectives. But one has to first know and understand who's the audience and what those business objectives are! Than we can put the mechanisms in place to measure, understand those metrics and take action accordingly. What's of interest here is M.Carrabis company, NextStage Evolution, come from the sociological, psychological and behavioral angles while I come from an IT background, applying my expertise to the field of web analytics and online business process optimization.

The result is interesting: two spot lights on the same object of study, making the whole picture a lot clearer.

Quality assurance using WASP: tag all pages

In a post from a year ago, I was commenting on the challenges of JavaScript tagging. Here, I want to share some insight and advantaged of doing web analytics implementation quality assurance using WASP. This will take the form of a series of posts, each one addressing a specific issue.

Tag all your pages!

First and foremost, we need to make sure all pages are tagged. Obvious isn't it?

What strikes me when beginning with a new client is how bad their web analytics implementation is. Missing tags is the #1 problem to look into. Sadly, the area that are left out are transactional areas: the outcomes! Why?

• It's more complex to tag transactions beyond the mere "page view" tags
• Content areas are often rendered out of templates, transactions require case by case tagging to be implemented
• Transactions are often under the realm of IT and changing them implies "negotiation" for resources, timely delivery, tests and answering any security considerations.
• In some cases, transactions are on a different host and even a different technology altogether
  

The WASP advantage

Using a crawler such as the excellent SiteScan by EpikOne to check if all pages are tagged is a start, but it's not enough (plus, that's fine for Google Analytics but none others). A crawler won't be able to log into secure areas of your site, fill forms and execute transactions. Most crawlers will look for a specific string within a page, so even if the JavaScript code is there it doesn't guaranty it will be executed (more on JavaScript execution in a later post). Using the debugger provided by the solution provider, such as the one from Omniture, is fine for ad hoc tests, but hardly usable and often hard to decipher. Using proxys such as Charles, ieWatch or Fiddler is interesting, but they work at a lower network level and are way too technical. The other alternative is to open your wallet and work with Maxamine to do a complete and thorough web analytics, security, web compliance audit.

Because WASP blends itself into your browsing session, you will be able to go into any area of your site and make sure the tags are correctly implemented, even checking if specific parameters are being set correctly: custom variables, events, segmentation, etc.

Simply put, I'm from the school who think quality assurance of transactions can hardly be automated. You have to define test scenarios and diligently go through them not only to make sure the transaction work as expected and provides the appropriate results, but also to make sure you are measuring them correctly.

Coming up in WASP

The next release of WASP will include a couple of long awaited features:

1. Site crawl*: from any page, launch a recursive crawl of all pages on that same site.
2. List of links to parse*: already have a list of URL you would like to scan? Simply open the file and WASP will visit all of them and report back the results.
   

This is perfect to check the non-transactional pages of your site. For transactions, you will still be able to use the passive page checking already offered.

You will also get two ways of viewing the information:

1. Web analytics implement view: what's currently shown by WASP
2. Web analyst view*: showing simplified and plain English information about the tags found on that page.

Another request that was on queue: results are going to be available as a CSV file with domain, page, web analytics solution found as well as the exact data sent.

* Those features will be part of an advanced version of WASP available on a subscription base/purchase only.

Is Web Analytics too marketing centric?

Last night I was sharing some sushi with fellow web analyst and freelance contractor Jacques Warren, chit chatting about web analytics and other things in life... well, mostly web analytics. Jacques is a strong believer of integrating web data with enterprise systems (see "The Big Integration"), and I'm a strong believer of business process optimization through proper use of Internet, and more specifically, web technologies. That being said, I'm no marketing expert, I'm just a tech guy who got out of his shoes to be more business oriented, applying 20 years of experience in listening to requirements, analyzing possible solutions, and making recommendations.

This got us talking...

In my opinion, web analytics is being somehow "hijacked" by marketing.

Read on...

I don't think it's a good idea to imply "web analytics" is “marketing optimization” (as in the eMetrics conference title); I think we should not talk about web analytics, we should talk about “business analytics”. Just like the web itself, web analytics grew from IT roots simply because it was "too technical" (1994-97), then marketing got hold of the web as a marketing channel (1996-2002), but sooner than later corporations realized marketing was "one" function of the business, so they created ebusiness strategies encompassing several business functions , including sales, customer support, creating extranets and intranets and so on (2002-). Soon we won’t even talk about ebusiness, because in the end the “web” will be blended in all aspects of the corporate functions and culture.

Then, this morning I read the excellent post from Paul Legutko, at Semphonic, "The Future of Web Analytics Consulting" and a few minutes ago, the follow up from Marshall Sponder. What a coincidence!

I think the same type of transition from IT to Marketing to Business-wide will happen, probably sooner than we think. What we call “web analytics” today, which has somehow become the stronghold of marketing, will continue to evolve. In reality, what we want to do is “analytics” using all scientifically valid methods and tools to optimize the business, and that includes not only web analytics for marketing optimization as we know today, but make extensive use of "multiplicity" to optimize all functions of the business. It's certainly no coincidence that Thomas Davenport, a thought-leader in analytics and business strategy, will be a keynote at the San-Francisco eMetrics.

Do you know ballet?

It's always exciting to launch a new site, even if I didn't do much to contribute as of yet! I have been enrolled as a web analytics consultant to work on the National Ballet of Canada a few weeks ago.

Analysis process

The initial steps went something like this:

1. Listen to the business mission, goals and objectives (regardless of the web site)
2. Determine their web analytics maturity: the who's, what and how
3. Identified current pain points and challenges
   
4. Identify key goals on the web that can translate into KPIs
5. Established an historical baseline and at the same time, tried to identify quick wins. The tool used was VisiStat.
   
6. Provided instructions for Google Analytics tagging as well as CrazyEgg for the home page
   
7. Now into the post implementation quality assurance and will asses the current site vs. baseline in the upcoming days/weeks.
   

Site characteristics

Other than content about performances and performers, the site includes a B2C component for online ticket sales, so the conversion process is obviously a key element in measuring the success of the site. There are numerous events, dates, different seatings categories, various discounts etc. so the event and seating selection needs to be as intuitive as possible. There's even a "view from your seat" feature. The website also needs to reflect the unique mood and experience that is ballet. Unique site design, some high-res pictures and short movies bring a touch of class to the whole online experience.

Shame on me, I never went to a ballet event other than The Nutcracker... nevertheless I can appreciate the qualities of this form of Art and I will plan on attending one of the upcoming performance.

WASSUP? Four elements of a succesful web analytics program

When we ask "WASSUP?" we're seeking a fair and honest answer, but what we often get is the default "yeah, fine". When it comes to web analytics, we also ask how it's going, and we want the right answer!

Here's the presentation I gave yesterday morning during our WAA/eMetrics Breakfast.

If you have trouble viewing the presentation, use this link instead. I'm experimenting with VoiceThread, an amazing way to communicate and get involved into a conversation, so I'm looking for your feedback! Click on the "record" or "type" to add your comments directly into the presentation!

WASSUP?!

• Web
• Analytics
• Strategy: Trinity, the classic from Avinash Kaushik about Behavior, Outcomes and Experience
  
• Strategy: Multiplicity: also from Avinash; using multiple tools and sources of information to come up with better insights
  
• Ultimate Team: multidisciplinary and empowered teamwork from the Business/Marketing, IT and Analysis sides
  
• Process: Systematic, with SMART objectives, Continuous and considering an ever changing environment
• ! Obvious? Easy! We're just having trouble putting it into action.
  

Online business in Quebec: culturally and financially distinct

Despite growth in online ecommerce and retail activity, it is not clear that Quebec businesses are moving to take advantage of the trend. Quebec-based businesses risk losing their traditional customer relationships to foreign players in the online world.

With the percentage of Quebecers' online increasing (71.5 % in November 2006 and 72.3% in June 2007 CEFRIO) ecommerce activity is on the rise as well. In November, 16 % of adult Quebecers spent 401 millions on products and services bought on the Internet. And we know Canadians buying online increased 34 % to 45 % between 2004-06. However, since the CEFRIO "Indice du Commerce Electronique" doesn't state which percentage of this spending is done in Quebec vs the rest of Canada or abroad, it is not clear that this growth is translating into increased revenue for Quebec-based retailers.

The past year have highlighted the efforts of US Giants such as eBay and Amazon in the French language capabilities and optimization of their websites targeting the Quebec market. It is expected that the increased Quebec market traction resulting from these efforts will continue to grow in 2008.

The truth is that Quebec retailing has traditionally been able to rely on our unique culture and language as a barrier against foreign competitors who would have tried to enter the Quebec market. That advantage is not as important online and Quebec retailers need to start acting now to protect their status of preference among Quebecers or risk losing it to competitors from not only out of province, but out of country.

The web presents a double edged sword to retailers in any market as it both increases the potential business opportunity as well as the scope of competitors that business is likely to face. The need to strategically address a retail web presence with clear cut commitment to web analytics and the use of that data to adapt the online offering is a key principle for success online.

"There are proven methodologies and simple and straightforward steps that retailers can take to protect and grow their market share in both online and offline transactions," says Jim Sterne, founder of the Web Analytics Association and Chairman of the eMetrics Marketing Optimization Summit (coming up in Toronto, March 31st-April 2nd). "The growth of web commerce is clearly both a threat and an opportunity to Quebec retailers. Active use of strategic web analytics tools can help Quebec retailers to level the playing field and help them successfully navigate the changes that will be thrust upon them today and in the coming months."

To address the threats and opportunities facing Quebec Retailers, the Web Analytics Association will be holding a one day breakfast seminar at the Delta Montreal, on Wednesday, February 6, 2008, 7:30am - 10am. The forum will allow attendees to share and discuss the challenges at hand, and what steps can be taken to improve the state of their online presence.

(Note: content of this post re-edited from "Language and culture may not be enough to protect Quebec retails in the online world")

Free Landing Page Optimization webcast from VKI Studios

Last week I had a very interesting chat with John Hossak, VP of Business Development for the Vancouver-based VKI Studios.

John will be doing a free webcast on Google Website Optimizer on Feb. 5th. He will go through the step by step process of doing landing page optimization:

• Determining an appropriate test methodology (A/B, Multivariate, etc.)
• Deciding which page elements to test
• Developing test pages
• Implementing the tests
• Analyzing the results

Based on his valuable feedback, WASP will include information specifically geared toward Google Website Optimizer. It will tell you if the page you are viewing is an original page, a variation page or a conversion page used as part of a test.

Researcher or practitioner?

I was going through my backlog of emails and blogs and noticed a very interesting article from Human Factors International about the difference between the researcher approach and the practitioner approach to problem solving. Here's what I found most interesting and readily applicable to web analytics.

Seek for solutions

Clients are looking for solutions. Simple. While the details are important, and knowledge of (and benefits of) various analysis methods are core skills or a web analyst, you should convey the findings - better yet, the solution - rather than the analysis leading to it.

Key skills of a web analytics practitioner

HFI cites a study where key skills of a UX practitioners (designers) were identified. I took the liberty to slightly modify them so they can be applied to the field of web analytics:

• Effective practitioners need to be able to identify, diagnose and execute a correct solution quickly.
• Practitioners need to know which methods and tools are out there and when to apply which, and just do it.
• Practitioners should be able to articulate the tradeoffs for favoring a specific method based on the business perspective. This means balancing the quality of the data against constraints like:
  • How much will the various approaches cost to deploy?
  • Which is fastest given our current state?
  • Which is the most powerful but least expensive data we can get to make the case (e.g., quantitative, qualitative)
• Practitioners need to provide concrete recommendations, not problems and findings. And the earlier the recommendations come, the better. Here, we want to highlight the notion of "continuous improvement".
  
• Conversely, practitioners need to set up and provide metrics on how a process or interface fares over subsequent updates and releases. That is the core concept of testing.

Business acumen

The HFI article concludes with this:

Practitioners need to sell in business terms. They suggest that practitioners (and training programs) should hone negotiation and audience monitoring skills, and the ability to recognize when there is a gap between what is being presented and what the audience is seeking. After all, the business wants answers: What do we do? How to do it? What is the projected business impact (ROI) of making such changes? They aren't always as interested in the method and data that lead up to that solution.

The complete HFI newsletter is available online, along with lots and lots of other insightful material.

Practical analytics: incoming traffic quality

The initial inquiry was something like this: "In Site Catalyst how do i calculate bounce rate for referrers? I am trying to figure out the quality of referrers to the site using bounce rate by referrers".

Step 1: Define

The first analysis step is to define the improvement goal that is consistent with customer demand and business strategy.

We can rephrase that into something closer to a SMART objective: "During the month of October, which referrers brought people who engaged with our site beyond the initial page?". This might be good enough, but the initial statement was about "quality", which ultimately needs to relate back to the business primary objective (or one of the secondary ones). For an ecommmerce site, it could be: "During the month of October, which referrers brought customers?". We could even qualify it further by stating a minimal purchase amount, life time customer value, etc.

Step 2: Measure

Here we have a very good example of the difference between reporting and analysis. Reporting would limit itself to each individual metrics such as Referrers and Visits and leave it to the reader's imagination to find a meaning to those numbers. Analysis aims to look at the correlation between various metrics and build up a good story around them.

In the second step, we want to look at the process and collect relevant data for comparison.

In this case, we have a couple of important elements we can look at:

• Visit: "A visit is an interaction, by an individual, with a website consisting of one or more
  requests for an analyst-definable unit of content (i.e. “page view”)."
• Page Views: "The number of times a page (an analyst-definable unit of content) was viewed."
  
• Referrer: "The referrer is the page URL that originally generated the request for the current page view or object." In our case, we will want to look at the originating domain, not the whole URL. So we will talk about the "referring domain" for "external referrers".
  
• Conversion: "A visitor completing a target action." Typically, it's easier to look at the referring domain that brings us the most conversions.
• Entry Page: "The first page of a visit."
• Page Views per Visit: "The number of page views in a reporting period divided by number of visits in the same reporting period."
  

Step 3: Analyze

We want to analyze the relationship and causality of various factors.

In order to analyze what is happening, we need to define two new metrics derived from the basic metrics commonly available in any web analytics solution:

• Bounce Rate: The number of visits that resulted in a single page view and then left the site. It can be defined in Site Catalyst as a new calculated metric of Single Access/Entries.
  
• Weighted Bounce Rate: Same as above, but gives more weight to pages that are viewed more often, thus pushing problematic pages to the top of the list. It is calculated from this formula: (Single Access/Entries) * (Page Views/Total Page Views).

In Site Catalyst, the only way to find out Bounce Rate by Referrers is to go under Paths/Pages/Most Popular Pages and select our newly defined Bounce Rate and Weighted Bounce Rate.

Then we can simply click on the correlate icon and select Finding Methods/Referring Domains.

Step 4: Improve

Optimize based upon the analysis and various design experiments.

Now we have all information at hands to make hypothesis and validate them. One could think we don't have much control over who links to our site. In this case, we found out some referrers were simply linking to a "deep" page while they should link to a page that comes up sooner in the process. Since those referrers were mostly partner sites, we could simply ask them to fix the links. They are actually sending us qualified traffic, but at the wrong step of an important conversion process! End result: frustrated customers, loss of revenue.

Step 5: Control

The last step is to control that our changes results in positive outcomes from a user perspective, but also from a business point of view. We can simply run A/B reports showing the months of October and November side by side.

We want to ensure that any variances can be explained. Set up audit at specific intervals to asses conformity and institute control/correction mechanisms.

Six Sigma

Here it is! Without knowing it, we've just went through a very simplified Six Sigma approach that can easily be applied to web analytics. Using the Define-Measure-Analyze-Improve-Control process, it is easier to keep the business objectives in mind and stay focused on actionable data that brings tangible outcomes.

Webcom-Montreal: optimization through web analytics

Yesterday, I had the pleasure to be on a panel at Webcom-Montreal with my friends Jacques Warren and Simon Rivard to talk about using web analytics for site optimization. Here's some highlights from this presentation.

What is web analytics?

Jacques suggested a definition, here's mine, which is slightly modified from the one proposed by Avinash Kaushik:

"Web Analytics is the analysis of qualitative and quantitative data from your website, the competition and business systems to drive a continual improvement of the online experience that your customers, and potential customers have, which translates into desired business outcomes (online and offline)".

I'm adding "business sytems" as an important source of information used for analysis.

Why use web analytics?

We were unanimous on that one! Through the use of web analytics, we aim to answer three simple questions. You might change the way those questions are formulated, but it's essentially:

• Who is coming to the site?
• Why are they coming? What's their goal?
• Were they successful in accomplishing their goal?
  

What skills are required to do web analytics?

I proposed three key elements to be successful in web analytics:

1. Business acumen: knowledge of the specific business area, the global strategy, and how this strategy is facilitated by the Internet. Understand the goals and the processes related to ebusiness and "speak the language of business".
   
2. Technology savvy: general web and internet concepts and best practices (and sometimes deep understanding!). The architecture and how the site is constructed, structured and applications are conceived.
   
3. Analytical mind: ability to analyze data of various nature. Correlate it to business objectives and provide recommendations. Communicate efficiently to all stakeholders.

Don't worry, I don't know anyone who can claim to be a specialist in all those three areas (if you know someone who claims that, be very suspicious!). That's why a multidisciplinary team is the best, or hire a consultant for areas where you don't have the skills internally.

How can we improve a site?

Once we have a better insight on the three essential questions, we can seek to improve the site. Simon suggested the following areas of improvement:

• Spotting technical problems such as bugs, broken links or server errors
• Understanding the level of interest (engagement) of visitors (where they come from, what's their interest, what they are doing on the site, etc.)
• Better qualifying target audiences (especially in the context of marketing activities)
• Improving usability and communication efficiency

And another one of my own:

• Improving business processes
  

Simon noted that in his experience, any time there's a red error message on a page, you drop another 10% of your users.

What bounce rate or conversion ratios are ideals for...

We knew someone would ask! Simon joked: 0% bounce, 100% conversion!

• For eCommerce, some studies mention conversion from visit to purchase of 4% as being a good objective, 8% is exceptional.
• But each business strategy being different, and each expression of that strategy on the web being specific, bounce rates, ratios and conversion rates should never (well, maybe not never, but very rarely!) be compared from one company to the other.

It's really hard not to fall into anecdotal demonstration of such and such case where bounce rate was 50% or 2%, or conversion for a specific transaction was 98% (yes, I have seen that!) or checkout was improved by 400%. Again, each case is different!

Where can I find more info about web analytics?

You are here! It's already a good start :)

• Locally: "[WAM] Web Analytics Montreal" and "[WAQ] Web Analytics Quebec" are two monthly events happening in... Montreal and Quebec city were a bunch of people in the field get together for a drink, network, exchange and hear about the latest industry news. Look at blog.immeria.net/search/label/Event for the latest web analytics events (local and abroad) or subscribe to immeria's blog feed to stay aware of local events in Quebec. If you are on Facebook, you can join the group "Web Analytics Quebec".
  
• Association: The Web Analytics Association foster the interest of the web analytics industry and offers content, events, networking, webcasts, etc.
• Education: The University of British Columbia offers the Award of Achievement in Web Analytics. I will be tutoring the Introduction to Web Analytics course in January, and Web Analytics for Site Optimization later on.
• Conference: eMetrics Marketing Optimization Summit is THE conference on web analytics. The next one is coming in Toronto in early April. I'm on the advisory board for that conference.
• Blog: The blog of Avinash Kaushik is renowned to be the best one on the topic of web analytics.
• Discussion group: The Web Analytics Yahoo! Group is also a great place to learn. But please, before asking a question, look for older posts, there's a big chance someone already asked a similar question!
  

I hope that helps, please don't hesitate to send in your comments!

Great minds thinks alike: Bryan Eisenberg

Let me share a recent experience. A few days ago I had the chance to spend some time with Bryan Eisenberg. Who? If you don't know who he is, read on. Otherwise, skip the next paragraph!


Bryan and his brother Jeffrey are the co-founders of FutureNow Inc., he is the author of several best sellers in the field of online marketing and conversion, a renowned worldwide speaker, and the guru behind the concept of "persuasion architecture". He is, along with a few others like Avinash Kaushik, what we would call a Guru. Don't worry if they don't admit to be one: guru, like leadership, is something others see in you, not something you can claim yourself.


I spent about an hour with Bryan. We talked about my goals as a freelance, discussed persuasion architecture, and I had the honor to take a glimpse at the internals of their process and tools. Persuasion architecture is all about process, but it's also supported by a tool (or a set of tools) that could elevate any analyst job to levels of insights that were unimaginable without actually tying up Bryan to a chair beside you.


Market Motive, the new internet marketing think tank, can now count on Bryan as a leader of online conversion. Great minds thinks alike: Market Motive is bringing the cream of the crop that businesses need to understand in order to get a competitive advantage online.

Desjardins General Insurance site launch

Over the years, I have participated in many web projects, smaller campaigns up to full ecosystem revamp. Today is the official launch of my employer's revamped web ecosystem: Desjardins General Insurance.com (or Desjardins Assurances Générales in French).

On the surface, it's a new design, new content, renewed branding and a much more pleasing user experience.

Under the hood, it's Microsoft Sharepoint Server 2007 (MOSS) using Content Management System, custom authentication and optimized HTML /CSS rendering of pages... and lots of little tweaks and hacks. Being "early adopters" of a new technology or a new version of a solution is always risky, and of course, we ran into our share of problems. But at the same time, MOSS elevated us to a level where a huge amount of work was saved and the technological platform is in a very good shape to push the envelope. In this project like any other, there is rarely a "best" solution, there is always what appears to be the most promising scenario under the circumstances!

From the organizational perspective, it's a big project involving dozens of people over nearly a year. Financial services and insurance are usually quite conservative organizations, so this is a welcomed shift toward ebusiness and a much better leverage of the Web. And it's not over yet!

As a senior ebusiness adviser/architect, I was much more involved in the early stages of the project: addressing the business requirements and turning them into realistic and feasable solutions. While I used to be deeply involved in project development, this time I had to watch the project evolve from a distance and stay at a 40,000ft view. Still, there's always a warm feeling when seeing the baby's face for the first time! (ok, a bit stretched in my comparison, I have two kids and it's not quite the same... but you see what I mean!)

Conversion doesn't start where you think

We might have implicitly known about this fact, but this time it is quantified in a newly released study by ComScore/Forrester: 50% of those who start a transactional process actually have no intention of completing it.

No wonder we often see 2% or 4% conversion rates when using a magnifying glass to look at our checkout process or lead generation forms.

That's why Persuasion Architecture stresses that conversion actually begin much earlier, and also why it's important to measure more than the Conversion Funnel. Conversion Beacons, Points of Resolution, Waypoints and Driving Points needs to be understood and measured. This doesn't mean you need a dashboard will a gazillion of KPI's! At the executive level, you still want to watch Purchase and Buyer conversions (see E.Peteron's "Big Book of Key Performance Indicators"), but at the tactical level, you need to dig much more to find the root cause of abandonment.

That's where a systematic improvement process kicks in. I was first introduced to SixSigma when working for a manufacturing company where it was first applied for production of recreational vehicles, and than extended in all spheres of the business. SixSigma was the rallying cry for the whole company (maybe sometimes too much!). But I can appreciate and apply this methodology to web analytics. In SixSigma terminology, I'm some kind of a Green Belt :)

When faced with a financial transaction, the 50% or so who didn't expect to complete the process gave the following reasons:

• 23% wanted more product information
• 19% where not ready to apply
• 14% wanted to see if they qualified for the product

Even when they had the intention of actually completing the transaction online:

• 12% changed their mind (what's missing is "why?")
• 11% had privacy/security concerns
• 11% wanted to speak to a human regarding the product

To sum it up, "As more consumers research and purchase financial products on the Web, the importance of understanding application abandonment will increase", said Forrester. You bet!

Thinking out of the box

One of the common skill mentioned in web analytics job postings is worded as "strong problem solving". Problem solving is a skill anyone can learn, but there are some ingredients that are essential: an innovative mindset is one of them.

One of the latest ChangeThis manifesto highlights the seven pitfalls to avoid in order to break the mold of traditional thinking:

1. Shortcutting: Avoid leaping to solutions; a "blink" solution often hide deeper causes. Although most problems don't require us to analyze them very deeply, when faced with more complex challenges we have a tendency to jump to conclusions.
2. Blindspots: Experience often leads the way and can make us blind to what we are less comfortable with. We have to force ourselves to break the mold and think differently.
3. Not invented here: Sometimes it's worth to consider and trust other's solutions.
4. Satisficing: When we satisfy from a solution that will suffice... we satisfice. We compromise, instead of going for the best solution, we go for the less negative one.
5. Downgrading: Lower the bar and it's a lot easier to achieve success! Avoid overselling the upsides and shuttering the downsides...
6. Complicating: Seeing things more complex than they really are is an easy justification to add cost and time and reassure us in our lack of comfort when confronted with the unknown.
7. Stifling: Dismissing altogether, or second guessing one's idea if favor of our own.

I've worked in environments where over half of those pitfalls where commonly seen without even realizing it was happening. Younger and less formal organizations often fall for the not invented here, shortcuting while older organization are more prone to satisfice and stifle. It's insidious, and it's often part of the corporate culture.

The web analyst is often an agent of change and an evangelist within the organization. Developing an innovative mindset and problem solving skills is rarely presented in formal training, but just being aware of the pitfalls is already a good start.

Practical analytics: persuasion scenarios

A typical site is composed of several conversion objectives: registering to a newsletter, purchasing something, finding the nearest store, or any activity that represent a positive outcome for the user and the site owner. As Bryan Eisenberg puts it, those represents the "conversion points" of a "persuasion scenario":

Schema from Brian Eisenberg

Conversion Funnel

Conversion beacons represent the first and each of the steps leading to a conversion point. They are typically represented by a conversion funnel. On a typical ecommerce web site, the conversion beacons are the steps of the checkout process (a linear funnel) leading to a purchase (the conversion point).

What you should look for in the funnel:

• Distraction: how many of your visitors engage in the conversion process but leave it to wander in another area of the site? Look for next-page flow that leads to other pages of your site.
  
• Abandonment: how many of your engaged visitors are simply exiting the site? Look for the Exit Page metric for each of the steps of the funnel.
  
• Errors: anytime an error condition is presented to the user, the likelihood he/she will exit the process becomes more significant. Look for page reloads ratio and next-page flow that leads to error pages.
  

Conversion beacons

A conversion beacon is a step within the conversion process. This step is more easily analyzed with a Form Overlay tool, where metrics are shown "on top" of the form. Each conversion beacon should be analyzed for:

• User errors such as skipping a required field or using the wrong input format: eliminating errors as close to the user as possible is usually the best approach. Avoid round trips to the server for simple validations, implement AJAX calls to ease the process flow.
  
• Abandonment at the beginning of the form: a typical indication of users being overwhelmed by the complexity or the length of the task at hand. Break complex tasks in manageable chunks and clearly tell the user about where they stand in the process.
  
• Abandonment at the end of the form: a typical indication of fear of the unknown. Clearly label action buttons ("continue" or "place your order" or "confirm"?) and again, indicate where you stand in the process.
  
• Clustered abandonment: high abandonment on a specific field (or group of fields) is a clear indication of misunderstanding (what do they mean by "Enter address category"?) or mistrust (why do they ask for my birth date to purchase a book?).
• Abandonment anywhere in the form: once you've nailed down all the other aspects, there will be a number of inexplicable abandonment and there might be no rational explication behind those.
  

Points of resolution

Points of resolution are those pages that triggered the user to take action. Next in the analysis process is to find out which pages contributed the most to your conversion objective.

• Previous-page flow: when looking at the first conversion beacon of your funnel, look at the previous page flow. This will indicate where people stood before entering the process.

Waypoints

Moving up another step, we want to see how each page contributes to the conversion. What used to be path analysis as become irrelevant in most cases because the number of combinations are virtually infinite and are not significant. So we rely on the Page Value instead:

• Page Value (called "Page Conversion" in HBX or "$ Index" in Google Analytics): this is typically calculated by distributing the value of a conversion to all the pages that where viewed before the conversion. Even in a non-ecommerce site, specifying a conversion value (say, 1$ for every subscription) will be particularly useful to calculate this metric.

Funnel Points

Those are the pages that puts you in a position to influence the user toward the objective. They can be entry pages or specific pages that are specifically geared toward the persuasion process.

• Entry Pages: Look for the entry points and how they contribute to your conversion.
• Page Value: Again, the Page Value will help us identify our winning pages, but the next one might be even more interesting:
• Pathfinder: Omniture SiteCatalyst offers "Pathfinder", this will basically identify a significant subset of the click path that contains specific pages (such as "what happened between my entry page and my first conversion beacon?").
  

Driving Point

Here, the goal is to find out which "source" brought us the most qualified traffic. This topic in itself represent one of the major web analyst tasks and typically consist of analyzing the following metrics as they relate to conversions (those who engage but do not convert) and bounce rates (those who stumbled on the site and did nothing else but leave):

• Referrers: which web sites brings us the most qualified traffic?
  
• Search Engines: where do people go to find about us?
• Keywords and phrases: regardless of the search engine, what was in the mind of people that came to our site?
• Campaigns: which of our marketing activities are the most successful. Be it banners, PPC or even offline activities.

Key Performance Indicators

A very common question is "Which KPI should I track?". Here we're talking about indicators that will instantly raise a red flag (or bright green light) if they change:

• Conversion Ratio: Each of your persuasion scenario, be it a purchase, a subscription or any other valuable business activity, should be measured as a Ratio of First Conversion Beacon to Conversion Point. Have such ratios for each individual scenario. Example: Shopping Cart/Purchase, or Subscribe/Thank You.
  
• Global Conversion Ratio: Regardless of the persuasion scenario (which transaction), track how successful your visitors are at accomplishing their goals (and yours!) by dividing Any Conversion Point/Visits.

Web Analytics implementation Quality Assurance

Here's a couple of references that should help improve the quality of your web analytics implementation:

• Web Analytics Implementation: Items Overlooked by Manoj Jasra
• Part 1: Your Web Analytics Data Quality May Stink and Here’s Why! by Judah Phillips
• Part 2: Your Web Analytics Data Quality May Stink and Here’s Why!
• Web Analytics and Data Collection: The Page Tag by Judah Phillips
• Web Analytics Technical Implementation Best Practices. (JavaScript Tags) by Avinash Kaushik

Which brings me to the next question: what would you like to see in WASP?

If you don't see the poll, please visit my blog or answer directly.

• More web analytics solutions. The current version of WASP detects 45 of them, the next one already has 50, and there's nearly 200 web analytics solutions on the market, some of them phasing out, new one popping all the time.
• Site crawl and implementation diagnostic. Missing and duplicate tags, JavaScript errors, duplicate page names, wrong location of code, etc.
• Market stats about web analytics solutions. Which solutions are leading the pack? Who's using what?
• Detailed implementation QA for a specific solution, which one? Each web analytics solution has its little tweaks and tricks.
• Others? What would you like to see?

Please fill-out the poll, send me an email or simply post a comment!

Some fun with a site design

Since I met Mr.Carrabis at eMetrics and attended his fabulous presentation about the human side of online marketing, I'm much more aware of how subtle design elements might affect the user behavior. So let's have some fun with a site my wife stumbled on.

What's wrong with this site?

Take a look at the picture on the right (click to see a larger snapshot). The site is actually in French, so don't worry about the wording: just focus on the design. What's your first impression? What's wrong?

Asymmetry

If your first impression was one of confusion (or maybe even disgust), you've guessed right. My 15 year old daughter wants to study in an art-related field (see one of her drawing below) and she learned that one of the most difficult thing to draw are portraits. Why? Because our brain is so good at interpreting human faces that anything that appears to be an anomaly will literally jump at our face (sic!). Humans have been doing it for millions of years to know if the person in front of us was hostile or friendly. You've probably heard about the little experiment where you take a person's picture and use Photoshop to create a half mirror representation of the face. We instantly detects there's something wrong.

Why do this? What could be done instead?

This company is about human resources. I don't know anyone there and don't have a clue about who designed it. My only guesses for such a choice are: a) represent the two sides of the job hunting process: employer/employee. b) be politically correct by showing mens and womens "at the same level". Furthermore, the animation makes us focus too much on the pictures instead of the left and right-side navigation.

There's a nice section where employers and employees gives short testimonials. Wouldn't it be great if their picture was taken and the blurb be put right beside it on the home page? A single display (maybe chosen at random) and previous/back buttons would be less intrusive than an automated animation. I would love to do A/B testing on that and see the results!