Wednesday, January 30, 2013

Web Analytics Framework

For those venturing into the world of web analytics for the first time, it is useful to have a framework to understand the process of transforming data into action.  At its most basic level this framework has five key components that include identifying business objectives, defining goals, associating KPIs with those goals, setting a KPI target, and then creating segments to drill down into various visitor types.
  

Following the framework below provides structure to the process of developing a strategy from web analytics.[1]

1.  Business Objectives: the foundation of the process is to identify the overarching purpose of the business.
2.  Goals: what are the goals related to these objectives?  This helps teams focus on what is important, so they can best utilize their time.
3.  KPIs: these are the metrics that are attached to our goals.  Some common KPIs are conversion rate, average order value, days and visits to purchase, visitor loyalty and visitor recency, task completion rate, share of search. 
4.  KPI Target: setting this target enables you to evaluate whether or not the target was missed or met.
5.  Segments: drilling down into the characteristics of various visitor type enables you to understand what various groups of visitors are responsible for helping to meet or not meet targets. 

Once the above framework has been developed, an important next step is to develop a dashboard to communicate this information to decision makers.  This is where value is created, as information is not valuable if actionable insights are not being developed.  Software exists that facilitates the development of dashboard.  One free solution is Straigent, which comes on the CD bundled with Avinash's Web Analytics 2.0 book. 


Following best practices will enable new analytics professionals to get up to speed quickly, and leverage the experience of others.  Starting with these principles in mind will enable your team to increase your efficiency, and focus on what really matters.  Below are 8 R rules to guide your strategic effort:[2]

1.   Start simple, and start with outcomes.
2.   Leverage metrics that identify success for areas where you spend the largest efforts
3.   If you are unable to segment a KPI, choose a different one
4.   "Brand" and "site usage" can be measured, loyalty is very important
5.   Make sure to include a couple of metrics that report the voice of the customer
6.   Tie your web metrics back to traditional metrics
7.   Always include KPI's that give you a window into competitive intelligence

Effective web analytics requires a process of continually developing new hypothesis, and challenging existing assumptions.  Because of this it is useful to go through this process regularly with a fresh mind once an initial program has been established.  As your business adapts to the changing world, your metrics and segmentation will likely change as well.  Best of luck with the process!

[1] http://matt-smedley.com/web-analytics-framework
[2] http://www.kaushik.net/avinash/rules-choosing-web-analytics-key-performance-indicators/



A Reasonable Start to an Analytics Glossary

People have a propensity to want to sound smart. Any time we can pick up some new jargon or buzz words we’re inclined to just toss them straight into our vocabulary and flaunt them like we’ve used it every day. After all, who doesn’t want to seem like they’re part of the ‘in crowd’ by using a term that nobody else had ever heard of?



Hajura, Will Robinson! Hajura!


Do all those words actually mean something? Absolutely! And not in the usual business school sense that makes everyone hate MBA prospects for thinking we sound like the smartest people in the room.

Nope, we actually need to understand what everything on the back end of analytics mean so we can talk with Developers. And we all know how hard it can be to relate to the IT department and not feel like a moron....



With no further ado, some actual helpful terms and conditions that go into analytics

Uniques:
These are individual people that come visit your site. It’s actually a measure of unique browser IDs that visit you, so one person could count as four unique visitors if they use FireFox, Internet Explorer, Chrome, and Safari from the exact same computer.

Returning Visitor:
It’s a stalker. Well, not really. It’s someone who came back to see your site again because you’re popular.
This is the reason why Visits and Unique Visitors are not the same thing. Cookies are a great way to track this.

Cookies:


 


Seriously, I have a video for everything.
When you look at a website, odds are your web browser stored something on your computer so they can watch what you’re doing. Ostensibly, this should be to serve you better but that’s probably not the case. It’s likely that they just want to know what you’re doing so they can make you buy more stuff. Ha. Seriously, there are a few different kinds so here are a few of them:
Session cookies should only last for this one time you’re on the site
Persistent cookies are how Amazon remembers that you really like clarified butter so they should recommend a bunch of different types to you next time you’re on the site. They’re also how VW knows to show me ads with Jetta TDIs in them because I stare at them ALL THE TIME.
Third-party cookies are probably the sketchiest one you’ll hear about. You can basically buy ad space on someone else’s site (maybe our blog?) and as the advertiser, drop a cookie for your own website on a user’s machine because they viewed your ad. Yup, it’s as sketchy as it sounds, and it’s probably why your computer keeps showing you ads for finallyfast.com

Session ID:
Remember that session cookie we looked at earlier? This is the ‘tag’ in that cookie that lets the website know that you’re you, no matter how far you go on the site. Yeah, that stalker video seems pretty appropriate now, doesn’t it?

Browser:
Why the hell does this even matter anyway? Who cares what web browser someone’s using to view your site? They all access the internet!
Actually, it’s a pretty important statistic. There are three primary ‘rendering engines’ that power web browsers:

But what does that even mean? Gecko is the stupid mascot from that insurance company, isn’t it? Don't worry, these actually correspond to something:
  • Internet Explorer = Trident
  • Safari, Chrome, Android’s Browser = WebKit
  • FireFox = Gecko
Soooooooo.... when’s it going to mean something? Believe it or not, your web developers need this information and it's going to impact your web budget. Big time. If they know what browsers your audience is using, developers can test to make sure it works. Websites don’t ‘just work’ everywhere, each one of these engines draws pages very differently and web developers usually have very strong feelings about them. Especially about old versions of Trident (namely IE 6).


Bounce Rate:
These are people who leave your site after only visiting one page. This could actually be a huge indicator of problems and you should pay careful attention to it. Why would someone leave a page after only seeing it once? Was your layout too confusing so people couldn’t find what they were looking for? Rather than groupthinking this one to death it could actually be worth doing some real live user studies to monitor what people are doing on this page that causes them to leave. A/B testing (ugh, a buzz word that hasn’t been explained!) could help resolve this.

Exit Rate:
Wait, didn’t we just say the bounce rate is from someone leaving? It is, but the difference here is that someone has been browsing through your website and happens to leave after this page. It can actually indicate success!
Let’s say someone looks at four pages. The fourth page is the ‘exit’ page, so you can check high exit rates to see if people are leaving on, say, a product information page because they’ve found what they’re looking for. This is much less alarming than someone leaving from the product selector page because they didn’t see anything they found enticing.

Campaign:
Don’t give me that blank stare. It’s an advertising campaign, only way more sophisticated. There are a lot of different options to run your ads online but the important thing here is to manage them. You can have tons of different things in a campaign, some of them are listed below.

Banner ads:
The big picture ads nobody likes clicking on. Want to know a trick the car insurance companies won’t tell you? Those banners sure hope you do.

Text ads
Google made ‘em famous, just look at the right side of your search window and they’re tailored to what you just searched for. And someone paid for it. Try searching for DUI lawyer, those tend to go for big bucks..

Keywords:
Speaking of DUI lawyer... this is when you actually buy ‘words’ or ‘terms’ from Google, Yahoo, or any other search engine. You can actually bid out for ad placement depending on what the search term is. Things that are potentially worth a lot of money, like someone searching DUI lawyer on a mobile phone, could sell for a hundred bucks a click (no joke), while a term like ‘best garbage bag’ could sell for 5 cents.

Conversions:
Oh boy, now we’re in analytics land. A conversion is when someone actually does what you want on your site. You can set it up when you build your campaign so that...

  1. A user clicks on a web banner
  2. A cookie (yeah! We used a term!) is placed on their machine
  3. You follow their session ID (oh man, another one?) across pages to get to the shopping cart
  4. They actually buy what they put in the cart!
  5. Profit! We’re rich! Richer than astronauts!
See, there should be something besides ??? before profit. That’s why conversion rate is so important, so you can see how effective a campaign is.

SEO:
Search engine optimization. Oh man, that sounds about as rough as it is. You can actually build your website so that search engines have a way easier time indexing them. Remember all that nonsense about Flash when the iPhone couldn’t support it? That was actually a big deal because search engines couldn’t index flash. Wait, what’s indexing?

Indexing:
Google basically makes up an absolutely, unbelievably huge database/library of everything on the internet. They have a web crawler that does it for them.


Wait, what?

Web Crawler or Spider:
It’s a piece of software that is designed to look at every single thing on the web. If there’s a link on your website, it’s like an invisible person who clicks on it and writes down where it came from and where it goes to.

Back to SEO:
So sites like Google and Bing (ok, Bing steals from Google because they’re lazy. See last week’s post for that kick in the jaw because I’m not linking back) actually use Web Crawlers to Index your website. They see what words are close to links or widely used inside your content, then they actually record how popular your site is. It all balances out and combines with things like your Paid Search Keywords (saw that up higher) to dictate how high your page ranking is in a search engine, or how much you have to pay for that keyword.

Page rank:
It’s actually where you appear in Google or other search engines. You used to be able to screw with the system by linking to something a bunch of times, which is why Google Bombing became an expression. Where are those WMD’s again? And what’s a Miserable Failure? If only I knew of some Great French Military Victories...

Meta Data:
That’s so meta. Actually, it’s hidden data on the back end of a web page that describes what’s on the front end of a web page. People used to ‘cheat’ Google’s page rank by listing a word (like ‘awesome’) on the back side about 500 times so, which could help create Google bombs. Don’t worry, Google’s smarter than that now, but it’s still worth defining these in your commerce products to help broaden their search appeal.


A/B Test:
An A/B test is actually facilitated by those cookies we talked about earlier. We discussed it in class so it's not exciting to go into detail, but it basically means you can show some users one page, and other users a different version of the same page, and then you can look at the data to see what version is more successful (based on what you wanted people to be doing, of course).

 
Some Commerce buzz:

Abandonment Rate:
Someone didn’t want to buy your stuff; the user added an item to his or her shopping cart but didn’t finish checking out.
This can actually happen anywhere in the checkout process, you should probably be aware of what pages in your checkout are showing dramatic increases in abandonment or dropout, that way you have something to fix and it makes you look smart!

Abandoned carts:
Yeah, it's the same thing as above.


ASC:
Average Shopping Cart. It’s an average of how much a shopping cart is worth in your store.

Alright, seriously, you made it that far? I should give you a sweet reward for diligence, but I really don’t have much in the way of prizes to give you. Here’s someone being excited to help bring you back to reality:




Data Visualization and Web Analytics, a Comparison


What is data visualization and what does it have to do with web analytics? According to Matthew Ward of the Worcester Polytechnic Institute, data visualization is “the graphical presentation of information, with the goal of providing the viewer with a qualitative understanding of the information contents.”[1] In other words, data visualization is infographic design. [2] For those of you who don’t know what that is, it is a special form of graphic design that is focused on the visual representation of information. I bring this up, because one of the most important things about digital analytics is understanding, or analyzing, the data. For that reason, these two topics have a lot in common.

Good infographics depict information in a quick and organized way. A lot of the time, a designer is dealing with a lot of information and is faced with the dilemma of how to show that data in a single page or in a single design. The infographic poster to the left depicts the average data consumption in the U.S. in one day. [3] That is a topic that definitely has a large amount of data connected to it. Somehow, the designer sifted through it, organized it and came up with the poster you see now.
Web analytics is similar. Web analytics is defined by Evan LaPointe, author of the blog Atlanta Analytics, as “the measurement, collection, analysis and reporting of internet data for purposes of understanding and optimizing web usage.” [4] The key words here are “understanding” and “optimizing.” If all you have is data, it isn’t going to be much use to your business. It is the understanding and optimization of that data that companies look for. It is how that data can be used rather than just ­collected. You and your client must be able to see the data. In a way, it is much like seeing the data in the data usage poster as discussed above.

            The objectives of data visualization and web analytics differ. In the first, the idea is to help others visualize data in such a way as to inform the viewer. In the latter, the goal is to take a large amount of data and understand it in order to make strategic decisions for a company. Regardless, they are both ways of informing others. Thomson Dawson of Pull Brand Innovation states that, “Marketers (and those who cling to big data) are prone to view brand design more narrowly than they should.”[5] The point Dawson makes in his article is that those who are in charge of business strategies often get stuck on a piece of data or an idea, and sometimes it takes someone from the outside with a new and broader view on the subject to make the changes needed to thrive. There will come a time when one of your clients, or the business you work for, will put you in a similar situation. When that happens, just remember, there is always a way to show that person or persons why the data lead you to that decision. It is just a matter of visualizing the data and presenting it in such a way as to persuade.


           




Footnotes

[1] Matthew Ward, “Overview of Data Visualization.” URL: http://web.cs.wpi.edu/~matt/courses/cs563/talks/datavis.html accessed: Jan. 30, 2013.

[2] Amy Balliett, “The Do’s And Don’s Of Infographic Design.” Last modified: Oct. 14, 2011. URL: http://www.smashingmagazine.com/2011/10/14/the-dos-and-donts-of-infographic-design/ Accessed: Jan. 30,2013.

[3] Bima Arafah, “Huge Infographics Design Resources: Overview, Principles, Tips and Examples.” Last modified: May 12, 2010. URL: http://www.onextrapixel.com/2010/05/21/huge-infographics-design-resources-overview-principles-tips-and-examples/. Accessed: Jan. 30, 2013.

[4] Evan LaPointe, “A Better Definition of Web Analytics.” Last modified: April 22, 2010. URL: http://www.atlantaanalytics.com/practicing-web-analytics/a-better-definition-of-web-analytics/ Accessed: Jan 30, 2013.

[5] Thomson Dawson, “Why Designers Make the Best Brand Strategies.” URL: http://www.pullinc.com/why-designers-make-the-best-brand-strategists/ Accessed: Jan. 30, 2013.­­­­



           



Tuesday, January 29, 2013

What is the Value of Facebook likes And Twitter Followers

Web 2.0 is all about user generated content (UGC). In web 2.0 the demarcation between the content provider and the content user is close to non-existent. The revolution created by blogs and social media like Facebook and Twitter has changed the way the information is shared between the content provider and the user. Until the late nineties in Web 1.0, the dialog between the content creator and the user did not exist. The author published through internet which was assimilated by the user, however the author was unaware of what the user did with the information.



                        (video ref: http://www.youtube.com/watch?v=6gmP4nk0EOE)

With the advent of social media, the author now has the opportunity to receive feedback in the form of comments, likes, tweets and followers. In a business perspective the response of the user (read customer) to a post by the author (read business) can be invaluable, provided the reception is tracked.



Avinash Kaushik in Web Analytics 2.0 talks about metrics that can be used effectively to track the customer response for a post either in a blog or a micro blog like Twitter. Out of which likes and retweets can be most prominent and easily tracked. However, the real value to a business or an individual can be realized when the metrics such as Facebook likes and Twitter followers be quantified into dollar amount. Dan Zarella in his HBR blog How to Calculate the Value of Like prescribes a mathematical formula to convert likes/followers into dollar amount. The formula is given below,

Value of a Like = (Total likes)/(Links per Day*30*(Average number of clicks/Total Likes)*Conversion rate*Average Conversion Rate

An automatic calculator at Valueofalike.com provides the dollar amount by entering these parameters. Just for fun I used the data from the first blog I posted using likes and number of visitors. I set everything else to the lowest score. The value for each like turned out to be $0.27.

The formula appears to be empirical, however the estimated monetary value of the like/followers metric provides great insights to the author or a business tracking social media metrics.

References:
  1. http://www.youtube.com/watch?v=6gmP4nk0EOE
  2.  Avinash Kaushik, Web Analytics 2.0: The Art of Online Accountability and Science of Customer Centricity
  3. http://blogs.hbr.org/cs/2012/11/how_to_calculate_the_value_of.html
  4. http://valueofalike.com/
  5. http://dauofu.blogspot.com/2013/01/social-analytics.html










Google Event Tracking and Bounce Rate


Overview
After last week’s post on the differences and similarities between Google Analytics event tracking and ­custom variables I updated my website to include various types of custom variables and event tracking. Everything looked awesome in analytics and was able to better understand my customer segments and behavior. However, after a few days I started to notice that our site bounce rate drastically reduced and was having a new record low. I knew I screwed something up.

The Problem
After some digging around and researching Google’s event tracking code online I finally figured out what was going on. In general, we consider a “bounce” as a single-page visit to your site. So when a visitor lands on your product page and reads about the product then decides to exit the website we count that as a bounce. However, let’s say you setup event tracking on the video clip in the product description to know how many people are viewing the video. A visitor again lands on the product page, watches the video, and then exits the website without visiting any additional pages. If we go by our general description of a “bounce” then we would expect this to be recorded as a bounce since the visitor did not visit any additional pages before leaving the site. Google event tracking takes a different approach that turns out to be very helpful. Using the default code for event tracking in the video example above Google Analytics would view this as the visitor interacting with your website and therefore would not count their visit as a “bounce”. On my website I had event tracking automatically executing on page load so I was getting a zero bounce rate for all those pages.

Is it a Problem?
First we have to understand that everyone’s site is different. For example if my site was a blog and I enable event tracking on some of the videos or other areas where visitors could interact then I probably wouldn't want to record visitors that interact with my blog as a bounce if an event was triggered manually. This is because it’s common for people to only view one page of a blog. You could in this example add an event if the visitor scrolls down to read more of your blog. That would be helpful to know and you probably wouldn't want to count it as a bounce if they then exited your website.

On the other side we have websites with pages that are special offers or an eCommerce site with the whole purpose of having visitors move to other pages on the website. In this case you would probably want to record any single-page visit as a bounce. So you might be asking yourself how can you track events and still report the visit as a bounce if it’s a single-page visit? Simple!

The Solution
Google updated its event tracking script to include an optional parameter called Non-Interaction Events. You can simply include a “true” to tell Google it is not an interaction hit and should not be used in the bounce rate calculations. Conversely, by not including this parameter or including the parameter with a “false” will result in the event being calculated in the site bounce ratio. Regardless of the approach you take this can better help you understand the benefits of using event tracking and have more control over your sites bounce rate calculation.

Reference(s):
https://developers.google.com/analytics/devguides/collection/gajs/eventTrackerGuide#non-interaction

Increasing Effectiveness of Web Analytics Reports



Increasing Effectiveness of Web Analytics Reports
                Our Optimizing Online Business group met yesterday to outline our final project and make assignments to complete the final report.  One of the most interesting issues that arose as a result of our discussion was not in the thesis of the project, or the specific tasks, but in the level of detail, and writing style associated with the work.  We believed that our report should be written as if it were going to be delivered to executives of the company whose website we are analyzing.  This posed additional questions: What level of understanding do we assume?  What level of detail is necessary for comprehension?  How do we emphasize the key points clearly without adding in unnecessary confusing technical information?  We want to ensure that our project outlines meaningful results for the company whose website we are researching, in a clear  and impactful way.  The following are some guidelines on writing Web Analytics reports.
                One of the most simple ways to improve a technical, or analytical report is to make sure it has a clear, sound structure[1].  If the reader is a technical novice trying to understand some dense material and he or she also has to try and decipher a complicated or muddled report layout the chances of comprehension decrease significantly.  The author can ensure good structure in their written work by outlining the report before writing commences.  Especially in group projects, this step will ensure that all members are clear as to their role in the writing.  This step is particularly essential in writing technical reports because you will likely have to build the foundational knowledge of a particular subject (such as web analytics,) in the beginning of the report so that understanding of the entire document is ensured. 
                Another important element to consider in technical report writing is to simplify your sentences. According to Allanalytics.com, "When you've been pouring your life into a project, you can become so familiar with it that it's easy to forget how much others don't know."  Writing complex, explanatory sentences in technical jargon is easy when you are knee-deep in technical research, however, you will likely lose your audience.   Keeping your sentences simple, and only introducing one new idea in each sentence will ensure your audience understands your material. 
                One last important aspect to include with a technical, web analytics report to ensure understanding is multimedia and visual elements[2].   It is easy to compile meaningful and impactful visuals using Google Analytics and Site Catalyst.  These graphics will help demonstrate your main points, and in some cases may be the only piece of your report that an executive looks at.  Most web analytics reports will be rooted in Google Analytics or Site Catalyst data, and both programs offer stunning visual representations of their reports, it is a simple and effective way to convey data in your report to include their visuals. 
                Taking these steps will help ensure your report is read, understood and put to use.  Our group plans to use these steps in increasing the efficacy of our final report for Optimizing Online Business.
 -Marian Bonar


[1] http://www.allanalytics.com/author.asp?section_id=2587&doc_id=257849&
[2] http://odtl.dcu.ie/wp/1999/odtl-1999-03.html