Occam’s Razor by Avinash Kaushik

[There is a updated version of this post: Top Ranked Blogs: April 2007.]

Welcome to the third edition of the Top Ranked Web Analytics blogs.

The evolution of the ranking system continues with a couple of tweaks to the ranking this time around. The primary determinant of the rank in the list below is still Technorati (click here).

To be considered the blog should primarily be on the topic of Web Analytics (50% or greater number of posts), as judged by a qualitative review (by me) of the last two months of posts.

Important: If you feel your blog has a higher Technorati ranking than 46k and it is not here please email me (blog at kaushik dot net). Especially if you are a non-English blog. Please contact me.

I was amazed at the change in the rankings each month (from June to July to this one). I did not expect so much flux in the rankings. Overall it is so fantastic that the number of blogs on the topic of web analytics continues to increase, as do the rankings of each blog. Viva la web analytics!!! 

Some personal observations:

• Any blog you see with a N/A in the July column was not on the list in July (and now they have the top three spots !).
• Hitwise inclusion was a tough one because they are not “typical web analytics”, atleast most people don’t think so. In my mind Competitive Analysis is very much a cornerstone of any solid web analytics programs (Business case: click here & Usage: click here).
• I am glad to see international representation continues with the E-consultancy blog.
• Standards work. In the first list there was correct critique that not everyone will use, or have claimed a blog on, Technorati. This list proves that if you go out and suggest a standard people will figure out how to get on the standard. (And I thank you all for that.)
• Competition is really tough in the blogosphere (hence gains of web analytics blogs are even more impressive). In mid-July it would have taken a blogger roughly 190 unique blog links to get a sub-10,000 technorati ranking. As of today a blogger would need approximately 230 unique blog links. Amazing.
• You would have needed 58 unique blog links to make it to the list above today.

A Personal Best Blogs Ranking:

Here is my personal ranking of the “best” web analytics blogs (using the criteria that they “Eat like a bird, and poop like an elephant”). They get the ORbAK Stamp of Thanks! (and maybe a golden star to put on their blog : )).

# 1: This Just In – Justin Cutroni
(I am amazed at how much Justin contributes to the community, soon he will work himself out of a job because rather than hiring him I’ll just use his blog to fix every problem I have!! : )

# 2: Increasing your website’s conversion rate – Robbin Steif
(The irrepressible Ms. Steif’s blog has a wealth of practical knowledge for any web analytics practitioner. Her honesty is to die for. : ))

# 3: Commerce 360 Blog – Craig Danuloff
(Craig shares a very unique real world perspective that I find very refreshing. He is frank, he adds a lot of value and his personality shines through clearly.)  

If you don’t have them on your RSS reader then I highly recommend them.

I’ll repeat the six “reporting tips” that I have collated in the process of doing these rankings: 

1. Global standards and benchmarks are great because people buy into them more easily
2. Simplicity always wins over complexity, because what people understand better they are more likely to action
3. Judgment should be applied with a lot of careful consideration because reasonable people might disagree with someone they don’t know
4. In any report context is king, provide the right context
5. Be aware of hidden agendas, your’s and those of others
6. Be open and up front with your assumptions
7. Solicit feedback from your report consumers and incorporate relevant feedback in future iterations

So what do you think? How else can the ranking or the criteria be improved? Please share your feedback via comments.

[Like this post? For more posts like this please click here.]

As you can imagine choosing a web analytics tool is a critical process because whether you choose right or wrong you will be usually stuck with it for a while. Also because we tend to overstate the importance of historical web data (a post coming on this one of these days) it is quite likely that a quickie divorce will not be in the offing.

So in a world where choosing a web analytics tools seems akin to choosing a wife / husband / significant other, this blog post offers a radically different recommendation on how to choose your mate, sorry web analytics tool.

Summary: The current process of choosing a web analytics tool is fundamentally flawed. Why not change the process to: Implement Free Tool -> Learn -> Upgrade Analytics Skills -> Fix Data Capture -> Evaluate -> Now Choose?

The way tools are chosen at the moment is through a extensive process that usually looks like this:

1. Collect all the business requirements (goals, strategy, KPI’s, reports, reporting schedule and so on)
2. Collect all the technical requirements (site architecture, servers, scripts, pages, IT needs and so on)
3. Ensure that anyone who could ever need any kind of access to any kind of web data is contacted (inside and outside the company) and their needs documented
4. Put all of above into a RFP (Request for Proposal), and add vendor financial stability, references, etc into the requirements
5. Send to vendors, set a very aggressive reply schedule
6. Receive RFP’s
7. From those weed out the “insignificants”
8. Selection of one vendor that meets the requirements by a esteemed committee
9. Implement (champagne celebration included)

The search process takes two to four months, implementation one to two, and there is one guaranteed result of this process eight times out of ten: You will pick the most expansive, and usually one of three expensive, vendor.

Roughly the same number of times you would have made a sub-optimal choice and you are looking into three, six, twelve months of stress and having to deal with management questions that sound like this: “how come you are using a quarter million dollars a year web analytics tool and you are not recommending actions”.

The achilles heel of the above process is that the process involved people who asked for the earth and moon in the requirements (most of whom will never even log into the tool) and it was very divorced from the rough and tumble real world of the web, website and web analytics.

The process is too long, too time consuming, the process itself is expensive (just count the people, process, and time commitments from your company) and you’ll always pick the most expensive and expansive tool.

As a antidote to this sub optimal outcome here is a recommendation: 1) Ignore the nine step process above and 2) Don’t do a RFP.

Here is a radical alternative that will set you on a path to finding your right soul-mate (I mean web analytics tool), and do so faster and cheaper:

Step 0: Assign optimal ownership. (Day One)

• Project lead should be the person whose neck will be on the line to deliver web insights (not reports), for this post this person is you!
• Provide him/her with a small group of one to two people who will put the tool to hard-core use.
• Email the entire company (only slight exaggeration) to let them know you are selecting the tool.

Step 1: Implement a web analytics solution. (Day Two)

• If you want to use web logging data go get ClickTracks Appetizer
• If you want page tagging data go get Google Analytics
• If you don’t trust either go get StatCounter (even if it gives you bare bones data that is ok)

It takes one hour to get your hands on any of these tools. Five minutes to implement GA (tag into site footer, hit save, go get a drink and wait for data). Couple hours to implement CT, most of which will be you locating your website log files.

[Going out on a limb the recommendation is to stick to a javascript tag based solution because in the end it is quite likely that that most solutions use it so you might as well take a plunge now. That would be GA or StatCounter above.]

Step 2: Start using simple reports and process of creating a intelligent audience in your company. (Day Three)

• Email your core users (whose job 100% depends on the website, so very few) a report that shows: Traffic (Visitors), Referring URL’s, Search Key Phrases, Top Viewed Pages
• After a week set up a in person meeting to get feedback and questions
• Create second revision of reports
• After a week ask for feedback, in person, go to step 3 (because the feedback will tell you that the reports are not enough, show wrong stuff, they want more! : ))

In three days you went from nothing to actually having real data flow into your company and you are getting smarter by the second.

Step 3: Teach yourself the limitations of web analytics, tagging, numbers not matching, need to go redo your website IA / URL’s / ID’s / Cookies / data providing facilities. (Day 17)

• By now you have found that the reason you can’t answer questions is not the fault of your web analytics tool.
• Make a list of all the problems (you’ll need someone with slight technical skills if you don’t have them), they’ll usually be: URL structures, missing data (need for new url parameters of cookie values), maybe updated javascript tags, etc.
• Prioritize the problems (you plus your core team)

Step 4: IT “rules”! Cross your fingers, dive in. (Day 27)

• Partner with your IT / Website Tech team to roll out the fixes you need to get your website to cough up data
• This is often a painful process, cash in any chips you have (borrow some if you have to!)
• Don’t forget to keep your reporting going and enhance as new data becomes available
• Slowly increase your circle of data users

(Step 4 can take a lot of or little time depending on the size and mindset of your company. I am going to optimistically say this will take a month, YMMV.)

Step 5: Do a honest and deeply critical self review of where you are. (Day 57 or infinity)

After two months is your challenge probably is:

• That you have no idea what to do with the data?
• That you find that you don’t know how to use the tool?
• That the tool does not provide you with what you want?
• That web analytics simply sucks?
• That nobody buys anything the tools is telling you?

In five steps you have accomplished the impossible: You actually have a lot of in-depth knowledge of the tool (Google Analytics, ClickTracks or StatCounter), you have data flowing into your company and you are fixing things on your website that would have caused problems for any tool (and you are doing this without a single cent going to your web analytics vendor).

Remember even with time in step 4 you have have moved further faster than in the standard nine step process.

What you do next depends on the output of your Self Review (step 5). My guess on where you’ll be:

• You’ll find out that reporting does not equal analysis and that you need a major upgrade in terms of the web analytics skills in your company [20%]
• You’ll find that data or tool are not your problem, it really is your company culture (both in terms of using data or getting your site tech teams to do things to give you data) [20%]
• You’ll find that Google Analytics or ClickTracks completely fill all your company web analytics needs [20%]
• You’ll find that ClickTracks or Google Analytics are not the right web analytics tools for your company [20%]
• You’ll realize that Web Analytics (clickstream data) is not sufficient to find web insights so you’ll take the money you would spend on web analytics vendors and spend it on Experience / Research analysis (see Trinity) [20%]

If the limitation is truly the tool then now you are ready for intelligent choice for a tough and practical selection process. My recommendations (whenever you are ready to chose a tool):

• Your RFP should now contain the specific and actual problems are are having with the limitations of your current tool (Google Analytics, ClickTracks). And it should only be about the tool. No vendor can come give you a warm hug and solve your problems with your inability to capture data or have people to do analysis. Those are your problems to solve. [UPDATE: For more on how to craft a optimal Request for Proposal (RFP) strategy click here.]
• Select differentiated vendors

• Remember the Big Three are pretty much the same in functionality you need (except in 1% of features that niche businesses will value), so if your final list is the Big Three then you might miss out on a real differentiated choice. 
• If you want a real comparison bring in a vendor that is radically different. So consider CoreMetrics, Visual Sciences (not HBX), Instadia, IndexTools, NetInsight, ClickTracks Pro etc, each of whom, IMHO, have something significantly different to put on the table.
• Do a real Proof of Concept (implement the final set of vendors on your live site and compare with the free tool you were using to see if there is real differentiation).

So why should you do what some random blogger is recommending? If you follow the five step alternative process suggested above you’ll benefit from the following outcomes:

• You were not paying through your nose, ears, eyes, etc to first fix the problems you had in your company. (Data capture, basic intelligence up-leveling etc.) If you pick a expensive and expansive vendor you are paying them to simply identify that you have problems, you have the option of doing that for free (and since the process of fixing your problems is months it is a nice chunk of change).
• You have created atleast a core group of people who know what Web Analytics is everything that is frustrating and joyous about it:

• IT: They learn that it is trivial to implement a web analytics tool. Javascript tag, copy, paste, save, done.
• Web Site Developers: They learn all the little things that go into providing data that is business critical for actionability. Parameters to be added to url’s, page names to be updated, duplicated links in pages to be “lid’ed” so you can track them etc etc etc.
• Report Creators: They’ll learn that web analytics is less about report writing and more about analysis and this list might be handy for them to evolve to the next level.
• Web Analysts: They’ll learn that that they can use any tool to find a answer because they are Analysts. And of course that they have ultimate job security and we love them so.
• Marketers: Magic does not exist. Forethought has to be put into doing campaigns and coordination with web site developers and analyst before launch so that things can be tracked after launch. And no the web analytics tool will not make them coffee each day nor give them a massage.
• Business Leaders: Do a true assess of if they have the right skills in their folks, major gaps in their processes and that the true cost of web analytics not tools, it is the people (see the 10/90 rule).
• You will have chosen the best tool for your company, with your eyes open and you would have upgraded your company’s Analytics sophistication in the process.
• You will be promoted to VP of something as a reward and now have new worries to deal with.  [There is a very very small chance that you would have been so frustrated that you are ready to quit!!]

Ok now your turn. What do you think? Is this truly a radical alternative? Could you, or I, ever pull this off? Would you change something about this approach? Please share your feedback via comments.

[Like this post? For more posts like this please click here.]

In July I wrote a post on the power of visualization to communicate massively complex amount of data in a very effective manner. The example used as the illustration was Death and Taxes by Jesse Bachman. So it was thrilling to hear from Jesse that there was now an updated version for 2007, which was improved in many significant ways to boot.

I am inspired by Jesse’s work and have a poster of it next to my Minard poster. In this post it is a pleasure to present the 2007 version of the visualization (I don’t like calling it a graph!), some observations and an interview with Jesse Bachman.

To begin with here is a summary of the US Federal Budget, to ensure we first present the complete picture:

Total National Debt: $ 9.35 Trillion (up 7%)

Total US Budget:  $ 2.8 Trillion

2007 Federal Non-Discretionary Budget $ 2.22 Trillion

• $ 586 Billion: Social Security
• $ 367 Billion: Unemployment
• $ 394 Billion: Medicare
• $ 276 Billion: Medicaid
• $ 243 Billion: Interest on National Debt:

2007 Federal Discretionary Budget $ 983 Billion  (this is illustrated in the Jesse’s visual below)

• $ 632 Billion (up 64%): Military & National Security Related
• $ 350 Billion (up 36%): Non-Military

2007 Total Budget Deficit $ 354 Billion (hence $2.22 Trillion plus $983 billion does not equal the total budget of $2.8 Trillion)

 Here is Jesse’s representation of the numbers above (except the debt):

The focus of Death and Taxes 2007 is only on the discretionary budget only because the is the portion that Congress has to re-approve every year and it is the part on which the Congress has more control to move money around (and hence directly impact our lives).

Of course you can’t see anything! Trust me it is beautiful.

Click here for a high rez version of the image (when you get there save a copy).

Click here for a “viewer version” of the image. Use the controls at the bottom of that page to increase or decrease the size of the image.

Click here to go FAQ’s on Jesse’s official website.

Couple of things that have been upgraded from the earlier version of the visualization are:

• One extra layer of detail, from Department to Agencies to Core Expense Groups. Think: Department of Defense (439 billion) to the Army (111.8 billion) to Procurement (16.8 billion) to Humvee’s (.582 billion).
• Change in the individual budgets, red for decreases and green for increases.

When you look at the high resolution version you’ll see details like these that were interesting to me.

Department of the Air Force (130 billion):

You can see contrasts like Personnel budget going down a bit while the overall budget and other expenses went up.

The Classified segment (let the conspiracy theories start!! : )) (44 billion):

You’ll also find the cost of the Global War on Terror (110 billion) on another part of the visual.

Since not many people think about it, Department of Agriculture (tiny 19.7 billion):

Marketing and Bureaucracy up, Conservation and Research down. You don’t see it above but Farm Subsidies up as well.

Finally something dear to all of us, life. Health and Human Services:

Overall up a bit, Bird Flu preparedness is new funding which is great but most other things suffer: Cancer Institute, CDC, Substance Abuse and Mental health, Administration on Aging (though I have no idea what this department does anything to do with aging should go up probably because of the aging US population).

There are lots and lots more details that you will find interesting, regardless of your political, social, economic affiliation.

For example (I apologize this is a rant): Having been a rather poor student it is sad to see funding for Student loans and Pell grants is down in big numbers (down 18% and 27% respectively), as is funding for Higher Education (down 30%). Look at Department of Education towards top right.

Here is my interview with Jesse that covers some of my questions about the visual:

1) What is your primary motivation behind doing this visualization?

Jesse:  A few years ago I was a bit disturbed by two things, how large the military budget  is, and also how it is presented in government pie charts.  I wanted to present an image that reflected the discretionary budget, where the action is, and how much of it is spent on the military.  I had no grand aspirations of making posters or anything, I just made the graph and put it on deviantART as personal artwork.  I’ve always done lots of political artwork.

2) Given all the “feedback” from the last one you did, why would you want to deal with all that all over again?

Jesse:  The first one I did for the 2004 budget and I sold prints online, didn’t sell very well but it got a lot of comments.  I forgot about it for a year and half until boingboing.net picked up on it, for the first time.  Interest exploded and I started getting comments from people outside the deviantART community.  Professors,  military contractors, all walks of life, and they generally liked it, although there were several reasonable detractors, of which I tried to address in the 2007 version. 

I decided to continue this project on a grander scale because it seemed to permeate beyond fellow artists… it resonated with concerned citizens.

3) What tool did you use to do the visualization? Why did you choose it?

Jesse: The graph was contracted entirely in Photoshop.  Due to all the photos and no real need for vector data this was the easier solution for me.

Unfortunately, government budget data is rarely in usable excel format so all the figures were plucked by hand from various PDF’s usually 100’s of pages long.  A friend wrote a simple program to calculate the diameter of a circle whose area is proportional to the master circle. All I had to do was input the items funding in billions.

4) Since our blog readers are all analytics people would you like to share some tips on why you choose this particular format to depict the data? Did you try others before you settled on this one?

Jesse: Well for me, the importance of the graph is not the amount of dollars spent but the relationship to other programs.  Dollar amounts in the billions are largely incomprehensible to the common tax payer without any frame of reference.  So I wanted everything to be proportional to the master budget, so that an item with a relatively minuscule budget looks relatively minuscule.  The bubble format was the first idea I had and it seemed to work out. 

Actually the hardest part of the creation was finding images to represent each item.  Some were obvious but others were agonizing, I mean what does one visualize when they think “operating program”?  Also, due to the connecting lines and all the text needed, arraigning the bubbles became quite an irritating jigsaw puzzle.  Places where I had to sacrifice form for function are the defense wide and HHS bubbles. 

Also I tried to maintain a clockwise alphabetical order for the departments which added some difficulty.

5) I noticed you are using Google Analytics to track your website, are you watching any key metrics? (Sorry had to ask, we are web analytics geeks!)

Jesse:  Not at the moment.  Boingboing.net caught me a little off guard since the site only went up on Saturday.  I do use a separate script to track my to-the-minute refers (Link: http://www.thebudgetgraph.com/refer/index.php).  I pretty much try to join any discussions I see going on and try to correct inaccuracies.  Also, this will help me avoid redundancies when I start some push marketing.

6) You have worked very very hard with a lot of very complex data. Any tips for the rest of us who do reporting and analysis (of course of a lot less important data)?

Jesse:  I always like seeing data thats not a pie, bar, or line graph.  A little forethought into the design elements can transform a mundane xyz axis into artwork people would want to put on their walls.  Or at least stand out in reports.  Be creative.

– – –

I have to admit that for me it is a lot more art than to many other people. You can buy a 24” x 36” poster for Death and Taxes 2007, click here.

I thank Jesse for taking four months to create this wonderful work of art and also for taking the time to do this interview for our little blog.

PS: Two months ago I became a citizen of this great country. I am very much on a democracy high and into exercising my rights (what can I say democracy is a drug). So one action I am going to take is utilize a service on the official site to send a copy of Death and Taxes to my US Congress representatives. For only $10, including shipping, you can write a letter and send the poster to your representative (the site will print your letter and mail the poster for you).

You could be a Democrat or a Republican or a Green or a Libertarian or simply an American. You will find things on the poster that will relate specifically to you, things you would want to communicate to your Senator or Congressman / Congresswoman. Click here for this service and exercise your wonderful right.

PPS: Click here to see my web analytics chart that is just a tiny implementation of the principles from Jesse’s visual.

[Like this post? For more posts like this please click here.]