Pick One, Just One Web Analytics Report, Go!

One"You are stranded on a deserted island in the middle of the ocean. You can only pick one web analytics report to take with you on this deserted island.

Which report would you take?"

There are so many things wrong with the above scenario!

While not quite framed that badly, I do get asked a question like that quite a lot. And it poses the same challenging thoughts as in the above scenario.

Just one? Why only one? And how did I get on the island? What website am I looking at? What kind of website? Is there a computer on the island to get fresh data? Is there fresh water? :)

All tough questions to answer. With so much choice it is hard to pick one, and if you don't get paralyzed then it can reveal a lot about you.

After having thought about it for a little while (ok for just a few minutes), regardless of the website and its purpose, I would pick this web analytics report as my golden choice…

Outcomes By All Traffic Sources:

greatest web analytics report on earth

I am sure you gasped!

Perhaps you're thinking: "of all the wonderful and vast number of metrics and dimensions this is the best he could come up with?"

Are you disappointed? [Is anyone out there absolutely struck by the astounding brilliance of my choice? No? Yes?]

Let me explain.

You are right, there is a veritable ocean full of metrics in our web analytics tools, be it Omniture or CoreMetrics or WebTrends or Google Analytics or ClickTracks or whatever else catches your fancy. But the report represents two things I care more than anything else:

1) Sources of traffic, and hence just a hint of customer intent.

I have come to believe if I have sources of traffic then it can give me lots of information (not all, just lots) about what kind of people I might be attracting by where they are coming from, I might also get other delightful clues.

For example I am sure you all know that I am a big fan of direct traffic. If you tag your campaigns correctly (and come on who does not do something so basic!), then direct traffic represents free traffic because it is people who come through bookmarks, typing in your url or other such activities.

It is also traffic that is familiar with you (hence they visit directly) and so it also typically represents returning visitors (good for business). In the above case I was happy that direct is so big (and notice it converts higher!), but if I was a business I would probably want a even higher number!

Then I know that stumbleupon and Wikipedia and del.icio.us represents, for me, brand new visitors who might be topically interested in recent stories (in your case, probably recent promotions or product launched or a interesting seasonal effect).

Finally the dominance of google is clear in this case and for me it is a good thing because it means I have done a good job of search engine optimization (SEO) and that is translating into a steady stream of new traffic that gets introduced to the blog. I was delighted that there is so much traffic from images.google, a validation of the time and effort I put into tagging each image with relevant descriptions etc.

Net net, I know a lot about who is coming and by that I know if my core acquisition strategy is working, if I am getting the right kinds of people on the site, and which sites are my BFF's!

traffic sources and website outcomes report

2) Outcomes, so I have a site, big deal, what's the impact of the site on me / my business / the world.

I find that most website owners / analysts do a very poor job of measuring outcomes, or even understanding it optimally. This is one of the core reasons many senior executives in companies don't take web analytics seriously, we the providers of data / analysis stuff our reports with Visits and Visitors and Time on Site and Bounce Rate and Referrers and Top Exit Pages and . . . . . . but nothing about explaining outcomes.

I am a big fan of outcomes. "Show me the money!"

I am a fan because if you measure outcomes accurately you will find multiple outcomesthat your senior executives suddenly care about your web analytics reports, they will ask you good questions, they will seek you out rather than your knocking on doors that never open.

Of course I like measuring outcomes because to shows the value of the website and its ability to add to the bottom line. And I don't just mean measuring overall Website Conversion Rate. That is a start.

I like measuring Macro Conversions and Micro Conversions. Truly understanding why a website exists and then measuring all the ways in which it adds value.

In the first screen shot above you'll see I have two goals listed (I actually have three). So I am measuring all of things that add value to me. I have also highlighted in pink the goal that I care the most about. Do that for your business and you'll be kosher.

Now looking at the intersection of the Goals and Traffic Sources helps me understand what sources are truly valuable to me. It helps me separate my real BFF's from the pretend ones! :) For me that is analytics.blogspot.com and Wikipedia and direct!

With absolutely no other data I know the kind of websites I want to have more relationships with, invest in getting relevant traffic, get more bang for my limited acquisition budget etc.

That's it. Here's a helpful visual of my view:

critical web analytics priorities

Surprised I don't want anything about the site? Leaving it to "magic"? :)

It is a matter of prioritization. In the worst case scenario if I am a good website owner / marketer / analyst then I would bother enough to get to know it really well. In that case I can take a stab at understanding what magic is happening that is cause the various traffic streams to do so well, or poorly.

It's not perfect but I can get by (especially if I can dissect outcomes closely).

Important Update: By "web analytics" I meant clickstream, you know Omniture, Indextools, Google Analytics, Statcounter etc. Not web analytics 2.0. So no surveys or competitive intelligence or website optimizer etc. This is to make the challenge even more fun. :) I am on the record saying that if I could only have one report I would take the 4Q greatest survey questions ever report! :)

Ok its your turn now.

Imagine you are stranded on a deserted island and you got lucky, its just you and Brad Pitt (or if you prefer, Angelina Jolie!). You can only have one web analytics report. Which report would you most want to have? Just one. No drill downs. No excuses. Just one single report.

Won't you please play along and share your perspectives with us? What would you choose? Why?

I'll send the person with the best answer (I decide) a signed copy of Web Analytics: An Hour A Day! [Check out this post for the winners of this contest!]

Thanks so much.

In case you really are stranded here's a helpful article: How to Live on a Deserted Island.]

[PPS: Completely off topic:
I recently resigned from my position on the Board of Directors of the Web Analytics Association. Some of you have inquired why. It is based on the number of things I have on my plate and where I can add the most amount of value. I have cherished my time on the WAA board, it is a wonderful industry organization that is in the midst of rapid evolution to keep pace with the ecosystem. I was thrilled to be a small part of it and I am confident that the fantastic slate of new Directors will help accelerate that process even more. I was and remain an active supporter and participant.]


  1. 1
    Adrien says

    It might sound stupid, but I would prefer to have a VOC report, as if it is the only report I do have, this VOC report will actually allow me to do some actions on the website (assuming that I have the comments included). Where my conversion report will tell me what is happening, but it will not help me increase my success

  2. 2

    Adrien: My apologies, in my rush to get this post out, just before I get on a plane to Spain, I totally forgot to mention one important thing. I was requesting your input on a web analytics report, as is corementrics, microsoft adcenter analytics, nedstat etc etc.

    I 100% concur with you on the VOC. If I had a full suite of everything web analytics 2.0 then I am on the record as saying I would only pick the simple three greatest questions survey.

    But I wanted to make this game more fun, and challenging. :) I have added a update to the post.



  3. 3

    I'm really having a hard time with this, Avinash. I'm trying to find a way to break the rules and take one report with me on a deserted island. Does a dashboard report count? This way I could add any widgets or any reports I needed to have with me, and that would be one report.

    But, I think you're looking for something much more specific. So I'll keep it simple and take the Top Landing Pages report from Google Analytics, found underneath the Top Content section. Each one of your entry point pages is listed, and the report itself only gives you three metrics – Entraces, Bounces, and Bounce Rate. I could at least get an idea of how my website's pages are being received, and if they are working in terms of keeping people interested in whatever it is I have on my website. Of course I'd want to then segment each page by source, and also by keyword…but this desert island doesn't have a segmenting option so I'd be screwed there. But at least the Top Landing Pages report is a start, and you gotta start somewhere I guess.

    Thank you!

  4. 4

    Well you probably stole the best clickstream one Avinash with your traffic sources/goal result combo report. I guess I would take the GA Benchmarking (beta) overview report [is this outlawed from the game also?]. While its not always easy to figure out a relevant category to place your site in it definitely provides some great information for showing execs about how well/crappy your site is doing compared to the competition. Which should result in more attention to your web analytics position either way!


  5. 5

    Joe: Given the "limitations" great choice, you are in the running! And you are right, no "dashboard" / data grab. :) As much as possible a standard report, not mandatory but as much as possible.

    Thanks for playing the game.

    Dan: Ahhh…. very interesting choice on your part. I would not have thought of that one. Excellent!

    {Is it not amazing that it is such a struggle to fine one awesome report that would cough up insights that are actionable?)

    I am hoping for more answers like Dan's, things one might not think of right away.

    Thanks! How fun!

    PS: Ok got to close the laptop, the air hostess is giving me "the look". :)

  6. 6

    If the bencharmking entry above is outlawed then I would take the same report you described traffic sources / goal completion, but also add in a trend comparing the current month data vs the last month. Trying to beat out last months results, if even by a little is one of my favorite motivations.

  7. 7
    Rob Weinstein says

    The more fundamental question is why we are taking a web analytics report with us to the deserted island. Perhaps the best answer is a really thick report to help keep the fire going (survival tip #4). :-) In all seriousness, though, I love the idea of measuring outcomes and think you're 100% correct. I've only recently started reading your blog (and watching your videos on YouTube and reading your book and building my shrine of homage to you … errrr, strike the last one :-) ), but I'm completely in awe of your enthusiasm, brilliance, and ability to distill complex issues down for the masses. Keep up the great work!

  8. 8

    After some thought I think I am going to say I would take the Google Analytics Campaigns by goal report. I do not mean to cheat the system by using a similar one to you but I have a few reasons.

    I would like to know how my e-mail / non digital campaigns are doing. I am spending money on them and I do not want to come home to find out I have no money. Also it is a particularly important report because it helps for us to know if the money we invested in campaigns is producing results or not.

    The other reason is I can tweak in some optimization into this because if I send out and e-mail campaign I will send it out to a small percent first, check the campaigns report to see which e-mail version produced the best results then send out the best performing e-mail to the rest of the e-mail list.

    I also like to point out flaws. My major flaw is that if the report was printed on paper then I would not have as much paper to use for fires as if I chose a larger report. :)

  9. 9

    I would take my HBX Campaign Hierarchy report. Assuming all of my campaigns are set up correctly, this report lets me know (roughly) who is coming in, how long they are hanging out, if they are buying, and roughly what we're making. All of this would help me somehow communicate from my deserted island what the heck is happening and areas in which to make modifications that might be useful.

  10. 10
    Emily Fazio says

    I like this game! I’m a locally owned, service orientated company, and I am very interested in growing my client base to increase my revenue. I am not interested in international traffic to my site, or traffic from outside my service area. While this does generate cool points and a warm fuzzy feeling, it does not pay the bills. My report in GA is ‘sources by state, compared to last month.’ This tells me how many potential or current clients visited my site, how they got to my site, and how well my goals were achieved per source, all compared to last month. There’s a lot I can do with this data! (This report can be viewed by site usage or by goal conversion. I think goal conversion by source view is most actionable, but there may be cases where plain site usage statistics could reveal a piece of the puzzle you may be looking for….yay for choices!)

    What is equally valuable is what is not on my report. Let’s face it, I’m on a desert island with Brad Pitt and looking at an analytics report, I am very aware of what is supposed to be driving traffic to my site. If I’m doing something that does not show up on my report, then it is not driving the visitors I need, and its time to drop the effort and put that money to use elsewhere. Suntan oil maybe??

    I understand that my report is basically your report, segmented down a level. I’m not meaning to cheat the system! I think it’s important to recognize that by choosing your report carefully, you can gather actionable data from what is omitted from the report as well, putting the analyst on the web analytics pedestal.

  11. 11
    Emily Fazio says

    Footnote! What is important to my report, is that I had a specific question in mind and tailored my report to get the best actionable data.

  12. 12

    In a classic case of over-thinking, I am wondering what I would be able to *do* about anything I learn in a report when I am on the desert island. Can I fix stuff? For instance – and I know this is sad – I first thought of things like "search results not found" and error reporting. I am thinking that while I am stuck on the desert island I have to hope that my content is just wonderfully evergreen (or highly reference-able like perhaps this blog ;) ) and that people are just self serving. Anything that gets in the way of them serving themselves is a big (albeit decidedly un-sexy) problem.

    For the record, I think that if Brad Pitt were there, I may suddenly find myself disinterested in web analytics. :) Hmmm.. maybe in the long run I am better off with a report that requires complex interpretation exercises. Can I start over? ;D

  13. 13

    For me, conversion rate by keyword is a most. Not only it give you insight on what your visitor are looking for, but also how well you deliver.

    But of course, I prefer to have plenty of choice when the time come to drill down further.

  14. 14

    Best Single Report Challenge…

    As you state, this question is difficult to answer as the type of site (e.g ecommerce/enquiry/publisher/blog) effects the chosen report, as each site has a different aim (e.g traffic/enquiries/subscriptions/visits).

    Assuming stats package is setup correctly i.e. all pages are tagged, Navigational search are accounted for, ppc links are tagged, internal IP`s excluded, then I would go for…

    * 1. Campaign Medium >> Ecommerce Tab >> ROI report >> sort by Profit descending (this report does not actually exist in GA, it is just a concept report).

    This shows at a top level which medium (e.g ppc/organic/direct/email/banner/cse/affiliate) is generating the most profit based on spend and secondly which medium should have funds increase/decreased (i.e which has best ROI).

    My point is it is not just about the report it is about intelligent use of data, when integrating with the stats package to allow for Profit & ROI reports. Similarly, if customerID is stored on login then…

    Report Screenshot: http://tinyurl.com/528o6n

    * 2. Visitors >> User Defined >> Ecommerce Tab sorted descending by Visits >> compare to per visit site average

    Shows most valuable customers and insight into buying frequency, also highlights visitors who have logged in returned to the website but not bought!

    Report Screenshot: http://tinyurl.com/4brgtd

    3. Printed version of Browser capabilities >> Java support. Why pick this? Because although this report provides absolutely no useful data and even confuses user who think Java is the same as JavaScript… The report would be useful on a desert island… as there is shortage of toilet paper ;-)

    Report Screenshot: http://tinyurl.com/4jq3qm


  15. 15

    The perfect report depends on the web site. I agree that if you still have to work hard on the incoming traffic, I guess "Outcomes By All Traffic Sources" is certainly a good choice.

    However, I would go for the "funnel visualization". I adore this report because it is (a) intuitively appealing and (b) gives you the opportunity to optimize your conversion path without much effort.

  16. 16

    ps: keep up the great work!

  17. 17

    The one report that I would take to the island would be the HBX Campaign Conversion report. It gives clickthoughs and conversions to all the campaigns tracked, so as long as you a) know what your conversion goals are and b) are tracking your referrers, your organic traffic and then your "campaign-y campaigns" you can see their performance against each other (adding context, a Kaushik NB) as well as being able to extrapolate out (or at least hazard a guess at) your general visits. So from the one report you could guess your traffic volumes, see how each of your campaigns is performing day by day (we get daily updates on my island) and most importantly assess their performance relative to each other to see who you have to give kudos to and which campaigns need a little love and attention once you get off the island.

  18. 18

    Hi Avinash,

    I´m analyst at a german portal and we have lots of local content, something like yellow pages and a catalogue of relevant internet addresses for all cities in Germany. Based an this content lots of our traffic comes from google and people are explicitly looking for a certain piece of information. So, similar to the task completion rate you suggested, I would take a report that shows me the percentage of people who left my site using a provided exit link. Thus I get a feeling, if me visitors found what they where looking for (because they clicked a link) or if they left unhappy using the back button or typed in another URL (no exit link). Not as nice as task completion rate, but the best I can get out of clickstream data.


  19. 19

    After seriously questioning my sanity for thinking about web analytics while stranded on a desert island with Brad Pitt…

    I get no points for originality. Before I read your answer I immediately thought "Referrer Groups!" Since my site is a publishing one, I've added a conversion rate for subscription, as well as subscriber/non-subscriber status to the standard visit, page views per visit, and bounce rate metrics so I know which sources bring new people and which ones bring already-subscribed customers.

    Alternatively, I'd look for a report that tells me how to change Brad Pitt into George Clooney… haven't found that report in my WA tool yet, need to talk with my vendor. ;-)

    A PS to your PPS — Your committee misses you, and we are privileged to have worked with you! But Robbin is awesome, rest assured we are in good hands!

  20. 20

    Some of the natural choices have already been mentioned by you and in the comments. And so I will go out on a limb to make this game interesting :-).

    If I own a site and am on a deserted island, I will use it as a time to do some Zen meditation on my site (away from the emails, blackberrys, meetings etc.). So from this aspect, I will take the Most Requested Content report with me. This will give me an opportunity to study the rationale behind the top content; if I have optimized them to give the customers a delectable experience when they visit; are there content pockets I wanted to drive the customers but is not in the report; think about my SEM/SEO and by the time I am rescued — to really have a plan to make the website a pleasure to visit.

    At the end of the day, I believe in the philosophy — Perform your duty and not for the fruits thereof. I believe that if one diligently strives to improve the various aspect of the site keeping customer needs in their mind (not their needs), their fanbase and bottom-line can only show an upward trend.

    P.S: And since I have Ms.Jolie at hand, I am going to ask her for advice on how better to Market the site. No one can beat the Hollywood stars when it comes to the Art of getting public attention :-)

  21. 21
    Melissa says

    After thinking just a bit about this, I really have to say that my favorite is the Coremetrics On-site Search Report. I find it to be one of the most actionable reports, and what I like to refer to as my low hanging fruit. If I only have 5 minutes in my day, I can always gain insight and take action on that data.

    I can easily see how customers are talking about the brand, what are they searching for, how many results we are returning to that search, and is their searching session influencing them to convert. I often sort the report by zero results returned and # of searching sessions. That way I can focus my efforts on the search strings that have multiple searching sessions, but we return no results.

    In addition to optimizing my on-site search tool, I can also use that data to optimize my PPC campaigns. Quite often customers are searching for similar terms on the search engines, so the data helps me identify opportunities to optimize my keyword buys.

  22. 22

    It's always good to hear you suggesting that people actually look for results instead of exquisite detail about random walks.

    Disregarding the game for a moment, I'm dying to know what the goals in your screen shot are. Do you have a goal of people getting to your About page?? And reading every one of your posts??

  23. 23

    I'm an analyst at an electronics retailer and we use WebTrends. I would take the Single Level – Forward report if I was on a desert island. I use this report almost daily to determine where my visitors are clicking on any page on our website. I also see how many users exit the page from this report.

    Using the forward path data, I can go to the page and see what, if anything, should change. Are the visitors looking for something else by clicking on the navigation too much? Is the visitor interested in something not on the main product pages? Then I try to give the visitor what they want.

    Usually I use this report in conjunction with the top entry page report and an entry page by referrer report to determine customer intent. However both of these reports have already been mentioned.

  24. 24
    Bryan Cristina says

    I picked referring sites even before I even got to the end of the end of the header.

    1) It's always interesting to see how vast a site's reach is.

    2) When you DO have access to other reports, they're a great segment to run things against

    3) And if I were on an island, the hundreds of very, very funnily named sites that always show up in those reports (look through them sometimes, it's amazing what's out there) would at least amuse me until I died from sun poisoning or drinking the hopefully supplied crashed cases of rum.

  25. 25

    Funny.. you say web analytics.. but it seems everyone is PPC.. centered.. we use analytics in SEO on an hourly basis..

    i use http://mcc.hitslink.com

    i would want.. the "Latest Visitors Report"

    it provides stuff like this for each visitor:

    6/3 7:02:24 PM
    Location:Las Cruces, New Mexico, US
    Referrer: http://www.google.com/search?q=sa...
    Search Term: satellite xxxxxxxxxxxxx
    No of visits: 1
    Network or Router Name:
    National Aeronautics and Space Administration

    my client is a government approved vendor of satellite services. some info omitted for business ethics.

    from this, we now know:
    1. where they ranked on the page.. #2 of 238,000,000
    2. what search term they used to find my client
    3. who was looking, when and what they searched for

    B2B sales cycles usually aren't something you can measure with a conversion report due to the high price tag and long sales cycle. However with this.. I can call the client and say… "did nasa buy something?"

    And because the client's sales team can watch this as well, they know where the lead came from.. =)

  26. 26


    Why are you proud of traffic from images.google.com? It has very low goal conversion rate — only 0.21% of users in this segment reached your "About" page.

    For me the facts that Google Images Search generated 20 times less traffic + that traffic doesn't really care about your goals — all that is a clear indicator that images don't really worth an effort.

    Wouldn't you agree?

  27. 27


    We recently launched a new version of our website and I am brand new to SEO and Web Analytics.
    We have been using Google AdWords for a long time without tracking conversions or goals because of technical roadblocks.

    I used to use Traffic Sources -> Keywords (Paid and Unpaid) vs Bounce Rate to try and get an idea what was going on.

    I think it was about as effective as making lemonade out of apples but now we are implementing goal tracking!

    Our online traffic is based on our advertising.
    I will finally be able to see which keywords are actually driving outcomes!

    Keywords -> Goals
    What do you think?

    Am I missing something?

  28. 28

    Last February I competed in the Omniture Web Analytics Competition at Brigham Young University. This competition uses SiteCatlyst data from a live retail website. Student teams of up to three people have a few weeks to prepare a presentation off of the data.

    My team had one "Golden Report" that stood out amongst all of the reports and actually guided the flow of about 85% of our presentation.

    This report was essentially a marketing channel comparison report. It compared all of our KPI's against each of the marketing channel and the site average. The site we analyzed was a retail site and our KPIs were Conversion %, Average Order Size, Average Order Value, and Yield. The report also showed the Revenue, Visits, and Orders (so that you could see the relative size of each channel).

    We found a tremendous amount of actionable data from this one report. It not only showed how each channel was performing, but it also showed us where each channel needed to improve. For example, if one of the channels had a high conversion rate, but a low average order size then we knew that for that channel we wanted to focus more on cross selling (these channels were converting but they were not selling as much as the other channels). If another channel had a low conversion rate then our recommendations for optimization was directed towards improving the conversion.

    Over 40 teams competed in this competition and my team took first place. This report was a key component in our victory.

    I know its a bit off topic, but if anybody is interested in learning more about the Omniture Web Analytics competition check out: http://ebusiness.byu.edu/web_analytics.php

    I also blogged about my teams strategy at: http://tuckerc.blogspot.com/2008/03/winter-2008-owac.html

  29. 29

    Good education.


  30. 30

    I thought I was going to get all the way through the comments without anyone else picking the report I had in mind but Melissa beat me to it. I was (and am) going to suggest the 'On-site search terms and results' report (through Nedstat).

    While other reports tell you what visitors to your site are doing and where they are coming from, this report can give you an idea of what they are thinking and that is more valuable. The report also gives you an idea of how your website responds to visitors when they tell you what information they want to know. It is one of the few reports that is directly actionable.

  31. 31

    Well, you took the very best one! Anything with outcomes is a sure-fire desert island pick.

    That said, I have a custom report that I am finding helpful. This report tracks onsite search terms segmented by constituency group.

    Our site serves four distinct (registered) audiences as well as prospects and other non-committed visitors. This report enables me to learn what each group is searching for and infer whether or not our search results are returning the pages they actually want. If there's a spike in a given search term, we may need to create content to meet that demand. If there are 50 visits but 400 page views for a given term, then there may be something I can improve about either the search results, our navigation, or our content.

  32. 32
    Kristen Nomura says

    Hey Avinash,

    Hope you're enjoying Barcelona!

    This post got me thinking that all the web analytics vendors should read this and re-assess how they tout their exhaustive list of features. People are constantly afraid about information they MIGHT need at some point that they lose focus of how to use the data from just a few key reports. So liberating!

    If I had to choose just one, it would be the Funnel Visualization report. The report you chose shows conversion rate, but that can disguise a lot if the site has multiple steps to conversion. What if I have a 5-step conversion process and 80% of people make it to step 4 and then abandon? The Funnel Visualization report would allow me to quickly identify where the process is breaking down.

    Of course, if I were rescued from the desert island I would then want to segment this report by traffic source to see differences in behavior. Well, hopefully I would do something unrelated to work as SOON as I was rescued, but you know what I mean :)

    "Our life is frittered away by detail… simplify, simplify." – Henry David Thoreau

  33. 33
    Shelby says

    This might be a bit too granular, but if I'm stuck on a desert island, I'm going to focus my efforts and try to do one thing really well.

    Using Omniture SiteCatalyst and knowing that our email campaigns are our highest converting campaigns, I'm going to use the Creative Element reports by Campaign and query for email campaigns.

    With this report, I can test email copy (especially link text) and test what is converting the best over time (I will be stranded for quite a while!). I'll focus my time on some major A/B testing of my email campaigns.

  34. 34
    Suresh Ramaswamy says

    Avinash, although you're looking at this through a WA lens, conversion by source is far too one dimensional. Typically 'Targeting' accounts for approx 40% of the response, thereby you are ignoring the impact of Offer/Reason to Respond (~40%) & Creative (~20%).

    I'd like to see source replaced by Unique Combinations of Creative X Offer X Placement and then check for goal conversions against them.

    Even in this scenario, the last click always gets the credit. I'd like to see the impact of latency on at least the last 3 unique combinations of above that led to the conversion, before I decide to optimize.

    I'd also like to add a cost/goal conversion, as spends vary, unless all of the traffic generated is Organic.

    Well, you said one report ;). Suresh

  35. 35


    The idea of a single metric is not so far fetched as you might imagine even though the very idea seems preposterous.

    Fred Reichheld of Bain and Co argued that there is only one worthwhile metric in business which he calls the Ultimate Question "Would you recommend to ao friend" – surely an easy question to ask in any online transaction.

    Fred argues correctly that there is no correlation between customer satisfaction or customer loyalty and growth. When you think about it, getting another person to recommend captures many of the prior interactions which must have been successful to generate advocacy and WOM.

    In reality the Ultimate Question is not a lead indicator though – it tells you what has already happened so it's not that useful in predicting and preventng adverse customer behaviour.

    It also does not tell you where customer losses occur or what to do about them.

    A better measure to take to a desert island would be a survey which asks for a rating of three questions –

    how well do you understand me and my needs?

    how well do you satisfy my needs?

    would you recommend me to a friend?

    Together these human responses track the degree of engagement with customers because only by engaging with customers can we understand where in the customer journey they decide to defect.

    Pure numeric web analytics should be complemented with basic insight into the emotional state of online customers.

    You can find a White Paper about the anthropology of human engagement at w ww.business-sensei.com

    The essence of engagement is in tracking how experiences (online or off) actually make people feel (ie alter their brain chemistry).

    So take the tripartite "understand – meet needs – advocate" assessment to your paradise island.



  36. 36

    Goals. See if Im making any money or not.

  37. 37

    I also think, that the Goal-Report is the best one. But: Unfortunately GA reports only four goal and I allready want to have more than four goal to report. Did anybody know if Google is working on that? I only get the informatione last year in Germany, that this limit isn't fix forever.

  38. 38

    Hi Avinash

    If I was on a desert island, I'd probably choose a weather report :)

    But since it has to be a web analytics report, I'd go for site overlay (click map), for the following reasons:

    1) It shows customer intent directly mapped to my website interface. I can see clicks, goals,… and I can take action and see what happens inmediately.

    2) It's an easy to understand report, so if I find somebody in the island, I will be able to explain them what I work on and they will understand me easily, and hopefully they'll find my job fun and hence they won't eat me :) – companies can be like desert islands when you're an analyst, so you'd better make sure you choose a report everybody understands, or you'll be alone with no resources, and that's the worst thing that can happen!

    3) It allows me to communicate with my website users: I can change an element of the interface, and see how the website users react (I could even be tricky and ask them questions, and see what answer they click on to get even more information! :)

    4) Since I have configured Google Analytics with several profiles with different filters each, I can see where users click and how they convert depending on where they came from (I have a profile for organic, ppc, referral and direct traffic) and who they are (I have profiles for registered users, new users, repeat users, …) – I am not sure however if you consider these as different reports, to me, it is still the same report. Anyway, the previous points are still good enough for me to make this choice.

    This was a great way of making me think, Avinash.

    Limitations are necessary to make you think and be creative. I looked for a quote to illustrate this point, and I found quite a nice one by Rollo May:

    "Creativity arises out of the tension between spontaneity and limitations, the latter (like the river banks) forcing the spontaneity into the various forms which are essential to the work of art or poem"

    Thanks for making me think, and sorry for such a long comment, but there are no comment-lenght limitations in your blog! :)

  39. 39
    Sam Simon says

    Avinash sorry to see you leave the WAA Board. Your contributions in helping drive the organization forward will be missed. As was mentioned in another comment, you have served the board admirably during your tenure.

    For too long the Web Analytics Association has been lead by the same people, who still seem to be in-charge. Having outsiders like you is key to driving change.

    Hopefully others will pick up the baton.


  40. 40

    Here's the longest comment on a blog in the history of mankind!

    My apologies for that. The last two weeks have been: SFO – London – Amsterdam – SFO – Frankfurt – Barcelona – SFO and I simply could not get to all the comments.

    I am absolutely astounded at the burst of creativity in the comments on this post, you all are simply awesome! And you have taught me a lot (as will be clear below).

    Without exception this is now my favourite post, due to your comments. Thank you, thank you, thank you!

    Here's my take on your take (and sometimes a small amount of push back or even a tiny argument or two):

    Pearce: Great idea in choosing the campaign report, especially good at stopping campaigns the company might be wasting money.

    (+ Jennifer + Erica + Tucker)
    My fear with just a campaigns report is that typically campaigns will account for 30 – 50% of the traffic (at most) and that means going blind on the other non-campaign traffic and hence a high price to pay. Non campaign traffic can be highly profitable (and of course "no cost"). Nonetheless atleast you will have a great handle on the paid traffic.

    Emily: I love your choice a lot because it illustrates two great points:

    1) You are so focussed on what you are solving for, local and the rest can go to.. well you know :), love the focus.

    2) Web Analytics is deeply personal, what works for someone does not work for another.

    Very nicely played.

    Angie: You have done a great job of showing how other outcomes that can be measured AND for good measure you threw in segmentation (sub and non sub status). That last one might be cheating a bit, but if anyone is allowed to then you are! :)

    Peter: Ahhh.. another very interesting answer that I would not have picked right away, and one that makes so much sense for your business. If I were to summarize it: Top Outbound Clicks and % of Site Traffic that clicks on outbound links. Awesome! Like Emily above, success is local and personal.

    Mark (+Kristen Nomura): Nice recommendation. I often say in my speeches that if I were to look for the fastest way of improving conversion then I would fix the funnel between Add To Cart and Submit Order buttons. Small number of pages, very high ROI. Though I have to fess up that as a single report, "on a desert island", my fear would be that it gives me too narrow a few of the website and its performance.

    But I suppose if you are optimistic about getting off the island quickly then the best way to use that time would be to fix the cart and the checkout process using funnel visualization! :)

    Phil: You get the highest number of points for suggestion #3 – you make a excellent point there, you are very funny, and you are thinking of things you will actually need when you are stranded! Bravo!

    Now according to the contest rules you get only one choice so I'll consider your first one, very nice mock up of a new report. I think it will be very valuable. Thanks for actually putting in the effort to create a image for our readers.

    Sébastien: Great suggestion about capturing customer intent (that is so hard to get in clickstream reports).

    But my fear would be that it would show only a small slice of what is happening on the website (unless you have a majority of traffic coming from search – for my blog it can be up to 50% so maybe you have something here).

    Ned: Nice! you even roped Angelina in to helping you! Now that is thinking outside the proverbial box! :)

    I was wondering how long it would take for top content report to come in, I think it came in much alter than I would have expected. You are right that it gives great insights into what people are seeing, but, purely and only IMHO, I worry about people only finding what we have. What about things they actually wanted? Top content report works if you believe that your site is perfect and that most (if not all) visitors find what they are looking for. For me, again stressing the IMHO, those are assumptions I hesitate to make (not after I have seen so many "sub optimal" sites in my life!).

    Dave: I have three goals configured for the blog (got to have something, else I'll hardly be a Analytically driven person!). I don't sell anything on this blog so the three things I care about are:

    Goal 1: All Posts: The number of people who find that link and go to my "site map" page. That page lists my posts in various categories. My hope is you'll find more relevant content and spend more time on the site reading it and perhaps discover that the blog is good and you'll sign up for the RSS Feed (my biggest goal on the site).

    Goal 2: About: The % of site traffic that sees my "about" page. Simple really, the blog is my brand on the web and hopefully the posts make you curious to learn more about me and organizations and activities I am affiliated with. Extends my brand a smidgen, and perhaps, maybe, makes me a bit more famous! :)

    Goal 3: Speaking Engagements: The number of people who visit this page. Perhaps you'll find a event close by and attend. This is good for me, I get to meet people who read my blog (like Gemma, her blog: Where Is Avinash When You Need Him?). It is good for the organizer because they pay me and perhaps I can send some leads their way.

    Not the most awesome goals in the world, my humble attempt at something.

    Eric: Great suggestion! You mentioned this one, and Pere in a later comment mentioned Site Overlay report, and it was a doh moment for me. I can't believe that I did not choose this (or site overlay). I think it might be because at some level I was thinking that I could only get one report and your report is about a single page and I would want more than one page in one report.

    Dennis: For it it was more that I love taking pictures, I put a lot of effort in creating them and so the fact that 1,417 people came to the site just through a image search and looked at them makes me happy. And if 0.21% converted that is good, mostly I am happy that I try to do SEO for those images and people look at 'em, so certainly a different kind of "conversion"! The "happiness conversion"!! :)

    James: I am surprised that even without goals you don't find that report to be helpful, typically if you look at bounce rates across paid and organic and for high bounces drill down to landing pages then it is really great at finding landing pages that stink. Of course adding goals will make the report even more valuable.

    If most of your traffic is from Search (Paid or Organic) then this is a great report to take on a desert island!

    Paisley: I have to fess up a bias: I am not a fan of real time data, and in as much not a fan of the "latest visitors report". Mostly because except for the smallest of websites there is no way that it scales (as in procurement, understanding and analysis of that report to get actionable insights). It also promotes a "single session" obsession, when in reality conversion is rarely a single session occurrence (as you say in your mention of long sales cycles).

    Phew! So glad I got that off my chest! :)

    Of course you have outlined a scenario that obviously works very well with your company. One other area where this might be of value is capturing in real time the website behavior (or search) info into a CRM system so you'll have more insightful leads.

    Thanks for pushing me hard to think about my own bias.

    John (+Melissa +Peter): I am a huge fan of internal site search (as you might well know!):


    If there are enough % of people using site search then it would make a great actionable report to take on a desert island (If not then I would worry that I don't have enough data about my site to truly have a macro impact). Segmenting it by visitor type is of course awesome!

    Shelby: My first thought was that the report might give you enough of a view of your website performance at a "site level". But if you are spending most of your acquisition budget on email then it might still be enough (even on a desert island!) to save costs and have a macro impact on the website.

    Pere: Great answer! I love the site overlay report and I can't believe I forgot it when picking my single report, and also for reasons you outline in your comment. I suspect it might have been because subliminally I might have thought that site overlay will show just one report (one page) and if I was on a island then I would want data for more than one page. But I suppose you can count it as one report and have all the pages on the site in it! :)

    Excellent answer below (and a fantastic quote at the end), it shows why you are such a legend in Spain!


  41. 41


    Hum true. I realize keywords reports can represent a small segment of your visitor. In my case I look at two websites, one with 2.5M unique visitors by month (great brand), less than 5% coming from search engines (not so great in SEO), and another one with 200 unique visitors by month, 80% coming from search engines.

    Also, lot of the keywords are brand related and not very useful to capture the voice of the customer. So it need cleansing.

    But I still believe this report (conversion rate by keywords) is the best way using only clickstream (a no no of course) to have a peek into customer mind.

    Like always, I guess the answer depend on the websites.

  42. 42


    Good blog post. I'd have to say non-branded natural search traffic. As an SEO consultant this is definitely the report that I hold myself accountable to for all my clients. This is what separates the men SEO's and the boy SEO's.

  43. 43

    How can I know the traffic in my site if I´m not an exxpert ?
    I use Alexa and Quantcast and the results are like "Day and Night"
    Any other recomendation for site with low traffic?

  44. 44

    For me it is all about the content report. I meant to trackback to this sooner, but I left town right after I wrote the post.

    Here is an excerpt of my post:

    My favorite report is my content report. Because that tells me what people actually care about, how much time they spend reading, where they are most likely to leave and how fast.

    I linked to the remainder of the post in my comment profile.

  45. 45

    Avinash, as always great post. This is an easy answer based on what you have taught me through your past blog postings. The single greatest actionable Web Analytics report is Visitor Satisfaction by site content area. This report combines both qualitative and quantitative information and basically screams at you to take action to improve the areas of the site for which it measures. It is almost too easy to take action upon, and action is the name of the game, isn't it?

    The content area dimension is to Web Analysts as the Visitor Segment is to Marketing Analysts. It provides a way to drill into your site deeper while not having to sift through the page-level details. True, the report may be scary or hard to understand for some, but the power that it possesses far outweighs the complexity.

  46. 46

    In B2B you better know what companies are visiting your website. Amount of traffic is less important.
    Know these visiting companies and the pages they visit in order to know their interest and be able to qualify them as leads or not.

  47. 47

    Hmmm. I would probably choose % Qualified Traffic vs. Total Site Visitors across different time. Making sure that the data table and chart show the actual values.

    Since qualified visitors are site visitors that you identified as a desired target or lead, you can choose to throw that ultimate equation into "qualified" visitors.

    For example: eCommerce — (Total visitors who bought your product) per Overall Visitors for the past two years. * Total revenue can be applied off the report to that ratio… Is that cheating??

    Lead acquisition site — Total Leads per Site Visitors. You can always assign $$ value to the equation as well.

    The point is, the ultimate report should reflect ultimate outcome against different date range so that your marketing or optimization efforts are reflected into the bottom line.

    If possible, maybe segmentation by traffic sources could be applied to the data as well. Showing as a bar graph in one chart…

    Thanks for this nice article Avinash, it really gets us thinking about the bottom line, and what to represent to key stakeholders.

    FYI, the site link attached to my site is one post I wrote about that kind of reflect my point here.


  48. 48

    Kris: This is a great idea, it is always important to stress focusing on the quality traffic.

    I am not sure that you are identifying "qualified" traffic. I think you are identifying converting traffic % (or outcomes %, as you say leads etc), and plotting it over time. This of course is very meaningful but it will only apply to a smaller fraction of your site.

    I tend to think of "qualified" traffic as, just as an example, say all the people on the site who were "in the game" to purchase something (and compare it to overall traffic). So maybe qualified traffic is all Visitors who saw a product page. Or came on the keyword Purchase Now. Or something like that. Then you can understand two things:

    1) Of all the traffic you get how many were actually in the game to
    purchase, hence qualified. This is very helpful to know.

    2) Of all qualified traffic what percent purchased? This tells you how
    efficient you are.

    It is hard to quantify qualified traffic, in a older post I had touched on that thought:

    Measure the Real Conversion Rate & “Opportunity Pie”


  49. 49

    I am with you on the goals/sources report being THE ONE.

    However, I am having trouble pinning down what Google actually defines as the "source".

    So, for example, if my goal is a product download, I would expect that typical behavior for a good proportion of customers would be:

    google search for products of the type they are looking for.
    Browse a few and bookmark
    come back direct to the sites they are interested in to download.

    So, in this instance is the source of my conversions "google" or "direct"?

    I have spent quite a while searching for a definitive answer to this, with no success – if you can help, or point me in the direction of some help, I would be really grateful.

    Kind Regards


  50. 50

    Seems you answered that question yourself? Is it not both?

  51. 51

    Clare: Here is the short answer:

    Like pretty much every other web analytics tool, out of the box, Google Analytics will always give credit for the conversion (download) to the last "campaign".

    Campaign is: a referring website, a email campaign, organic click, ppc ad, affiliate promotion, twitter link, etc.

    Here is some quick scenarios to really help you nail this down… :)

    PPC click -> email link -> conversion. Credit (the source in this report showing conversion) goes to email link.

    Email link -> PPC click -> Organic click -> conversion. Credit goes to organic click.

    Email link -> PPC click -> Organic click -> Direct -> conversion. Credit goes to organic click.

    Direct visit. -> conversion. Credit goes to direct.

    Hope this helps.


  52. 52

    The important thing is knowing where it came from originally tho.

    Avinash, what about these "not set" issues? Is there a way to simply not have these displayed?

    My dashboard is looking very different recently, partly due to reading most of your posts n latest book. ('actionable insights' love it!) But many of these, in both keywords>>goals and in ppc>>goals for example I still get these "not set" figures.

    It seems these are actually 'totals' and they don't have an effect on the other stats (apart from percentages) but they are very annoying.


  1. […] Another FOT, analytics king Avinash Kaushik, takes a more direct approach to analytics this time, encouraging folks to“Pick One, Just One Web Analytics Report, Go!” And don’t forget, you can read the thinks interview with Avinash Kaushik here. […]

  2. […]
    About two years ago, Avinash Kaushik asked “if you had to pick one web analytics report on a deserted island which report would you take.” I totally agreed with his selection of the outcome by all traffic sources report as I am also a fan of outcomes. (Show me money!) But what if I got stranded on this island with an iPad and because of this I got to cheat only a little? This post is about my one favorite report in each of the top three web analytics tools that I would.

  3. […]
    About two years ago, Avinash Kaushik asked “if you had to pick one web analytics report on a deserted island which report would you take.” I totally agreed with his selection of the outcome by all traffic sources report as I am also a fan of outcomes. (Show me money!) But what if I got stranded on this island with an iPad and because of this I got to cheat only a little? This post is about my one favorite report in each of the top three web analytics tools that I would.

Add your Perspective