Thursday, February 28, 2008

What Web Site Optimization tool are you using?

What Web Site Optimization tool are you using?

I conducted an informal poll on my blog to see what Web Site Optimization tools are the readers of my blog using on their site. I only had one question, I will be conducted a more formal survey in future. For now I wanted to share the results of this poll.

Here are the results:

The results are based on 65 respondents and respondents were allowed to choose more than one option.

As we see Google Optimizer by far is the leader used by 43% of the respondents. No big surprise there, it is free easy to implement and easy to use. “Other” , which referred to the tools other than those mentioned in the poll, was second with 21% of the respondent. Followed by Offermatica, In-house tool, Optimost and Touchclarity.

During past few days I have talked to many companies who are now starting to seriously look for a website optimization tool. So far either they had no A/B testing or used in-house ad-hoc A/B testing tool (hack).

What tool are you using?

Monday, February 25, 2008

Ad Targeting on Gmail appears to be using more than just Gmail data

There has been a lot of speculation about how Google will use all the data that it collects about visitors to its various services. Last year I blogged about how Google can use data it collects on Gmail, Google Search, Adsense, Adwords, Google Analytics etc. to do Behavioral Targeting and then continued with few more posts on Google and Behavioral Targeting. Google was first denying the whole notion of it getting into Behavioral Targeting but last year they introduced Personalized Search using Web History and then introduced in-session targeting on their Search Engine Result Page (SERP).

Now, it’s Gmail’s turn.

Users of Gmail are used to seeing text ads on the side and top of their Gmail mailbox. These text ads are based on the content of the email that a user is reading or gets. I was surprised when I recently saw an ad that was neither relevant to the emails I get nor relevant to my Gmail profile. It was however relevant to some of the search made in past by me on Google and relevant to me (maybe), which Google could have guessed from my name (maybe).

The ad I saw was from a Travel Company that specializes in tickets to India, and interestingly the ad was in my native language “Hindi”. Yes that’s right, the ad was in “Hindi”. My Gmail account, on which I got this ad, is for my semi-business use and never get any personal emails other than from my wife, who uses email address. But I have never got an email in “Hindi” in this account. My conclusion is that Google pulled my name, Web history from Google Account (that includes Google Searches, none of which resulted in me viewing “Hindi” pages) and email address to conclude that I might be interested in an ad from an Indian Travel Agency. None of my searches or my emails will indicate that I know Hindi so it was purely a speculation by the Google Advertiser or Google to think that I knew Hindi.

Here is the ad that I saw

Have you seen something similar? Comments? Questions?

Wednesday, February 20, 2008

Long Live Web Analytics

Ian Thomas, the man behind Microsoft Gatineau is making some bold predictions about the future of web analytics in his post titles Web Analytics is Dead. Long Live Web Analytics blog. He made two predictions:
  1. In three years there will be no Web Analytics vendors, but Web Analytics will be everywhere.
  2. In five years, all Web Analytics software will be free.

It is a great post and I like the way he is thinking. However I do not agree with him, here is my reply to his predictions:

  1. In three years there will be no Web Analytics vendor, but Web Analytics will be everywhere - I completely agree that Web Analytics will be everywhere in next few years. This is already happening, as he mentioned and provide several examples. However, I disagree that there will be no Web Analytics Vendor. Microsoft, Google, Oracle, Atlas, Doubleclick etc. will (or already do) provide web analytics as an add on to their products but there will still be a need for one web analytics product you can rely on to make strategic decisions. Can you imagine having 15 different web analytics solutions that all give you different numbers? There will still be a place for deeper and strategic web analytics tool. I agree that the functionality of web analytics tool will change and they will continue adding more products. They might not be known as web analytics vendor but web analytics will be a core product that they will provide. They will be providing more than web analytics and they are already moving in that direction.
    (Side note: I also agree that Omniture will buy an ad serving company, I made the same prediction last year . I also predicted that Oracle will get into Web Analytics, because I believed that web analytics will become essential part of marketing automation, Online advertising, CRM etc. and Oracle won’t want to be left behind. And actually about 10 days after making the prediction they did buy a company that has web analytics product).

  2. In five year, all Web Analytics will be Free – There is no free lunch, we all know that. As John Levitt commented on his post, most of the tools that provide web analytics as an add-on product, provide very light analytics and reporting. Their web analytics offering is to support or enhance the value of their products and offerings, so that they can keep up with competition. There will still be a place for deeper and strategic web analytics and that will come for a price. Web Analytics maybe subsidized if you buy other products from a company like Omniture but it won’t be free (bundle discount).
    A lot things can change in less than 5 year, Omniture can start using the web analytics data collected on the sites to create an online advertising network (they have pixels everywhere, they can buy an ad serving company, use touchclarity and offermetics to deliverer right ad to the right person at the right time) and then I can see that they can provide free web analytics tool if you let them aggregate user behavior and do targeted advertising. (My suspicion is that Google will be doing that soon too).

What do you think?

Monday, February 18, 2008

British ISPs get into Behavioral Targeting

According to Herald Tribune three major internet service providers in Britain (British Telecommunications, TalkTalk and Virgin Media) have jumped into the growing Behavioral Targeting and Online Advertising space.
Phorm , the company behind the BT technology, created an online advertising platform called the Open Internet Exchange. The company will use ISPs data to target right ads on the websites (publishers) participating in this network. These three providers, that represent two-thirds (66+%) of internet access market in Britain, have agreed to provide the customer’s surfing data to this internet exchange. The exchange will be open to any website that wants to join so smaller publishers will also be able to derive value from behavioral targeting.

ISP based BT has certain advantages compared to traditional (cookie based) BT Networks (e.g. Revenue Science, Tacoda, Blue Lithium etc.)

  1. Reach - ISP based BT networks have a bigger reach (granted major ISP participate in it - the one by Phorm does have that reach) compared to the likes of Revenue Science and Tacoda. Even though Revenue Science and Tacoda claim to reach 60%+ internet users, I don’t think they do collect data on and target 60+% of internet users. With their partnerships with other ad servers, they might be capable of reaching 60% of internet users but that is to just to serve any ads not collect behavior and serve targeted ads. ISP on the other hand can collect data on all their users and hence show targeted ads.

  2. Relevance – BT’s promise to provide relevant ads to the online visitors. Networks that work with large number of different kinds of sites (different verticals) can collect wide variety of user behavior data and accurately identify users segments. ISP based BT networks have the potential to collect much richer data (because of their reach) than any network like Revenue Science and Tacoda ever can. Revenue Science and Tacoda collect data on select sites (those that participate in the network) while ISP’s collect data on any site that is accessed by their customers. ISP can better understand their customers’ behaviors and hence serve more relevant ads than traditional BT networks.

Privacy Issues

However, ISP based ad networks also pose a bigger privacy threat than traditional BT networks, as I wrote in my post on NebuAd, another ISP based BT network. NebuAd, one of the first company to enter into ISP based advertising. NebuAd responded to my post and said that the data is anonymized so there is no issue of privacy. Phorm promises the same level of anonymity but I still think that the chances of privacy leaks are more in an ISP based network than they are in a traditional BT network.

Phorm also says that consumers are are in control, they can switch relevance 'off' or 'on' at any time at a site called, site that educates users on how ISP based advertising works. This is somewhat in line with what I predicted earlier this year, where I said:

Behavioral Targeting will continue to grow this year, however, there will be greater push for protecting consumer privacy. The privacy concerns will result in two things:
  1. Clear instructions (or links) on Behaviorally Targeted Ads that will allow behaviorally targeted visitors to opt-out of Behaviorally Targeted advertising.

  2. Opt-in system – Some networks (maybe new ones) will move towards opt-in rather than opt-out (I favor opt-in over opt-out as I wrote in past. So I am making this prediction that this year networks will pay attention to it). A new type of networks or services might come up which will allow users to be an active participant in BT and control who can use their online behavioral data and how they can use it.

Phorm is moving in the right direction by providing proper education (I mentioned the need for education in my post on Google and Doubleclick) and an opportunity to opt-out in an easy way. I think soon we will see the actual opt-out link on the ads served by Behavioral Targeting networks.


Thursday, February 14, 2008

Skills required for a Web Analyst - Part I

Web Analytics has become one of the hottest career fields and it is becoming hard to find people with web analytics experience. So if you can’t find people with web analytics experience then what do you do? Well, I interviewed a few of web analytics professionals and one of the questions I asked them was regarding the skills they think were important for a web analyst. In this 2 part series I have compiled their responses to that question, which should help you in understand what skills you should look for when hiring a web analyst.
If you are a hiring manager looking for web analyst or somebody who wants to start a career in web analytics, this article is for you.
So here are some of the responses:

I believe that an ideal Web Analyst should poses blend of Business and Technical skills. He should be articulate and understand the online business quickly.
-Trinadh Rao, Country Manager Web Analytics Association, India Web Analytics Manager at Franklin Templeton

A great deal of tenacity and, being sharp enough to make the connection between right and left brain items.
-Daniel Shields, Web Analyst at

In my opinion a web analyst needs to understand the data and site dynamics of their particular website. Should have decent technical knowledge, good customer focus and attention to details. Should intuitively pick insight from an otherwise burgeoning deluge of data.
-Jaisiri Chetty, Asst. Manager (Insight),

In terms of education any advanced course in statistics should work as a passport to Web Analyst job. He/She should be genuinely good at collating different industry information.
-Apurba Sen, Product Manager at India Ltd

A sense of abstract thinking and Art (seeing the whole picture at once) is important for Web Analytics work.

I think what a lot of companies are asking for has nothing much to do with what is needed to be truly effective. In order to be effective as a Web Analyst you don’t usually need a degree in statistics or be trained as a rocket scientist (yet that is what all the job descriptions I’ve seen ask for – like a big long laundry list) you need the ability to understand what some one needs to measure, what the goals are, and a technical mastery to the tools, the web analytics platforms being used. The least important thing is to know the tools beforehand – because anyone can learn them fairly quickly who is not brain dead (yet this is precisely what most interviewers ask for first – knowledge of the tools)

The most important thing, in any job, is being effective in the job you’ll be hired for –being trusted, that you can deliver what say you can deliver on…and a lot of that is based on trust, on conveying confidence, conveying authority. The technical part is more like the icing the cake, as far as I’m concerned. I go for rapport, gut feeling, intuitive knowledge and visualization of what my clients want and need, and even what they don’t know to ask for, but which they still need…I try to give them…and this is not just at IBM, but all my work is done like this.

You know you’re effective when you’re allowed to work on the “big problems” in your organization. It’s a feedback loop. You need to get trust of higher ups so they’ll let you get the relevant experience that actually makes you valuable in the marketplace (so that you can “move up”). The paradox is, believing in your self first, is necessary in order to get any kind of trust and buy in so that you’ll be allowed to work on the big stuff.

You also need people to like you – and for some people, that’s easier to achieve than others. I won’t say that people that are disliked are not effective – they can be also, but they’re probably miserable and less effective than if they were liked.

But none of these skills is actually what is asked of you in an interview – yet some interviewers will make note of them, nonetheless and the one’s that do are the one’s to work for.

-Marshall Sponder, Blogger at

More to come in part II.

You might also want to check my blog post titled "Starting a career in web analytics", that I wrote more than an year ago.

What skills do you think are important? Let me know and I will add them to my next post.

Monday, February 11, 2008

Behavioral Targeting Standards Consortium

Revenue Science is announcing a new initiative call Behavioral Targeting Standards Consortium (BTSC). The purpose of this consortium is to put some standards definitions and best practices for behavioral targeting. This effort hopes to the form an association of diverse key industry players who will help shape these standards.
According to a press release by Revenue Science
“As the behavioral targeting industry continues to grow, confusion surrounding the technology has also expanded along side it. How does behavioral targeting differ from audience retargeting? What constitutes an audience segment? How are segments constructed and where does all this data come from? How should behavioral targeting be measured? These are questions the industry is hearing more and more. Without standards around these issues, it can be difficult for advertisers to plan and execute behaviorally-targeted media buys or measure performance.“

As ad networks are spring up every day claiming to do Behavioral targeting I believe there is a need for such a consortium because all behavioral targeting companies are making up terms to confuse advertisers, publishers and consumers. Last year Revenue Science touted that are reaching over Billion behaviors a day. “Reaching Billion Behaviors” is something I called confusion because it is not clear how you define a behavior. In a blog post last year I showed how you a network can reach billion behaviors just by having one user view 30 pages on a site.

I am not too sure if a behavioral targeting network like Revenue Science is the right entity to lead this initiative. According to Media Post article, in 2004, Tacoda proposed a set of 22 standard behavioral targeting segments to simplify the media-buying process--ensuring that advertisers who bought "Auto Buyers" as a segment on one site would get an audience that exhibited the same quality of behaviors on another site. Competitors were concerned that the standards would favor Tacoda's operating model. Other companies have also attempted to put forth their own standards since then but still too much ambiguity in the market.

Last year Revenue Science released a site called, a site aimed at educating consumer, publishers and advertisers on what Behavioral Targeting is. The sites’ promise was to provide articles, events, and success stories from industry experts and eventually grow to include an interactive section where users can share their questions, ideas and experiences with behavioral targeting. It is almost one year and all I see is case studies and discussion from Revenue Science only and that too is very minimal. Maybe it is lack of coordination in the industry and this consortium will help or maybe it was just a short term marketing ploy.

Let’s hope this consortium does take off. If you are interested you can join the consortium at to join, and the consortium's first meeting will be held at OMMA Global Hollywood in March. Keep in mind Revenue Science will have you email (the thank you email come from marketing).

Sunday, February 10, 2008

Web Analytics Surveys

eMetrics Survey

Jim Sterne is conducting an eMetrics survey about web analytics usage and tools. The purpose of this survey is to find out which channels marketers and analysts are measuring, how they are using that data to optimize marketing and their main objectives for marketing optimization this year.

The survey is really short; it has only 9 questions and should not take more than 5 minutes.

Here is the link to take this survey

Bounce Rate Survey

Recently I conducted a survey to understand bounce rates of different kinds of sites. Currently I am in the process of analyzing the results. I hope to post the results within 2 weeks.

Do you have a survey idea? Send it to me.

Thursday, February 07, 2008

Online KPIs – Back to Basics

Those who have been doing web analytics for a while know how important it is to define proper online Key Performance Indicators (KPIs). But believe me, there are a lot of marketers who are confused about online KPIs, difference between KPIs and metrics and how to define them. So I am going back to basics with this post.

What are KPIs

Web analytics tools collect a lot of data and provide a lot of metrics and reports. In fact most of the web analytics tool vendors proudly talk about number of reports that can be created in their tool. These reports, metrics and data might look interesting but we all know interesting is not necessarily important. KPIs, on the other hand, are the important metrics; the metrics that provide a view into the health of the business and are tied to the business goals. They allow business owners to focus on the things that are important to drive their business. Key Performance Indicators tell a business owner whether he or she is meeting their business goals or not. Good KPIs provide context and hence are usually represented as ratio, percentage, indexes etc and not as raw numbers. KPIs drive actions within an organization.

KPIs are specific to a business role. So, not all people in the organization have the same KPIs though all the KPIs should ultimately be tied to overall business goals. The CEO has a set of KPIs, a merchandising manager has a set of KPIs and a marketing manager has yet another set of KPIs. However, all of the respective executives (departments) need to be defined keeping overall business goals and CEO’s KPIs in mind.

Another way to understand KPIs is that they are the metrics that make people freak out when they go in the reverse direction from the expected and call for immediate actions.

Since so much is riding on the KPIs, it is very critical that you pay due attention in defining your KPIs. Understand what business goals are and then think about what activities and/or user behaviors relate to your business goals. Put together a list of all the metrics that will measure those activities and/or user behaviors. Weed out the unimportant metrics, figure out what are important metrics and what are critical few (and hence KPIs) that have an impact on the business goals. Note: For your analysis you will need to look at more than your KPIs to provide you a bigger picture. Remember, all KPIs are metrics but not all metrics are KPIs

Characteristics of KPIs

Dennis Dennis R. Mortensen lists following 7 KPI characteristics on his blog “Visual Revenue”
  1. a KPI echoes organizational goals

  2. a KPI is decided by management

  3. a KPI provides context

  4. KPI creates meaning on all organizational levels

  5. a KPI is based on legitimate data6. a KPI is easy to understand

  6. a KPI leads to action!

Those are all great characteristics of KPIs. I however differ a little on point number 2. In my opinion great KPIs are those that are agreed upon by those it directly impact and will be taking actions so they are not just handed down by the upper management. And as I said above they should all be tied to overall business goals.

How many KPIs should you have?

I don’t think there is any rule but in my experience you should limit it to no more than 6.

Reporting on KPIs

KPIs should be presented in an easy to consume dashboard. Web Analytics tools have built in dashboards but most of them are limited in terms of the functionality and flexibility. My recommendation is to present KPIs in a separate dashboard that not only shows KPIs but also trending and brief analysis. Without trending and analysis the KPIs might not provide a complete picture. Excel, PowerPoint or third party dashboard tools work the best for reporting the KPIs. Since they are outside the web analytics tool they also allow you to integrate other data sources, as needed.

Books on Web KPIs

Eric Peterson has a great book on the subject, called The Big Book of KPIs

Monday, February 04, 2008

Bounce Rate Survey needs your input

Have you wondered how the bounce rate of your site compares to the others in the same industry? After I wrote the blog post about Bounce Rate I got several emails asking me if I had data or info on the bounce rate in their industry. Here is your chance to participate in an effort to collect bounce rate and conversion data. Please take few minutes to fill out a small survey.

I will publish the results on this blog and in Web Analytics Association’s newsletter.
Take Bounce Rate Survey and help me gather this information which will benefit all of us.

Note: Here are the results of Salary Survey that I conducted last year.

Note: I will be closing this survey on 5th, so please participate it in now.

Friday, February 01, 2008

Widget Ads and Offsite Analytics

Earlier this year in my Web Analytics and Behavioral Targeting Predictions for this year, I wrote
"Offsite Analytics – With proliferation of Widgets and Rich Media ads a need to measure, understand and optimize the user interaction with widgets and Rich Media ads will become important. Web Analytics will include offsite measurements i.e. user interactions with widgets and Rich Media Ads"

Yesterday I came across an article on iMedia Connection that validates my thinking. The article talk about an The online widget Ad by Paramount Vantage to promote “The Kite Runner” a movie This ad unit (widget) allows users to interact and make a ticket purchase for the movie without leaving the widget i.e without going to advertiser’s site. As you read the article by Media Post, you will see that they (agency/advertiser) are measuring the interaction rate (not sure what exactly is considered interaction maybe number of visitors/visit who clicked inside the widget/to number of visitors/visit who saw it) and the conversion rate. According to the article: "The first trial campaign saw interaction rates at 14 percent compared with other online campaigns where 0.5 percent is considered successful."
They also say that ticket sales ticked up but did not say how much.

As ad units like these become common there will be a need to measure more than just interactions and conversions. There will be a need to measure where in the whole conversion process visitors dropped off, what they searched for (cities, zip code, times etc), what other pieces on the widget distracted them, need for A/B and Multivariate testing etc.

I also believe that it will become common to use Behavioral Targeting in these widgets. Let’s say a widget like this was promoted by Fandango and they were promoting tickets for several movies. Knowing the user past behavior they could put the ticket sales for the movie genre that fits the users past behavior instead of any random movie.

What do you think? Will ad units like this become common or this is just a fad? What about behavioral targeting within these ad units? Yes, I know privacy is one of the issues but that can easily be worked out, if the use the opt-in model, as I said in my predictions about Behavioral Targeting.