Monday, December 31, 2007

Unnecessary Outcry over Adobe CS3 and Omniture Tracking

On Dec 26th a blog post on Uneasysilence.comshowed that when a user launches CS3 suite adobe calls a server on Omniture. According to the article
When you launch a CS3 application the application pings out to what looks like an IP address - and internal IP address: 192.168.112.2O7.

This created a lot of paranoia among Adobe users and led to a lot of blog posts and comments on these blog posts (Also see http://valleywag.com/338011/wear-tinfoil-hats-when-using-adobe-products). Everybody started talking about how evil adobe is and so on. It appears that we, Web Analytics community and online marketers, have a huge task of educating users about Web analytics tracking and quelling this kind of paranoia.

Finally one product manager from Adobe stepped up to clear all the confusion.

Adobe Product manager wrote a reply on Adbobe Blog. Here is what he wrote:
According to Doug Miller from the Adobe.com team, "Omniture is Adobe's web analytic vendor for Adobe.com. There are only 3 places we track things via Omniture anywhere in or around our products.":
  • The welcome screens (these things) in some Adobe apps include a Flash SWF file that loads current news, special offers, etc. These requests hit Adobe.com servers and are logged, like regular browser-based traffic, by Omniture.

  • Adobe Bridge embeds both the Opera browser and the Flash Player, both of which can be used to load Adobe-hosted content. These requests are also logged.

  • Adobe apps can call various online resources (online help, user forums, etc.), and those requests are logged. [Update: To clarify, those contacts are made only if the user requests them--e.g. by choosing Help->Adobe Exchange.]


This, as far as I've been able to discover, is the extent of the nefarious "spying." If I learn anything else when more people get back on email, I'll update this post.

Let me start by saying that the kind of tracking Adobe appears to be doing is pretty harmless to you end users. Now let me ask a question to all these people, who became so paranoid about Adobe and Omniture Tracking. “Do you know that you are being tracked at a lot of places?” I am sure you have done one or more of the following

Connected to the internet – Do you know that ISP track of what you do online? You should read my blog post titled “ISP Based Behavioral Targeting.
Visited any site on the internet? - A lot of sites (and in fact they all should) track user behavior to create a better experience for users and to help them in their business goals.
Installed a toolbar – Do you know that their activity is being tracked by toolbars you install?
Used any social networking site where they volunteered all sorts of information.
Used a credit card. – Yes they have whole history of what you bought, when and where.
Bought a product on any major retail chain, used a credit card or a club card. – They keep track of what you buy, when and where too.

Since you are tracked everywhere, why is there a paranoia about being tracked by Adobe? Actually the kind of tracking Adobe is doing is not even close to the information you are giving away via other activities (some of them mentioned above). The kind of tracking adobe is doing is to understand the usage of their sites and provide a better experience for the users. The way I understand, Adobe is looking at user behavior at an aggregate level and not at an individual level, most of the companies doing web analytics do not look at individual user behavior. Most of the web analytics tools, like Omniture, use an anonymous cookie to track user behavior. This anonymous tracking usually looks at aggregated data for entire user base (or few segments) instead of an individual and hence do not invade your privacy.
So I request you to stop this paranoia about tracking by web analytics tools. They are helping you to have a better experience on the web.

Note: I am not associated with Adobe in any way. I have never worked for them in any capacity and do not know anybody personally at Adobe.

Comments?

Friday, December 28, 2007

Web Analytics Jobs Trends – 2007

Web Analytics is one of the hottest career fields these days. Organizations are realizing that web analytics can no longer be a part time work and requires full time dedicated staff thus pushing the demand for web analysts.

This year, since making my Web Analytics jobs predictions in Januray, I reported on open job positions every month till August of this year. After August I stopped reporting on the jobs because I did not see any major changes and there was nothing exciting to report. Since the year is coming to an end, I thought I will close out the year with another report to show where we are and what to expect in 2008.

Before we look at the numbers please note that the Dec numbers are taken today i.e. 28th while the rest of the number reported were taken on the 1st of the month. As you know I use two job aggregator sites Indeed.com and SimplyHired.com for the data. Both of these sites collect open job positions from individual company sites and from job boards such as HotJobs.com. SimplyHired.com also provides job boards called job-a-matic, like the Job Board I have on my blog. SimplyHired.com job board allows individual bloggers or site owners to quickly create a job board specific to their site’s content.
Note: Those who are curious to know what to expect in terms of salary, I will be posting the results of Job Survey on 1st of Jan 2008.

Let’s take a look at the numbers.

When I first reported the open Jobs in January, there were 1024 open positions that had the word “Web Analytics” in them. In the very first month the open jobs were up to 1711, a jump of 67%. It is quite possible that due to holidays not many positions were advertised and when people were back from holidays they started advertising open positions and hence we saw this big jump. Well this year, as of Dec 28th, there are 2068 open positions, that is 102% increase from January this year. I don’t see much change happening in next 3-4 days so January 1st numbers will be about at the same level.
Looking at the trend in 2007, we can expect a big increase in open positions in February and the whole year



Note: Month in the above graph represents the month when the data was gathered. A lot of job positions are never listed on any job board or company sites. These jobs are filled by networking and referrals. So key to finding a job is increasing your network. Let people know that you are interested in Web Analytics.

If you want to start a career in web analytics and don't know where to start, check out my article starting a career in Web Analytics and my Web Analyst interview series to see how others got started in web analytics.

Which tool experience is in demand?

Omniture remains the most sought after tool experience, followed by WebTrends.
In this month, Google Analytics displaced Coremetrics from 3rd spot and is behind Omniture and WebTrends.

In 2008 I expect Google Analytics to be in hot demand. As Google releases new features and opens up the APIs it will become more complicated to implement and use.


Stay tuned for the Salary Survey results to be published on January 1st.

Thursday, December 20, 2007

FTC Proposes Behavioral Advertising Privacy Principles

To address consumer privacy concerns associated with Behavioral Targeting FTC proposed privacy principals.

The purpose of this proposal is to encourage more meaningful and enforceable self-regulation to address the privacy concerns raised with respect to behavioral advertising. In developing the principles, FTC staff was mindful of the need to maintain vigorous competition in online advertising as well as the importance of accommodating the wide variety of business models that exist in this area,” according to its proposal “Behavioral Advertising: Moving the Discussion Forward to Possible Self-Regulatory Principles. The proposal states that behavioral advertising provides benefits to consumers in the form of free content and personalized advertising but notes that this practice is largely invisible and unknown to consumers.

Below are the principal they proposed:

  • Every Web site where data is collected for behavioral advertising should provide a clear, consumer-friendly, and prominent statement that data is being collected to provide ads targeted to the consumer and give consumers the ability to choose whether or not to have their information collected for such purpose.

    Sooner or later that is going to be almost every site that you encounter. Since there is no common definition of Behavioral Targeting any targeting (since it will uses onsite behavior, geo or any data collected from users ) can be considered behavioral targeting.

    Give consumers the ability to choose whether or not to have their information collected for such purpose - it is not clear if they mean opt-in or opt-out.

    I am in favor of providing an opt-in instead of opt-out. In my post on Google and Doubleclick privacy concerns, I wrote:
    I believe that if consumers are provided proper education (I will write about consumer benefits in one of my future posts) than they can infect benefit from Behavioral Targeting. It will be a win-win situation for all the parties involved. Proper education and disclosures by advertisers, publishers and networks will ease the concerns regarding Behavioral Targeting. Consumers have the right to opt out of Behavioral Targeting but what is lacking is proper education on how to do so. The networks currently opt-in users by default; however, in my opinion the proper process should be opt-out by default and opt-in if user chooses to opt-in, just like we do for emails and newsletters. This process will move the burden from users to the advertisers, publishers and networks.


  • Any company that collects or stores consumer data for behavioral advertising should provide reasonable security for that data and should retain data only as long as is necessary to fulfill a legitimate business or law enforcement need.

    “Reasonable” is very vague since every company can define it’s own explanation of reasonable.


  • Companies should obtain affirmative express consent from affected consumers before using data in a manner materially different from promises the company made when it collected the data.

    This is to safeguard against changing privacy policies. Since almost all the privacy policies have a clause which says something like “We reserve the right to change this privacy policy. New privacy will be posted on this page”. It is hard for consumers to keep track of what has changed since they agreed to the privacy policy.

  • To address the concern that sensitive data – medical information or children’s activities online, for example – may be used in behavioral advertising, FTC staff proposes:

    • Companies should only collect sensitive data for behavioral advertising if they obtain affirmative express consent from the consumer to receive such advertising.

    • FTC staff also seeks comment on what constitutes “sensitive data” and whether the use of sensitive data should be prohibited, rather than subject to consumer choice.


    My opinion: Sensitive data should be prohibited. However it won’t be easy to define what constitutes sensitive data especially when it has to apply to various countries and cultures. Sensitive information in one country might not be sensitive in another country or culture.


Comments? Questions?

Wednesday, December 19, 2007

NebuAd’s response to my blog post Part II

In response to my post on ISP based Behavioral Targeting, I got 2 response from NebuAd, one of them was posted at http://webanalysis.blogspot.com/2007/12/nebuads-response-to-my-blog-postin.html. The latest response is for one of the issue that was not answered in previous response:

The ISPs are completely passive in NebuAd’s model. In addition, there is no pop under, and the ads NebuAd sells to do not take over the publisher’s inventory. More specifically, NebuAd's technology does not include any type of overlays that affect publisher content or ad inventory without their knowledge.

Consumers do not see any more ads than they would otherwise see, and they are standard ad types, such as banners ads displayed only where you would expect to see them. The use of pop-ups and pop unders remains a publisher decision. With NebuAd’s solution, the ISP role is completely passive. They merely allow NebuAd’s equipment to reside on their network and are not involved in the advertising process. No data is shared between the ISP’s data systems and NebuAd’s data systems

NebuAd purchases online display inventory from publishers, mainly through leading ad networks. Targeted advertising that the consumer sees is placed there on the ad inventory that the publishers regularly sell to ad networks. These ad networks work with NebuAd to identify and deliver the right ad for the right user, allowing for the publisher to get a higher price from the advertisers who want the targeting. NebuAd’s revenue is generated by advertising sales during this process – the ISPs do not sell any of the advertising.

Monday, December 17, 2007

NebuAd’s response to my blog post

In response to my post on ISP based Behavioral Targeting, I got the following response from NebuAd:

Below are a couple of quick points from NebuAd’s CEO Bob Dykes to explain and clarify some of the information.

There is no information shard between NebuAd and the ISPs - the only involvement between the two is the agreement that lets NebuAd place the appliance in the ISPs network. To further ensure privacy, NebuAd does not collect the websites visited and map those back to the specific user. Instead it converts, via an appliance located in the ISPs network, the key user identifiers, such as IP addresses, to a one way random number so that the central servers see this and not the original identifier.

NebuAd works by listing categories (e.g. “Cars – SUV – Lexus”) and noting if random number goes to a site, or perform a search, that is related to the category. If yes, then it notes that interest mapped to the random number, but do not map the URL’s visited, just the interest. This is why, since it doesn’t even have the info on sites visited, there's no mechanism to map the random number to specific URLs
Since NebuAd and the partner ISPs do not exchange data, the ISPs do not see the categories each random number visits, and NebuAd does not receive specific customer information, so there is no way for either NebuAd or the ISPs to match specific customer information with even the categories of information associated with the randomized numbers. NebuAd also does not retain the raw data mapped back to the anonymous user profiles.


They have not yet provided any response to the following point that I made in my previous post:

It is also not clear to me if the ISPs will work with individual publishers or networks and provide behavioral data to power their ads on publishers inventory or if they will override publishers inventory with their own ads (which will probably cause sudden death of ISP based targeting) or if they will do popups (pop under) creating new inventory. NebuAd does however have a service for publishers where publishers can use their services on their own inventory; however I am not clear how ISPs plan to use it.

Sunday, December 09, 2007

ISP based Behavioral Targeting

In an articles titled Watching What You See on the Web Wall Street Journal talks about ISP (Internet Service Provider) based behavioral targeting.

ISP based behavioral targeting idea has been kicked around for some time and NebuAd is one of the first company that made a product know as “deep-packet inspection boxes” for ISP to track user behavior online and then serve ads based on these behaviors.
This kind of targeting enables ISP’s to be a player in growing behavioral targeting market and generate a new stream of revenue.

This kind of technology is beyond simply using anonymous tracking. ISP do have a lot more information than just the browsing behavior. They have name, location, age, social security number (SSN). They know what time users login to their machine, when is the internet being used, what kind of sites are visited at what times, which sites provided information before a user made a purchase etc etc. This is far more information than companies like Revenue Science or Tacoda has and obviously can provide better targeting than Revenue Science or Tacoda can do.
However this also raises far more privacy concerns than companies like Revenue Science and Tacoda raise.

According to the article
The technology does raise privacy issues. The Internet-service providers often know other information about consumers, such as their names, locations and age and income ranges, which can be very valuable to potential advertisers, especially when combined with Web browsing habits. "Some of these [Internet equipment] guys are traveling in dangerous territory," says Emily Riley, an advertising analyst with Jupiter Research. "Should one company have all of that data in one place? It's a little troubling."


Other than user privacy there is another huge issue that this article did not talk about. In a network like Revenue Science or Tacoda publishers and advertisers (data providers) have to opt-in to participate. If a publisher/advertiser does not want to enable advertisers to use their data then they simply do not participate in the network. Advertisers can also choose to just use their site’s data to be used to only power their own advertisements. E.g Delta airlines can choose to participate in a retargeting campaign on a network like Revenue Science. They can retarget all the users who viewed fares to a particular destination but left the site without buying the ticket. To do so they will allow the network to collect information on all those users whom they want to target and then only allow the network to use those behaviors (users) to target their ads only. Alaska airlines cannot use Revenue Science and target their ads based on the behaviors on Delta airlines network. This is an explicit agreement between the publisher/advertiser and network.
However in case of ISP based targeting; data providers (publishers, advertisers and other sites) don’t have to opt in. They are opted in by default. Using the example above, a user’s behavior on Delta airlines site (and also information about who clicked on Delta’s ads across internet) is captured without Delta Airlines explicit approval. Now, ISP’s can use that information to power Alaska airlines advertisement and drive all those users, who could have purchased their tickets from Delta, to Alaska airlines. I am sure Delta won’t be happy about it. This applies to every single site on internet, they do not have an option their data will be used and in most cases to power competitors ads, this is a huge deal. I think it is, what do you think? I am sure there will be advertiser backlash too with this kind of technology.

It is also not clear to me if the ISPs will work with individual publishers or networks and provide behavioral data to power their ads on publishers inventory or if they will override publishers inventory with their own ads (which will probably cause sudden death of ISP based targeting) or if they will do popups (pop under) creating new inventory. NebuAd does however have a service for publishers where publishers can use their services on their own inventory, however I am not clear how ISPs plan to use it.

As I predicted earlier this year Behavioral Targeting has become a very common term among marketers. To cash in on this phenomenon a lot of new technologies and companies are springing up, I expect this trend to continue in 2008, we will see more innovation in coming month. Mobile and TV behavioral targeting is next in line too.

Questions/Comments?

Sunday, December 02, 2007

Behavioral Targeting and Affiliate Marketing

Behavioral Targeting (BT) provides a great opportunity for Affiliate Marketers and Networks to reach right customer and increase the conversion.

What is Affiliate Marketing?

Affiliate Marketing is way for advertisers to reach potential customers and only pay when a visitor takes some predefined action. Predefined actions range from a sale to registration. Recently Pay Per Click has also been added to affiliate programs. Generally, the website (publisher) places an advertisements on their site and when a predefined action results by a visitors referred (clicked) by that ad, the advertiser pays the publisher a % or a fixed amount. I won’t go into further details of affiliate marketing but you get the idea.

As you know now that one of the major benefits of Affiliate programs is that an advertiser gets an inventory on publisher’s site without paying anything. Advertiser only pays when a visitor becomes a lead or a customer. This is like sales people on commission only program.

However there is even bigger benefit than just getting free inventory. The ability to put an ad serving (and tracking) code on these affiliate sites (publishers). Just think about how many sites Amazon, eBay, commission junction, linkshare etc. have their codes on. Amazon was the one of the early advertiser who realized the power of Affiliate programs. Google, not matter how you look at it is also a type of affiliate program, in most case advertisers get free inventory and only pay when a click happens (and now Google has pay-per-action program which is true affiliate program).

The widespread tracking code on the internet provides a huge opportunity for affiliate networks and advertisers (like eBay, Amazon) to engage in behavioral targeting and reach potential customer with the right message at the right time.
Let’s take a look at a scenario to further understand how this can work.
A user visits eBay and looks at certain products and bids on a camera but lost the bid. Next day eBay gets another seller who is selling the same camera. How can eBay reach the user who lost the bid? Well one way is they can send her an email (She might not be actively checking my email). Second is tell the about that camera when she come back to eBay (that might be too late). Third is they can use their affiliate site to put that camera offer right in front of her if she happens to visit any of the affiliate site (and there are a lot of them). She has shown interest in the camera so why not use your affiliate code to put the right offer in front of her. The code is right there make use of it. Some might argue this is crossing the privacy line but that is whole another issue to discuss.
Amazon already does similar targeting by its Omakase links but it is not true behavioral targeting. They use the keywords or browsing history which goes back since they day they were born. I see offers from Amazon that are not even relevant to me anymore, they were at one time but Amazon still thinks I am interested even though I have not clicked on those products in years.

Affiliate networks like Commission Junction and Link Share are in great position (eBay uses Commission Junction). They know what kind of sites a visitor (or customer) visits, what kind of products/offers he/she is interested in by looking at their click or purchase behavior. Can’t they just automatically put offer s/products in front of customers?

The way most of the affiliate networks work today is not very efficient for any of the parties involved. They need to step it up a bit. Here are my thoughts on all parties involved how Behavioral Targeting (BT) can help them.

Publishers: A publisher has to decide what advertisers they want to sign up for, this is their guess on what will work. I have spent months trying to figure out what will work for my customers. Also then the link generation process is so difficult. Say a publishers sign up as an affiliate for Resaturant.com (check out SeattleIndian.com, it has restaurant.com banner), now the publisher has to go find the products that are relevant to their customers if they want to target with right banners. In this case maybe it is restaurants in Seattle. It is such a pain that majority of publishers don’t want to deal with it and just put a generic banner. Won’t it be nice if CJ or Restaurant.com in this case was able to put the right offer based on Geography and other browsing behaviors on either CJ network or Restaurant.com site itself?

Customers (visitors): Customers (Visitors) don’t have time, they will only click on ad which will be relevant to them. A generic message from Resturant.com or eBay or Amazon is not enough. Show them a message which will make them click right away. If a customer has already been to Restaurant.com and looked for Indian Restaurants in Seattle and now when the customer is in Seattle (Geo location) surfing on this affiliate site, why show her a generic message? Why not show her offers for Indian Restaurants?

Advertisers: Even if advertisers are getting free inventory now, it is not going to last. Inventory is limited, publishers are going to replace their banner with something that will make them money (AdSense has replaced several affiliates). So grab this opportunity. Put relevant message, gone are the days when generic message would have worked (did they ever worked though). Use the browsing behavior on your site (or a behavioral ad network, your affiliate network will have to offer behavioral targeting soon) to target the right messages.

Affiliate Networks: Why are they still operating using the old model where they make the publisher s guess what products or offers will work for them (Google is following this model too with their pay-per-action model). Why can’t you let publisher put some generic code on their site and then serve ads based on user behaviors? I have seen a lot ads sitting on a publisher site not generating any sales, few clicks, who is benefitting from all this? Not the affiliate network. Not the publisher. Advertiser? May be. Visitor? May be. Now if you had placed a right message (determined from user past behavior) then chances are that all of the parties involved will benefit. It is worth trying. Test it. Use the data you collect, the data from advertisers' site, the data from publishers' site. It is not going to be easy but it is doable. Acquire a behavioral targeting company, use their technology and network too.

Affiliate marketers and networks have a huge opportunity to cash in using Behavioral Targeting. They just have to move fast.

Comments, Questions?

Here are some more posts on Behavioral Targeting.

If you like this post, you might want to subscribe to my blog feed. Click here to subscribe to Web Analysis, Behavioral Targeting and Online Advertising