The Wall Street Journal did a whole series on “What they know”. Online marketers - especially when it comes to ad networks, web analytics and behavioral targeting – are depicted as evil spies and Cookie Monsters. Call it sensationalist, biased or even lies and fallacies; it depicts what is generally perceived outside of our little web analytics industry. If we listen to the WSJ, even underlying concepts of Web 2.0 such as embedding widgets on a website is evil. Their analysis revealed content (and most likely, cookies) being served by a 3rd party… Bad, bad, bad! It goes beyond web analytics, but when threatened and called upon, people react and as is often the case, external events forces us, web analysts and the WAA, to act.
A fundamental misconception of web analytics
Over the year I have taught over 700 people on the fundamentals of web analytics trough courses at UBC, ULaval and my workshops. Oftentimes, a fundamental misconception about web analytics takes the form of "With the help of Google Analytics (or whichever tool), I can find dimensions such as number of visitors from a city, the source and frequency of their visit, recency of visits, keywords used, etc., however, how can I find "who these visitors are?" How can I find their contact details so as to start a personalized initiative?"First, we need to understand the fundamental difference between Web Analytics and Customer Relationship Management (CRM).
Web Analytics
Web Analytics is about a “statistical population”, "segments", "margin of error", "outliers", etc. The goal is NOT and should NEVER be to identify unique people, but to look a trends and patterns of online behavior shared by different groups of visitors. Web analytics pertains to the realm of inbound and onsite marketing and customer-facing optimization: campaigns, web sites, social media, etc. Period.We are not in the business of tracking users’ journey on the Interweb.
Behavioral targeting
Behavioral targeting doesn’t necessarily look at unique users; it looks at pattern of similar user behavior to determine what worked and what didn’t. However, admittedly, there are more advanced solutions where specific individual behavior is analyzed and the persuasion scenario tailored accordingly.This can be done without ever using any PII.
Customer Relationship Management
CRM is about one-to-one customer relationship, keeping track of “touch points” for each individual person, regardless of the communication channel they use – phone, email, web, etc. At its core, the goal is to improve the relationship with prospects, clients & customers by optimizing processes around all communication channels, all touch points, at all time.We are at the opposite end of the spectrum: Personally Identifiable Information (PII) such as name, phone, email and even more are critical to CRM.
Ethical & legal aspect
It is unethical and illegal in most cases to collect uniquely identifiable information such as contact details (phone, name, SIN#, complete address, etc.) without express user consent. Although it is technically possible, especially with advanced solutions, to collect custom data such as PII, contact details should be collected within corporate systems and used within the limits of the applicable laws (US, Canada and the European Union have laws & guidelines regarding when, how, for which reason and what you do with the collected personal information – see Privacy and PIPEDA for examples).Offering a customized experience on Amazon because it recognizes you is CRM, not web analytics. Facebook happily using all the data you give it, reselling it or making whatever they want with it is CRM, not web analytics – risky business, but done with user consen – well… kind of. Services using Flash Cookies, be it ad networks, web analytics tools or web sites themselves storing data without user consent and rebuilding it even after the user consciously deleted it is illegal in my book… and it seems I’m not alone thinking that.
My take
Eric Peterson drafted a web analytics code of ethic and I’m glad he accepted to publish it under the umbrella of the Web Analytics Association instead of on his own. After calling on the WAA to be the voice of the industry, it would have been awkward to do otherwise! Lots of people talked about it on Twitter, commented on the blog, bloggers posted their own point of view and some vendors, like Webtrends, publicly took a stance (tip of the hat to them!). On my end, I updated my privacy policy to reflect the proposed code of ethic.Certainly all good things. However, the WAA as an organization has about 1500 members, the whole web analytics engaged practitioners and consultants’ base represents about 5000 people worldwide. Let’s be realist… The WAA is a small fish compared to other, well established organizations such as the IAB, and honestly, pretty much insignificant when faced with structured privacy lobbying groups. It seems web analytics vendors are still more interesting in competing, purchasing and killing one another than joining a coalition and lobby for the future of the industry.
Other associations are already ahead, like the IAB Privacy self-regulation compliance icon which is being developed in collaboration with the Future of Privacy Forum. The WAA simply doesn’t have the resources and lobbying power. Considering Microsoft, Adobe and Yahoo are already members of the Future of Privacy Forum, I think the WAA should simply rally the IAB and the FPF.
3 comments:
Stephane, Thank you very much for posting this article. As a consultant in the field of analytics and behavioral targeting, I firmly believe that it is very important that we defend ourselves against sensationalism (WSJ article).
Having a clear policy establishes a line in the sand that we can all point to when explaining and defining our industry.
Dan Piche
Stéphane - I am curious where re-marketing falls into all of this. For instance abandon cart emails. Clearly you have to ID the person who abandoned the cart to send them an abandoned cart email. Or, perhaps closer to web analytics, re-targeting via email based on what the user viewed on the site. Again, you need to tie the activity (data provided by a WA solution) to the individual user.
In that case while you are sending out information based on a segment (people who viewed) the data is still collected at the individual level.
If you call this CRM and not WA then you are left with the issue that to do the CRM you need the data collected by a WA solution (which we as web analysts would be instrumental in setting up the tracking)...so then were are you?
Would love your thoughts on those points! Great post!
DtTall: I would put it simply: re-marketing without user consent is not appropriate. Period. If I have created an account and abandoned my cart, then I have consciously entered a business relationship (by creating an account) and it's fine to re-market. However, say I'm filling an insurance quote: it will ask for a bunch of PII - one of them might be my email. The user should not be re-marketed unless it was specifically mentioned something along the lines of "you can continue later" and ideally, even ask for "do you want your quote to be stored for later review".
I might be oversimplifying it and every business will find good reasons to keep, store and reuse data... I think as an analyst we should guide them and tell what appears to be acceptable, ethical and legal (or not).
We need to make a distinction between the technology and the good practices of web analytics. Is your WA tool the best one to store PII? Certainly not! At best, your WA tool will store the key to a back-end system holding the PII.
Collecting data at the anonymized user-level (i.e. a random user id, first party, persistent cookie) doesn't mean storing PII. If the person enters a process where an ID can be generated and tied to back-end data, WA will help you segment and extract those keys and I think that's fine - depending on the business use you make out of this joint data!
I don't believe in the "web analyst" role anymore - I believe in the "analyst" role. And in this role you might have to work with WA data as well as CRM data - and you know what - it's perfectly fine.
I'm sure that would make for a great round-table discussion at eMetrics or over a bear during a 5@7! :)
Stéphane
Post a Comment