On The Trail

http://www.coffeesunanalytics.com/web-analytics/37/ — Interesting stuff. Particularly the observation that dial up users have a lower average order size. I guess I’d expect that…

With all this talk of cookies wouldn’t it be nice to actually see one?
It’s pretty easy to see the cookies that are set on your machine. If you’d like to see them (and you use Windows XP or 2000 and Internet Explorer) go to C:\documents and settings\YOURNAME\Cookies (where YOURNAME is your login name). They’re stored there as text files and you can open them up by double clicking on them. The information contained inside is usually pretty boring, but really useful for both you and the site that sets them.

If you use Firefox it’s a little easier to see your cookies. From the Tools menu choose “Options”. Pick the “Privacy” icon on the left and expand the cookies section by clicking on the little plus sign. You can see all your cookies, their values and expiration dates by clicking the “View Cookies” button.

So we’ve established that cookies are important and many people delete them. What should an online marketer do about it? Well, first lets talk a little about kinds of cookies. I’ll try to be non-technical here:

1. First Party Cookies. Cookies can only be “seen” by the site that sets them. First party cookies are set by the site that you see in the address bar of your browser.
2. Third Party Cookies. Often when you visit a site the page will display images that are hosted by a different (aka third party) site. These images can set cookies just like a regular web page and they are known as “Third Party Cookies”. This is the type of cookies set by many ASP style web analytics solutions and banner and affiliate networks. These services are quickly switching to first party cookies, but many still use third party cookies.
3. Persistent and Session Cookies. Persist ant cookies are saved on a client’s machine until the expiration date specified by the site that set the cookie this can be years or decades in the future. Session cookies are erased when you close your browser. First and Third party cookies can be either persist ant or session cookies.

Modern browsers (such as Internet Explorer 6 and Firefox) have settings that can discriminate against various kinds of cookies. Many websites will not work correctly without some sort of cookies enabled, so all but the most paranoid people will accept first party session cookies. First party persist ant cookies can be “demoted” to session cookies by the browser, although the perceived value of having a site remember you when you return may keep most people from doing this. Third party cookies are in the worst shape since the user sees no value in them worth the perceived privacy cost.

So, what to do:

1.
Use a login to identify users if at all possible. Logins have the advantage of identifying a person instead of a computer. So someone who visits your site from home and work will be accurately counted. Unfortunately, logins are not possible for most sites.
2.
Use first party cookies. A higher proportion of people will accept these and they are much less likely to be deleted by spyware or adware programs.
3.
Make your site offer rewards for allowing a cookie. An excellent example of this is Amazon.com. For many people (including me) Amazon’s personalization is an excellent motivation to keep their cookie around. Of course, even though personalization usually gives the visitor a better experience the downside is that it can get expensive.
4.
Nothing. Web Analytics does not produce totally accurate numbers anyway. Trends in these numbers are the really useful and important information. As long as the proportion of users who delete cookies stays relatively constant the trends will probably mirror reality. (Best line at a vendor seminar: A Web Trends Representative after telling us that log file analysis tools like Web Trends are commonly off by 30%: “That’s why we call it Web Trends”)

Some people have proposed using other visitor identification methods like storing information using the Flash plugin or going back to a IP address user agent combination. In my opinion, the former is underhanded and borderline un-ethical, while the latter is much less accurate than using cookies (see the RedEye report). Realistically, a combination of #2 and #4 will be the prescription for most sites. Being aware of the problem, particularly when interpreting deferred conversions and lifetime value numbers is the first step.

As mentioned in the previous post two reports (1, 2) have found that about 1/2 of all users delete their cookies at least once a month. Why does this matter to an online marketer? Well, cookies are used in a variety of ways to measure the performance of online campaigns. If you’re trying to measure conversions for a particular campaign the user cookie allows you to tie together the initial clickthrough from the campaign with the conversion event (such as a purchase or an email newsletter signup or a white paper download). If the visitor doesn’t accept your cookie they’ll be counted as a clickthrough from that campaign, but they won’t be counted as a conversion, so your conversion rate will be artificially low.

Measuring deferred conversions (where the conversion occurs days or weeks after the initial clickthrough) is particularly hard if many of your visitors are deleting their cookies on a weekly or monthly basis. This number will be depressed even more than the immediate conversion rate since people who delete their cookies between clickthrough and conversion will not be counted. The “Lifetime Value” of a visitor is also affected since chances are good that their lifetime is less than a month.

Non-analytical marketing tools are also impacted by cookie deletion. Targeted content techniques often use cookies to determine what the user has viewed in the past. Most severely affected are banner ad and affiliate networks. They track people across multiple sites and set cookies from well known domains. Ad blocking and spyware blocking software target these cookies specifically. I suspect that the deletion rates for these cookies are higher than average.

There are the problems, what are the solutions? Tune in tomorrow for more…

Cookies are the online marketer’s best friend. Basically, they are a tiny text file that a website can write to your hard drive. Although it sounds vaguely sinister, it really isn’t. They are simply used to recognize a user as they travel from page to page. For example, when you log into Amazon they will set a cookie that tells them who you are. When you move around the site your browser will send back this cookie to Amazon’s web server each time you ask to see a new page — that is how the Amazon knows to put your gold box up in the top right corner. All of Amazon’s cool personalization features are enabled by the cookie that Amazon places on your computer. In this case, cookies are good for the site marketer and the site visitor.

Cookies can also be used to keep track what pages you visit on the site. There is little percieved value in this use of cookies by most web users, but it is the standard way that web analytics programs identify visitors and repeat visitors to your site. And here lies a serious problem facing web analytics today. Since users do not see the value of cookies a large proportion of them delete them with regularity. A recent Jupiter report (payment required) and another older report commissioned by RedEye in the UK have cast doubt on the reliability of cookies for web tracking. Their basic findings are remarkably similar: approximately 1/2 of all web users delete their cookies at least once a month.

What does this mean for web analytics? Tune in tomorrow for more.