Re: facebook: referring domain instances VS tracking code visitors
- First question:
To answer directly, let's consider an example. You have said that the variable expires after seven days. Imagine that the user clicks through to your site from Facebook on Monday, and acquires a tracking code in the variable in question here. Then, the user comes back to your site on Wednesday and Friday. Because the Facebook-related tracking code hasn't expired prior to the second and third visits, you will see a total of three visits tied to that tracking code for this user. In other words, when you asked, "Do instances from a tracking code count all visits from that user over the full 7 days, from any site, or something?" you were correct.
A simple way around this that I've seen many SiteCatalyst users do is to take a separate Custom Conversion (eVar) variable and put the same tracking codes into it, but set the expiration differently. For example, if eVar1 is currently storing the tracking code and is set to expire after seven days, you could use eVar2 to store the same tracking code, but set it to expire at the end of the visit. This can help you understand both long-term and short-term traffic and conversion from campaigns on Facebook and elsewhere.
The Visits and Visitors metrics are joined at the hip. It's actually fairly difficult to make implementation errors that affect these metrics beyond the explanations that you've read (users not accepting cookies, exacerbated by IP pooling situations). The one that comes to mind is s.cookieDomainPeriods. Does your domain have three periods in it (e.g., .yoursite.co.jp), or two? If it has two, then you can ignore this whole answer. If it has three, make sure that you've got s.cookieDomainPeriods set in your s_code.js file, and that it is set to "3". If this is not the case, we won't be able to set cookies and you'll end up with a somewhat higher difference between Visits and Visitors. I huddled with a colleague, and this was the only implementation-related mistake that we've seen affecting the relationship between Visits and Visitors.
I hope this helps. . . let me know if you need any additional information.
Omniture, an Adobe company
--- In email@example.com, "mstjohninjapan" <mstjohn@...> wrote:
> First Question:
> Trying to measure the business impact of facebook to our ecommerce site in April. Getting two different stories from Omniture's referring domains report and tracking code report. I understand revenue will be different, as revenue counted in the referring domains report is "direct" only, whereas with a tracking code it counts revenue if a purchase is made within 7 days. (CID set to expire after 7 days and last CID clicked). But traffic I don't get. If I add up all the 88 tracking codes from facebook, I get 44k "instances". If I look at visitors or visits, it shows around 22k for both (actual numbers multiplied by a factor, ratio same). A huge difference. Do instances from a tracking code count all visits from that user over the full 7 days, from any site, or something? I need concrete answer to the question, "how many people did facebook refer to our site in April?"
> Second Question:
> My visitors metric is always slightly higher than visits. I have read "why" this happens, but it still doesn't sit well with me. Im not very technical, but I have a feeling it has to do with something during implementation. Just wondering if anyone else has "fixed" this problem.
> I would love to be able to say "last month, on average, each visitor had 2.6 visits"...not 0.9