Any experts in site analytics here?

Need some insight into campaign analysis

[wolfgang]

If you have real world experience in using HBX (webside story) even better, but as long as you've managed a Google campaign and tracked using analytics software I'd love to get some advice and insight from you.

Initially I'd like to start by gauging whether our ratio of New Respondents and Repeat Responses are normal. If we had a total of 515 new respondents and 463 repeat responses for a total of 978 responses and we paid for 918, clearly a portion of the 515 click on our campaign more than once. What is normal for a campaign, given that there is variation from industry to industry? THanks.

[corbing]

Averages against other websites in analytics are tough. Everyone is different, and in this case, every keyword/campaign is going to be different. Instead of trying to figure out what is "normal" for everyone else, I would try to figure out what is normal for you and your campaigns and then figure out a way to improve your metrics so that the old normal is now underperforming.

Take a look at the people who clicked through more than once. What were the different search terms they used to get through? They probably didn't find what they were looking for the first time and went back to refine their search again only to find your adword again. Or maybe they searched for the same keyword both time? Perhaps comparison shopping? Use that information to evaluate and improve your landing pages to make sure they accomplish your goals. There is a usually a wealth of information is seeing two search terms from the same user that will help you improve your pages for the next customer.

[wolfgang]

How do you identify people who clicked through more than once and tie them to a keyword? I'm not sure we have that level of granularity. We just implemented HBX in mid-March and I didn't get campaigns set up until late March.

Also, is it reasonable to assume that if half of our first-time keyword ad respondents got to the site, went back to Google to search some more, found more of our ads and clicked at least once all within a 30 minute time-frame, thus registering as only one visitor but equaling two or more paid clicks?

[eWhisper]

HBX currently doesn't support keyword granularity for PPC campaigns. This should be rectified in Q2 with the release of 'bid', a PPC plugin for HBX.

They support campaign level information, and you can assign multiple conversion types per campaign. Therefore it is easy to compare an email campaign vs AdWords vs Overture vs each banner ad to determine advertising program effectiveness. However, you have to do quite a bit of digging to really get into keyword conversion granularity.

HBX is generally used as a JAVA based analytics program.
Here's a good thread about server side vs JAVA tracking which might enlighten you: [webmasterworld.com...]

HBX is known for good customer support, and if you have questions setting up some tracking, you should call your rep (if you have HBX you have a rep).

One interesting thing about HBX is the filtering settings. Setting up filtering settings does cost more, but it can provide some very good comparison metrics.

There really isn't a 'normal' in regards to repeat visitors and repeat conversions. A content site will naturally gather more repeat visitors than an eCommerce site. Repeat conversions often depend on how you define a conversion and what you're selling (selling toothpaste will gather more repeat customers than selling computers).

[wolfgang]

Actually, it's important to note the difference here: I did mean New Respondents and Repeat Responses (not conversions). Someone who comes in through an adwords ad for the first time is a new respondent and if they come in again via the same tracking url (same day or later), they are treated as a repeat response to the Adwords campaign. My question was: I paid for 918 clicks. Is it reasonable to assume that if we had 515 new respondents and 463 repeat responses, that 403 (918-515) of those repeat responses were also part of the new respondent group (515) and that they simply clicked on more than one of our ads. And is it also reasonable to assume that they could have done all their repeat ad clicking in one session, resulting in HBX only reporting them as 1 visit?

If my assumptions are unreasonable, then the only other alternative is click fraud, which given our low average CPC (.09) is unlikely (but not out of the question).

You seem to know what you're talking about, I appreciate the feedback. Thanks.