Determining Site penetration by adword

I think this is what these number mean.

[qfguy]

I have been trying to determine which adwords cause the deepest penetration into our site. Because of the in which arena we operate and the complexity of our offering CTR is not the best measure of relevancy.

So, Here is what I did and I *think* i am interpreting the numbers correctly. I placed the same 'page view' conversion code on every page of the site. So if I then generate a report of page views and conversions per word i can calculate the ratio of page views to clicks. This gives me an average number of pages viewed broken down by adword.

If i am interpreting this correctly (and I think I am) it has been an eye opening experience. Our most popular word (highest CTR, clicks and cost) has a ratio of 1.1 meaning that practically all the users who arrive on that word leave after seeing the home page. Some other words have much better ratios like 8 or 9.

I am no statistician but I imagine I should somehow factor out how a word with less traffic might 'skew' these results.

[danieljean]

That makes sense to me. I haven't yet created the scripts for automating the statistical analysis, but in my last Adwords campaign I looked at how often people clicked on another page from the logs. An alarming number of visitors did not go past the landing page. If the CTR measures ad relevance, percentage of visitors that do something after the landing page measures the page's relevance.

I am working on some major changes but when done I will run the campaigns again. With several hundred clicks a day, I won't need to wait more than 24 hours to get statistically significant feedback- so I might do a few changes before I consider it "good enough".

Of course, you could increase pageviews by visitors or CTR without changing your conversion rate, so it's always tricky calculating ROI.

[qfguy]

Yes we do a high cost low traffic service oriented business. My techinque makes Google's conversion code ROI meaningless.

[AdWordsAdvisor]

...that practically all the users who arrive on that word leave after seeing the home page...

This brings up the point that, if you site is organized appropriately, it is often better to send the user to the page pertaining to what she or he has already searched for, rather than the home page.

Why make them search again, after all, for what they've already searched on?

I say this knowing that in some cases there will be a compelling reason to send the user to the home page. But if not, then consider sending them to the page that is most targeted to what they were looking for in the first place.

Just my $0.02

AWA

[qfguy]

Thanks I already do that for some words. Does not apply in this case however.

[cline]

qfguy, that sure looks like a rough way to do the analysis. Why don't you just buy something like ClickTracks and analyze your log files? (Or hire a consultant who does this sort of thing to do it for you).

With ClickTracks you can drill down by click source and search terms, then analyze time on site.

I would really recommend reserving conversion tracking for measuring key behaviors, such as buying or making an inquiry.