Hey David,

Thanks for your post. While the Ah-Ha moments are rare, we hardly ever forget them.

I agree with your comment “Bounce rates should only matter when you check your landing pages” to a point as I have seen cases where due to the structure of the website itself, the bounce rate can become very important upstream of the landing page, especially when customers exit because they can not navigate to the landing page.

Also, in regards to using drastic changes in your A/B tests, I thought you might find this blog I read from the Performable Web Analytics blog earlier this summer interesting.

http://bit.ly/htd90P

I think their blog post really sums up the fact that unless you have very significant volume, incremental testing just might not be the way to go. Testing two completely different pages might show you results in weeks that would take months if you did them one at a time.

Look forward to seeing more posts from you on more of our blogs.

Take care,

Chris

The aha-moment came when I ditched bounce rate in this scenario and looked at exit rates in funnels. Generally these two are tricky metrics as they produce deceiving averages and mean completely different things. Everyone has to exit a site at a certain point or most likely time. So exit rates really don’t matter unless you look at a defined funnel (i.e basket to thank-you). Bounce rates should only matter when you check your landing pages. Then the dissection of traffic source makes sense as you can then potentially identify why you get such a high bounce rate of these (misleading PPC ad, crap ad-copy, lost scent, etc). But again, be weary of these. A 100% bounce rate from mysuperlinkbuildingsitethatinstalltstrojans.com might not be too bad if it is only 3 visitors.

In General though I fully agree with the principles discussed here, 80% rule. It will be much easier to show your senior management you did something great if you improve something that makes it onto the dashboard report than increasing the revenue of a tiny referral by 1000%

I tend to run A/B test initially though with drastic changes. The danger is that if the change is too subtle you won’t see any improvement and management will say “screw testing”. So rather than changing the hover state of a button, change the whole lot and test a much bolder change. Yes we all look for (and read about) the million dollar button change. Chances are your business doesn’t make enough of those dollars to actually produce something like that and you will probably make more money improving your site overall.

Hi KC,

I kind of see your point here, however I think it is important to consider the level of investment (both time investment and fiscal investment) required to increase the conversion rates for a specific traffic source.

If a 10% conversion rate increase can be achieved with a similar effort for both the high-revenue and the low-revenue channels, a company would see a better return on their investment by focusing on the high-revenue channels.

However, it seems that depreciating returns do exist in the world of channel optimization. It may be likely that your high-revenue traffic has already approached a conversion rate limit, and that additional optimization efforts would be better focused on the lower-revenue, lower-conversion channels. Therefore it is important for companies to do a cost/benefit analysis for each traffic source when preparing channel optimization strategies.

Cheers,
Andy

I use the x-parameter tracking at the source by creating a link that drops a cookie right from the get go and then redirects to a specific landing page. This allows me to track all the way from the source – PPC, Email, General Advertising, Other Promos – to the conversion. I hope that makes sense.

Thanks Neil.

You mentioned using x-parameters before entering the cart, but how do you track the person all the way to the source, especially when the source doesn’t go directly to the cart? It’s important that you can carry tracking information through your website and landing pages to the conversion point or else you aren’t getting the full picture.

And if you have suggested topics for future blog posts, visit our Facebook page to post suggestions: http://www.facebook.com/pages/Building-Keystones/153683638001551?v=app_2373072738

Cheers,

craig.
http://twitter.com/buildkeystones

I have some questions regarding the Pareto principle. The principle as you mentioned states “look at the traffic sources that are making up around 80 percent of your revenue. The reason is simple: the higher percentage of revenue a traffic source is bringing in means the more likely a small conversion increase will improve your bottom line.”

What if the area where 80% of the revenue generated is only 3% of the total traffic coming to your site? Does this principle still hold true and, if so, could you elaborate why? When I think about it in this sense the Pareto principle doesn’t add up. If this were the case, I would be more tempted to optimize on the 97% of the traffic that is not converting. To put 97% in perspective, consider a website that has 1 million unique visitors every month.

Regards,
KC