(Part V of a Series on Web Analytics for Search Engine Marketing)
There are lots of ways the web analysis can supplement or improve your overall Search program. Many of these take advantage of the unique view that web analytics provides across PPC, SEO, external advertising and direct traffic. Today’s installment is on Conversion Proxies – tracking engagement with a site when there are either no explicit paths to conversion or the number of conversions is too low to be used meaningfully.
When a business is multi-channel and/or conversion is offline, then optimizing PPC buys can be nearly impossible. This is a significant problem for both large site "brick-and-mortar" players and smaller sites with online conversion but with volumes that are too low to measure with statistical meaning.
The most common solution to this problem has been to let the buyer "optimize to clicks." There’s probably nothing an agency likes better than to "optimize to clicks." Doing so is incredibly easy and – in almost every case – much worse than no optimization at all. If you just pick the words you thought were significant to your business and used a flat bid for them, you’d almost certainly end up with a much more intelligent program than if you optimize to clicks. The reason? Optimizing to clicks will almost always lead to significant over-buying of the worst qualifying terms. An online stock-broker, for example, will end up buying lots of "free stock quotes" and not much "open trading account." Why? Because you’ll get more clicks. I cannot emphasize this enough – optimizing to clicks is stupid – consistently worse than no optimization.
But is there an alternative? With web measurement, there’s actually a very good alternative. By building "Conversion Proxies" – measures of engagement based on specific behaviors or quantities of behavior - you can provide comparative measures of qualification by Search Term (or Ad Group) that are nearly as good as (and in some cases better) than true conversion. These can be complicated, but often even very simple measures do a dramatically better job of optimization. We were recently able to show a client the dramatic difference in optimization that just looking at 2+ page views – visitors who did anything beyond the landing page – made. Typically, we like to deploy multiple levels of engagement. These usually range from simple numbers of page views/sessions to more complex packages of behaviors that reflect increased engagement with a site.
By comparing Land Rates and Engagement Rates at the Ad Word and top keyword levels, we are able to provide vastly improved optimization information to the buying agency.
Why might a Conversion Proxy be better than actual conversion? In one sense, it never will be. At the end of the day it’s always conversion you care about. But unless your conversion volumes are enormous, there will always be a significant chunk of your SEM program for which conversion numbers are too small to analyze meaningfully. A conversion proxy can greatly deepen the number of terms and ad groups than can be analyzed – so if the proxy is a decent one, the end result can be an actual improvement in overall program optimization.
For PPC programs where online conversion is non-existent or a poor indicator of overall performance, this technique is absolutely essential – and can vastly improve the real performance of a PPC buy.
Comments