Measuring Search Engine Marketing (II)

At Clickshift, we think a lot about the analytics behind web advertising.  I’ve written before about how some of the math impacts the measurement of search engine marketing.  Macro trends impact a lot of this measurement too, but I haven’t seen much of that show up with clients we meet.

For instance, the average price paid across a sample of 200,000 keywords for our clients was 85 cents in the first quarter of 2006.  In the second quarter, that price climbed to $1.10.  The quarter did see a slight rise in average position of the ads too, so we’d expect some rise in price (to do this analysis for real, we’d need to control for the average position).  All other things being equal, a client with an ROAS (return on ad spend) of 500% in Q1 would see that ROAS decline to 387% as keyword prices rose by 29%.  (As an example: $1 spent in Q1 would have made $5, but in Q2, it would take $1.29 to make $5 – for an ROAS of 387%.)  Most people, especially with help from companies like Clickshift, can keep the returns close to steady even with the volatility in keyword prices, but the problem gets much harder as the time horizon gets longer.  Plus, their management expects them to improve each year – and a 10% improvement in ROAS in the face of a 50% rise in keyword prices gets pretty tough.  

We’ve heard potential clients say “last year, I could get a user for $50, so I want to be able to get a user for $50 for the next year”.  Maybe they will, but if keyword prices in their category are up 100%, they’d have to do twice as well to hit their old target.  Of course, if keyword prices fall, they should be able to spend less to get a user, not more.  To really measure search engine marketing, then, we need to use a baseline, probably by category.  Sort of like measuring the performance of a mutual fund by comparing how much better or worse it has done than the index of its asset class.

Fathom has done some work on an index.  Unfortunately, that index has seen a bunch of criticism, and may not be a great baseline.  We need an industry index that is broad-based, category-specific, and built with appropriate sampling.  Only then can online advertisers really understand how much the macro environment impacts their individual efforts.  


Related Tags: Fathom, keyword price index, search engine marketing, online advertising