Measuring Up Your Keywords and Conversions

As legend has it over 2,200 years ago Archimedes was challenged by King Hiero II to discover whether his new crown had been made from pure gold or mixed with lesser metals. Figuring out how to measure the volume of the crown with its irregular shape presented Archimedes with a major task.

Today's PPC managers face a similar measurement challenge. Namely, how do you model exactly how much each keyword contributes to each of your conversions? And given that PPC drives nearly 1 in 5 visits to UK retail websites [1], such models are more and more likely to find their way onto CFO and CEO desks.

At one end of the spectrum is 'last click' attribution. Simple and straightforward, the only keywords that count are those that lead directly to a conversion.

Most people would concede however, that this method is not the ideal way to measure keyword conversions as it doesn’t consider how much of their previous search activity led to the final purchase. The 'consumer journey' has already been the subject of much discussion. But what then is the other end of the spectrum? How do you attribute important metrics like new customers and lifetime value to a keyword? A 90 or 180 day cookie, tracking everything and anything? The data could be overwhelming.

Research commissioned by Google indicates that as many as 40% of conversions occur after 4 weeks from when a shopper first performed a search related to the goods or services they purchased. Yet no one wakes up in the morning and says to themselves: “Today I will start my journey to buy a new coat, it will take 18 days for me to finally get there and buy it”. You make a start by seeing what products are available; take ideas from various sources and then when you think you have found the best item and deal, you make your purchase.

Is this because we all love browsing, or because it can be difficult sometimes to find exactly what you want? Like many things, it is probably a combination of the two. The Consumer Decision Journey recently published in the McKinsey Quarterly [2] discusses these moments of influence and how the 'purchase funnel' may be a different shape from what was once thought.

One thing that should gain the consensus of agreement though is that the consumer journey is path dependant. That is, each subsequent step is dependent on the experience from the previous point.

Accepting this, how much can each of these prior steps claim to have led to the final conversion? Not an easy answer given the total number of steps, the media involved and the time taken. Within the overall context of search however, comScore can shed some light on the path dependant nature of conversions within three major consumer categories [3]:

Metric	Clothing	Computers	Mobile Phones
No. of searches per conversion	5.7	5.0	4.7
No. of site visits per conversion	11.9	12.6	9.4

So how do individual Retailers start to better understand and act on the challenges posed by accurately measuring their own conversions?

You don’t necessarily need another Archimedes, but certainly someone who will take responsibility for devising how you go about measuring your conversions. In this case, using data from your web analytics package to discover exactly how each keyword pays it way. There are numerous bid management programs on the market too that can make life easier; not to mention third party consultants who can help install and make sense of analytics. Today, retailers of any scale just have too many keywords to do this manually and hope to achieve their own “Eureka!” moment.

[1] Hitwise UK, Shopping and Classifieds category, August 2009
[2] http://www.mckinseyquarterly.com/Marketing/The_consumer_decision_journey_2373

[3] comScore custom analysis- UK Population, 15+ Home and Work, Q4-2007 activity

Posted by Jonathan Marsh, UK Retail Team