Lastminute adjusts Google Adwords approach after attribution test

A Google Adwords experiment by Lastminute.com has found attribution for non-brand PPC is being underestimated by 43% by conventional analytics.

A Google Adwords experiment by Lastminute.com has found attribution for non-brand PPC is being underestimated by 43% by conventional analytics.

The online travel agent ran a geo-experiment test over 12 weeks between October 2011 and January 2012.

It took a sample of cities which generated over 1,000 bookings during the trial period which it divided into two categories, one of which formed a control group.

This control group had its Google Adwords campaigns switched off for the 12 week test period.

The data was crunched to take ensure there was no bias in terms long-term cookies picking up pre-existing marketing activity and for issue around precisely determining customer location.

Lastminute said the experiment found that booking trends from the cities in the experiment were closely correlated to the overall picture.

And two key drivers were identified: the first being a seasonality trend and the second how customer behaviour in the cities was linked to offline marketing, demographic profile and brand awareness.

Lastminute said the test showed how the ‘last click wins’ rule does not give proper value to channels that appear early in the conversion funnel, such as those used by customers to research and compare offers.

Reporting the results of the experiment, Lastminute said: “The core results of the study revealed that for every tracked conversion that was previously attributed to Google AdWords, 1.43 conversions should have been attributed in this way.

“In other words turning off the non-branded search terms in Google Adwords will results in a 43% larger decline than reported by Lastminute’s tracking systems.

“Thanks to this insight, Lastminute’s next steps include adjusting the brand’s paid search budget and engaging in further investigation into the causal effect of Google Adword activity.”