Buy Branded Keywords? A Case Study on Traffic, Conversion and RPV
Part 3 of 3
This is the last part of my deep dive into a question that is asked quite often. Brand keywords have seemed to be a conundrum for many in the Marketing and Brandings worlds. This series of posts and research has yielded some interesting data. Like any good testing and optimization they answered the questions posed and raised many more. It also seems timely as yesterday Ad Age and Search Engine Land reported on Travelocity CMO Jeffrey Glueck's discussion of the topic at the IAB Performance Marketing Forum. Now, the final verdict (for this client at least).
Challenges: An OTTO Digital client (major online retailer) has been spending thousands of dollars a day on PPC ads for its brand name keywords. With no competition on ads for these terms and with the recently updated layout of Google’s natural results on brand queries, they wondered if there would be any negative effects to traffic and sales if they stopped buying these terms. The hypothesis is that the same traffic will end up at their site and behave in the same way regardless if the paid ads are present or not.
Thirty branded keywords were tracked across Google, Yahoo and MSN over two weeks using Offermatica for monitoring behavior and reporting data. One week the keywords had search ads associated with them running and the other week they did not. There were no issues of seasonality for their business over this period and no promotions were ongoing or started over the test period.
Traffic: Traffic levels were incredibly stable over both weeks. Having the top sponsor ads present did not factor in acquiring more traffic to the site. In fact, not having the ad present resulted in more traffic on Google, Yahoo and MSN. The margins in traffic were so close that it is hard to gain confidence in this effect but it is counter-intuitive. Here are the traffic numbers in unique visitors as measured in Offermatica.
Conversion Rate: We pulled a few interesting metrics from the conversion rate data. Most interesting to me (besides the fact that Amazon.com still has not been awarded by Google category sub-links) was the fact that the category sub-links under the main Natural result link had a lower conversion rate than the main link.
This click effect was consistent when ads were present or not. Best practices would tell you to drive the user further down the site experience by landing them on the Category page (what I think about best practices). This is especially the case if users can get directly to a category with affinity from a brand query. Counter-intuitively, clicks on the primary natural result link had an over 30% better conversion rate than clicks on the sub-cat links below. Let’s slice the data on conversion rates with ads present vs. not.
Net Revenue and Revenue Per Visitor: This data is by far the most interesting and the most important. It’s one thing to look at the impact on traffic and conversion but the determination of whether to buy the branded keywords ultimately relates to the incremental revenue credited to them.
What is clear is that not only are the ads paying for themselves, they are driving huge chunks of additional revenue and are major lift in Revenue Per Visitor (RPV). It is especially interesting to see this broken out by channel. The chart below shows the incremental lift in dollar sales and RPV gained by having the search ad in conjunction with the natural result.
Conclusion: Running the PPC ads alongside the natural results for brand keyword queries provides a huge lift in orders and revenue for our client. The paid search ads clearly have a tremendous influence on user behavior. The ads did not drive more traffic but they drove more traffic that purchased.
This is interesting because so much of search engine marketing is about finding people at the “right time. This intent based marketing has built SEM. This data shows that search ads also have incredible power to influence post-click actions, possibly even triggering intent. Another possibility is that it is not the ads themselves but merely the presence of the ads that causes such profound conversion and RPV impact.
Bear in mind however that these thirty keywords were brand terms. It would be very hard to find an analogous non-brand situation (sole top sponsor result and top ranked natural result) that could remain static enough to collect the data set needed for confidence. However, clearly this is an area that requires more study. I’ll be diving a bit deeper into this data during my presentation at Search Engine Strategies, Ad Testing Research and Findings coming up in New York on April 12th. I also hope to blog about results for other OTTO clients in different verticals in the coming months. With so much brand money coming into SEM this is an area that should continue to be of interest for sometime.