“Measuring Google AdWords Performance” – An interview with David Rodnitzky CEO of 3Q Digital, a Harte Hanks Company
David Rodnitzky is founder and CEO of 3Q Digital, a position he has held since the Company’s inception in 2008. Prior to 3Q Digital, he held senior marketing roles at several Internet companies, including Rentals.com (2000-2001), FindLaw (2001-2004), Adteractive (2004-2006), and Mercantila (2007-2008). David currently serves on advisory boards for several companies, including Marin Software, MediaBoost, Mediacause, and a stealth travel start-up.
David, a tried-and-true business axiom is: you can’t fix it if you can’t measure it. You went a long way toward answering this challenge regarding Google AdWords spending with the Lin-Rodnitzky ratio. Can you explain the ratio to us?
The ratio divides the cost per acquisition (CPA) of all non-brand queries by the CPA of non-brand queries with at least one conversion during the measurement period (usually 30-90 days). For example, say the overall non-brand account CPA is $50 and the CPA of non-brand queries with at least one conversion is $25. $50 divided by $25 = 2.0 – that’s your Lin-Rodnitzky Ratio.
We generally say that a ratio of 1.5 to 2.5 is a good range. Below 1.5 suggests that you aren’t experimenting enough, and a ratio over 2.5 suggests you have a lot of queries that are costing you money without providing a return. So, a $50 overall, a $25 one-plus and a ratio of 2 suggests that the AdWords are performing well.
When did you first develop the ratio and start using it?
One of the main ways performance agencies like 3Q Digital win business is through account audits. A client gives us access to their AdWords account and we dig in and find opportunities to improve performance. The challenge with audits is that the recommendations can seem somewhat qualitative. There is a built-in bias to find problems; the agency has a financial incentive to tell you that things are bad so that they can win your business!
We wanted to create a benchmark that we could explain to the client before we did the audit – to provide a quantitative, non-biased perspective of their performance. In this way, the client can make more sense of our audit findings.
We created the Ratio around 2012 and it’s been great to see its broad adoption throughout the industry. There definitely have been critiques of the ratio. I think the one I hear most is that it doesn’t work for certain types of accounts. For example, some companies buy keywords that have a very low conversion rate, but are important from a branding perspective. There’s a whole thread on Reddit about the topic.
In general, generic benchmarks are always going to perform generically – the Lin-Rodnitzky Ratio was never intended as an absolute. But benchmarks are still helpful to directionally inform teams as to whether they are at least in the right ballpark.
Is the Lin-Rodnitzky ratio still useful if companies spend millions rather than thousands of dollars on AdWords?
The utility improves as the data set gets larger. If you are spending $100 a month, the statistical significance of any data is going to be quite low. Most of our clients are spending north of $100,000 a month, which provides plenty of data to analyze.
A frequent error that people make when calculating their L/R ratio is this: the ratio only applies to non-brand queries. If you include brand queries, your ratio will look great, because brand queries have an exponentially higher conversion rate than non-brand. The L/R ratio was specifically designed to evaluate the effectiveness of non-brand, because that is where you can really determine the efficiency of an account. In most large companies, brand accounts for a substantial percentage of the spend, so you need to be very careful to exclude brand traffic from the analysis.
Note also that there are plenty of other ways to determine the effectiveness of your AdWords spend. For example, most ecommerce companies use return on ad spend (ROAS – revenue divided by ad cost) as a bottom line measure of account success. It should go without saying that having a very low L-R ratio but a terrible ROAS is not a win for your company. It either suggests that you are optimizing for low revenue purchases or you are bidding too much for your queries.
Similarly, for very large companies, CPA is usually driven by requests for more information, whitepaper downloads, or similar activities that are very top-of-funnel. If you only optimize CPA, you can get into trouble by encouraging “looky-loos” rather than serious buyers. For example, if someone types in “free enterprise software”, they might fill out a form, but they aren’t very valuable to your business.
So again, the ratio is not an absolute – it is a guide to help you understand your performance.
Do you have any suggestions on how marketers can be more predictive or prescriptive with AdWords planning?
The most beautiful thing about AdWords is how quickly and affordably you can test, refine, and improve your results. A lot of testing can be done with just a $500 monthly budget. For example, let’s say you want to test different ad copy. At $1 per click, you can run 250 clicks to Copy A and 250 to Copy B and have a high level of confidence toward picking the winner. You can do the same sort of test with your landing pages, geography, devices, time of day, day of week, etc, etc!
The other cool thing about AdWords is that all your historical data is available 24/7. So, every day that you run campaigns, you are aggregating more and more data that you can mine to predict future success. To do this properly, you need to either be a SQL pro or have a combination of API access, a data warehouse, and business intelligence.
INTERVIEW by Mark Blessington.