Online Marketing Crimes and Misdemeanors

Analytics and online marketing guru, Avinash Kaushik, came up with a particularly memorable quote at last week’s eMetrics Marketing Optimization Summit: “I came, I puked, I left!”

Formerly with Intuit, and with a background decision-support systems (large multi-terabyte business intelligence tools), Kaushik is an author and a very motivational public speaker. When I reviewedhis book, “Web Analytics”: An Hour a Day,” in August, I thought the title was deceptive because it’s very much a marketing book, as opposed to pure analytics.

As analytics evangelist for Google, Kaushik strives to help people understand the value of data-driven decisions. At the metrics marketing summit, he and I worked together on a few sessions at the conference, which were not only very informative, but also a lot of fun.

His take on conventional advertising as a “faith-based initiative,” which assumes advertising will translate into revenue, is thought provoking. Backing up that statement, he elaborates: “As soon as you implement a Web analytics tool you will be 30 percent smarter within a week than you would be in any other channel.”

“Bounce rate,” almost Kaushik’s mantra, is a topic he scrutinized during his presentations. Very few people, he said, measure this very valuable metric. Online marketers, he said, talk at length about click through rates, cost per metric, and cost per acquisition, but very few marketers ask: What’s my bounce rate? Kaushik’s defines bounce rate as the portion of people who come to a Web site and only see one page. He sums it up, little more crudely and humorously, as: “I came, I puked, I left.”

It’s a crime, he says, the number of marketers who use paid search to drive traffic to their Web sites and then fail to measure the number of people who leave instantly. Average bounce rate is around 60 percent.

Up until quite recently, tools to highlight this metric weren’t built into analytics vendors’ tools. However, as Google’s analytics evangelist, he’s quick to point out that Google Analytics made bounce rate permanent in every single report and by every single dimension. In a single view in the Google analytics tool, the number of top landing pages is highlighted; next to that metric is each page’s bounce rate.

Cannibalization is another term Kaushik frequently uses. He says this occurs when marketers bid on their branded terms, such as Wal-Mart bidding on its own name. Say, Wal-Mart is already number one in the organic results for a search on its name. But the retailer also has paid listing above that. People click on the paid listing to get to the same page they would have done if they clicked on the organic listing. So, in that sense, paid search cannibalizes traffic from organic.

A number of online marketers may disagree with this premise because they believe being number one in both paid and organic is good for branding purposes. Test your theory, he says. Turn off your paid listing for a month and see what happens to traffic.

During one session where we presented together, an audience member asked Kaushik how to choose an SEM agency. I found his answer to be most interesting. While working at a large company he asked 10 vendors to pitch for the account; each gave a two-hour presentation. He then whittled the list down to three finalists, each of whom had a different paid search strategy.

One took the portfolio or automated model, which calls for letting mathematics take care of business.

Another method, described by Kaushik as the “hamster” model, referring to hamsters running very fast around a treadmill. Basically, this is human-based approach, which could be outsourced to somewhere like India, for instance.

The third approach is a rules-based model. Humans manage keyword, while other work is based on mathematics.

Kaushik’s approach gets interesting here. Each of the three vendors received a budget to manage paid search for six months. After that, the company would choose the vendor that outperformed the others. I thought this was a fair and constructive approach for selecting an agency.

Embrace incompleteness in a data-driven environment is another theme promoted by Kaushik. While managing the expectations of senior people a couple layers above you, he said, it’s not necessary to get the very last drop of data to make a good business decision. People involved in analytics can get too deep in the weeds trying to be perfect.

It reminded me of a friend who once said to me: “Don’t let excellence get in the way of the very good.”

You can hear my complete interview with Kaushik here.

Podcast: Play in new window | Download