When Claude Hopkins wrote the classic treatise on Scientific Advertising in 1923 he proclaim
ed that copywriters must acquire the most detailed knowledge of their clients’ products that is possible. Today, Hopkins would demand that Web writers, designers, producers, media planners, search optimizers and marketing managers apply the same discipline to understand the science of online analytics. The key would be to master the Key Performance Indicators (KPI’s) that define the measure of success in E-commerce. The early advertising pioneer who wrote “The beer that made Milwaukee famous,” might cheer about the way Jason Burby, director of Web analytics for Seattle-based ZAAZ has formed the team that is making Seattle famous. In this case, the fame is coming for analytics-based, performance-driven online marketing strategies that are powering results for Converse, T-Mobile, Alcoa, Microsoft, Qwest, E*TRADE and Reuters, to name but a few.
To say that Web analytics has gone mainstream is an understatement. Google has pulled the wraps off of Google Analytics (GA) and offers every Web marketer a free ride on the Analytics Express train formerly known as Urchin. The Google goliath has since named ZAAZ among a handful of Authorized GA Consultants. This month, longtime ZAAZ partner WebTrends launched a dramatically enhanced and expanded Website performance suite dubbed WebTrends MarketingLab and ZAAZ is once again in the driver’s seat. The Web firm also maintains a close relationship with Omniture and Offermatica as well.
What makes ZAAZ analytics tick? We caught up with Jason Burby, a well known ClickZ analytics’ columnist, analyst, strategist and industry expert during one of his recent commutes between offices in Portland and Seattle and covered the territory of Web measurement in some detail. What Burby espouses is more than just theory. As Jason is quick to point out, if you are not prepared to use the findings that are generated through analytics research by optimizing your Web site, you are wasting your valuable time and money.
Seattle24x7: How does ZAAZ incorporate Web analytics in its Web development and management processes?
Burby: ZAAZ is highly active in the analytics consulting space as a part of our core business of design, development and Web strategy. Web analytics enables us to gain knowledge and insight to help our clients become more successful. We’ve also formed a number of very close relationships with tool makers over the years. ZAAZ has had a long term involvement with WebTrends as one of the founding partners of their WebTrends Insight Network. For the past two years we’ve had a top-tier relationship with Omniture as well. We are also a certified Google Analytics consultant. Our approach isn’t as a reseller of these tools, instead we focus on the consulting and best practices areas where companies come to us to help them get the maximum value and return from these tools.
Seattle24x7: Can you outline the process you go through in helping clients to implement an analytics’ program?
Burby: We work with our clients in a four-step process. The fist step is defining goals and metrics. First, we work with the client to determine what are the overall business goals? What role does the Web play in the business? What are we trying to get people to do in each section of the site, or even down to the page level? At this step in the process, we talk about how to define KPI’s, what are the specific KPI’s relevant to your business, what are the main drivers of the business? Then we break that down further. We ask What is the information behind that? What’s the metric that we need to understand all of this?
The second step is then to configure reports within the tools that are aligned with the goals and KPIs we mapped out in step one. Too often we see companies jump to Step Two and get overwhelmed with the thousands of reports that are available now, without really understanding which specific reports are most relevant to their objectives.
Step Three then is analysis. This is where we’ll dive deep into the data to look for areas that are underperforming and thus create new opportunities for improvement. By monetizing and prioritizing the opportunities, we’re able to forecast the financial upside that specific changes to the site or to campaigns can have on the business.
Finally Step Four is optimization, where we help organizations take action on the data. This final step is where the real value of analytics and the potential of the Web is realized. We’ll create hypotheses and test different element, either through simple A/B tests of say a product detail page or campaign landing page, or more sophisticated tests using multivariate testing solution like Offermatica, where several elements are tested via recipes and dynamically you can understand which combination of elements, such as pricing, promotion, visual design, copy, color, button style, layout, .etc are contributing to better performance and conversion from your web visitors.
This process isn’t a 1-time and you’re done scenario, it’s a continuous improvement process and is ongoing. One of the things we try to communicate to all of our clients is that small incremental improvements over time can lead to significant returns to their business.
Seattle24x7: Can you give us a simplified example of how a Key Performance Indicator might be defined for a client?
Burby: As a purely hypothetical example, let’s look at Converse which is a ZAAZ client. The KPI might be the number of shoes sold per day, or the revenue tied to those shoes, or an average order value. But we need to look beyond that. What we’ve got to look at is what are the supporting metrics that meet that KPI? The second step, then, is configuring the tool to give you the data you need.
Seattle24x7: Using the case of shoe sales, what would be the supporting metrics you’d be looking for?
Burby: Well, for instance, one KPI that any e-commerce Web site might look at is purchase conversion rate. Revenue is another KPI and that can be revenue-by-day or revenue-by-week. Average order size is also something that is going to be important to analyze as is profit margin.
Another thing we’re often looking at would be store locators. We’ll also watch sign-ups for newsletters and study the impact of newsletter sign ups to getting people to acquire shoes or clothing down the road. We will generally leverage a lot of competitive data using resources like ComScore and Neilsen in addition to attitudinal data by topic surveys. The concept of tying in all of this behavioral data is really only meaningful if we’re driving towards changing and optimizing the site for business.
Seattle24x7: What other kinds of KPI’s should Web marketers be monitoring?
Burby: In some cases, the KPI is not about conversion. In cases like ad spending for sites that are supported by ads, you want to focus on recency metrics, how often can you get people coming back, and evaluating page views per visit.A new book entitled The Big Book of KPI’s written by Eric T Peterson does an excellent job of breaking down a number of KPIs for different types of Web marketers.
Seattle24x7: You’ve written that an integral part of the process is about recognizing opportunities?
Burby: The third step in the process is where we get into prioritization of opportunities. This means analyzing the data based on the goals we have set. When we look at a site for the first time, regardless of who the client may be — Intel, Washington Mutual, E*TRADE — they may have done some degree of optimization but many are still looking at their data for the first time. When we go through the analysis, we might identify 7-10 opportunities that we feel are the most significant improvements to the business interaction based on the initial engagement with the client. At this stage, we’re stepping back and estimating what would happen if we were to adjust these behaviors. Which has the greatest forecasted ROI? That’s where we’re going to focus.
Seattle24x7: How do you calculate the value and importance of various opportunities?
Burby: The method that we find is most effective with our clients is to monetize these online behaviors. Some could be revenue-specific, like an e-commerce transaction. Others could be producing calls to a cell center as an opportunity. Others could be looking at business leads. We will ask ourselves, do we think we can move this particular conversion or this particular metric up 2%, or 3%. We start to get an upside range. We believe we can get this particular metric, through optimization, into this range. Then we look at the level of effort on that and calculate out a forecasted ROI.
Seattle24x7: The key is not only recognizing opportunities but acting on them?
Burby: We had a case with a very large Telecommunications last year where we had two hypotheses. One that we were providing too much information on a product page, and another that we weren’t providing enough information. The analytics data isn’t going to tell us exactly what we need to do, it just tells us there’s a problem. So, as a next step, we actually put together a series of tests of different layouts for the product page that emphasized different aspects of the offer. You want to define your success metrics in advance.
As an example, you can look at a product page, any product page, and ask whether the the call-to-action is strong enough for someone to easily see how they add a product to the cart? An instance of that today may be within online banking. Banks are eager to get people to sign up for online banking and use Bill Pay as it greatly increases their retention of a customer. So when you examine the site, you want to identify several things: Are the calls to action to sign up for online banking strong enough? Do we need to test those? Should we position them in different parts of the page? Should we have them look different? Should we test the call-to-action?
Another way to apply variable testing that people can comprehend pretty quickly is a Web form. If I go to sign-up for something, or to check out on a purchase, and I confront a form that’s six screens long, that can create a major stumbling block. We may be able to capture the same information but have the forms layed out differently to make it easier for people to visually understand the breaks and what they need to enter where. We might also cut down on the amount of information being requested if it’s not absolutely required and see what effect that will have on conversion.
Seattle24x7: What tools do you use for variable data testing and what do you suggest for marketers who may not have the budget for more expensive tracking solutions?
Burby: We have a close partnership Offermatica. There are other tools to measure variable page tests as well such as Optimost, and ATG offers a tool within their content management system. At other times, we’ll build simple A/B scripts that redirect traffic.
Seattle24x7: How might a marketer go about setting up a simple redirect test?
Burby: For a smaller client, we recommend using redirect scripts which will drive traffic to different pages. We tend to want to see three versions of pages, where, for example, product1.asp is your control and product2.asp and product 3.asp is what you’re testing. In that case you’re just going to be testing one component on the page. It’s a manual way of doing it which means it’s a slower way, and you can generally test one thing at a time, but it’s much, much better than not testing at all.
Seattle24x7: How much should clients commit budgetarily to their Web Analytics efforts?
Burby: We recommend to our clients that they strongly consider allocating 5-10% of their overall Web budget including online marketing and online media to optimization. For a smaller company that may not be a lot. For a medium to large scale company that’s serious about online, that will be a more serious number. Of course, there’s no true ROI in analytics if you’re not actively using the data. Unless you get in and actually make changes to your site, you’re wasting your money. The biggest thing is to get in and try it. Get in and start somewhere. Take Amazon for example. Everything Amazon does is for a reason. So when you start looking at a site that has never been optimized, it’s pretty easy to really dig in with behavioral, attitudinal and competitive data and identify some opportunities that are going to pay off big.
Seattle24x7: Now that Google Analytics has gone mainstream as a free resource, how would you advise Web analytics newcomers to use it effectively?
Burby: One of the nice things about Google Analytics is that it’s pretty easy to jump in and get started. With GA, you can set up four goal conversions within the tool. Picking just the top three conversions on your site, or even two, and setting those up is a great place to start. If you’re selling product, you could begin by looking at your overall conversion numbers. If you’re generating leads, look at your lead conversion. If you’re considering customer service questions, look at that conversion. Take one specific desired behavior — purchase and leads usually cover most of them, and figure that one out.
The other thing we will look at when attempting to understand site performance is Exit Ratios. Not what are your top exit pages, but what pages bounce the highest percent of people that visit them. It’s one of the easiest indicators anyone can use to help identify problematic pages and potential opportunities for improvement.
Seattle24x7: Would you advocate using an exit survey to solicit opinions on the visitors experience with a site?
Burby: You can do that on an entry or an exit depending on the situation. The problem you can fall into pretty quickly is collecting too much data. After you’ve looked at your analytics and identified a particular problem you can then step back and ask how are we going to solve this, what do we think is going on? That’s when we’ll generally set up a survey. If you just set up a blanket exit survey, or install surveys all throughout a site, you can get some great information out of it, but too often you will be unable to digest or use that information.
Larry Sivitz is the Managing Editor of Seattle24x7.
