Let's Get Analytical

Online advertising generates a tremendous volume of data — and that’s one of its big strengths. Our nascent medium has always been lauded, especially during the big boom period, for its awesome potential to track actual responses quickly. Sure, direct marketing (whether TV, print, or otherwise) has been tracking responses for years. But certainly no other medium affords the optimization opportunities available with Web marketing.

The question is when to stop. There are impressions, click-throughs, back-end action tags, page views, session times, clickstreams, repeat visits, recency, frequency, duration, stickiness, and so on. What it means to most people — even some of the savviest marketers around — is simply data overload.

What does it all mean, and how do you know what is most important? How can you possibly ever hope to consider all of the various data elements available when making your optimization decisions? It’s scary, and it requires some creativity.

CTR Is Out

In the early days, CTR was king. It is the easiest metric to track but has quickly fallen out of vogue — and with good reason. Very rarely do you encounter marketers who set out to simply get people to click on their banner ads. When the brand manager is discussing results with her supervisor, do you think she could get away with saying: “I spent $250,000 online last year, and I got more than 500,000 clicks!” If I were that supervisor, I would want to know what that meant. Who cares how many people clicked? How many people bought something? How much revenue did those 500,000 clicks generate for us? That is all that matters. So it boils down to watching what the users do after they’ve clicked on the ad.

I think we can safely say, for the most part, we as an industry have moved past CTR. Smart marketers are looking at other, much more important data. Many, though, are still not going deep enough.

Tracers and Spotlights and Actions, Oh My!

Most marketers are using some kind of back-end tracking, primarily using a tracer image/action tag from their third-party ad-serving company. All of the major ad servers support this kind of tracking, and many can do some enhanced analytics based on data gathered by the tags.

These tags are great for tracking a specific action or sequence of actions on the marketer’s Web site. For example, by tagging the thank-you or confirmation page, a marketer could see how many people driven by the ad campaign signed up for a newsletter or purchased a product.

Advanced implementations might look at a multistep process, tagging several individual pages. Or your ad server might offer view-through tracking (users who saw but did not click on your ads, only to later hit one of your tagged pages). Or perhaps you can get even more complex with things such as site overlap, unique users/reach, frequency, and so on.

But most action tag systems fall short when it comes to analyzing what the user does throughout his entire visit on the site. Theoretically, you could tag every page of your site if it were relatively small, but, still, most ad-serving reporting systems lack the ability to measure critical site metrics, such as session time and clickstream.

Web Site Analytics

The best insights come when you integrate ad-serving data with Web site analytical data. Site analytics platforms have come a very long way — many of them began largely as technical tools and provided little in the way of actionable marketing data. But on the market now are some really robust systems that can help marketers engage their prospects and optimize both the Web site and promotional efforts — based on critical metrics that can tie directly to the bottom line.

These tools can be particularly useful when your products are not sold directly online. For example, a pharmaceuticals company clearly cannot simply offer its products online via an e-commerce engine. So these companies tend to provide product information and background, condition or disease state details, self-evaluation tools, and materials to take with you to the doctor.

But how do you know what metrics are important in these cases? We typically recommend a combination of analytics:

• Qualitative research such as intercept surveys or focus groups to assess mindset and the intentions of users before, during, and/or after the site visit. This kind of analysis can help you determine which sections of the Web site are most likely to generate the desired action and why.
• Behavioral analytics using some kind of site analytical platform. This gives you insight into what users are actually doing (as opposed to what they say they are doing) on the site. This way you can verify, to a certain extent, the information you gather from your qualitative analysis. You can also pinpoint hurdles in the decision-making or purchase cycle and try to eliminate them by tweaking your site. You can consistently watch behavior to ensure any such changes have the desired effect.

This kind of research and analytics can be a little on the expensive side. But look at it this way: You’re spending all of that money both on the development/maintenance of your Web site and on the promotional efforts to drive traffic there. How can you afford not to know if the site is effective at getting users to the next level?

Larry Everling is off this week. Today’s column ran earlier on ClickZ.