Analytics Tracking Performance – Beyond the Page View

Here we are, last session of the day. Let’s make this one simple and clear, okay? Mostly because I can’t manage anything else. My wrists are killing me and my battery is dying. Moderating this session is Joe Laratro. Speakers: John Marshall, of Market Motive and Scott Orth, of GTS Services will be teaching us all how to make the most of an analytics program. Steve O’Brien from Unica Corporation couldn’t make this session because he got stuck in London.

John Marshall begins. He’s got a fantastic accent. He’s going to be dissecting a single click. Hopefully, he’ll do so very slowly. He founded ClickTracks and was constantly asked about accuracy and tracking and all sorts of questions with complex answers. This presentation is about answering those if he can.

He’s making a few assumptions for the purposes of this presentation. This is going to be a fairly average scenario, it’s a PPC ad with a functional Web site using a modern Web analytics tool. He’s not covering ROI breakage–it’s really complicated–or cookies and cookie deletion.

So to begin: We click the ad. The browser goes and GETs the page. At that point, Google counts the click. Then you get redirected to the destination URL of the site with the full URL.

Error can occur here–1% of the time the browser drops the requests.

When you go through a redirect, the referrer is supplied by the browser. When the server redirects and 301s the DNS lookup happens “almost certainly”. It’s already at Google so there doesn’t have to be a DNS lookup there. Your site as the advertiser isn’t stored in the browser, so they have to do a lookup. In one percent of the cases, the DNS fails. It relies on their ISP not your server.

The referrer stays the same when you go through a redirect. It’s always the page the human being saw.

Site redirects again to hide tracking URLs. If it’s a JavaScript redirect then 100% you’ve broken your tracking. If it’s a different kind of redirect then there is a 1% browser drop failure.

Browser requests the landing page now. 5% cached pages error here leading to a total of about 8 to 10 % error from Google’s tracked clicks.

(For JavaScript, it grabs the elements, does another DNS lookup, execute tracking JavaScript, another DNS lookup on the data collector. The total error rate for JavaScript is probably about 10% compared to Google’s tracked clicks.)

Y’all, I understood about three words of that.

Scott Orth comes to my rescue with something less technical. Scott is starting with this statement: “Don’t Assume.” And he’s got a video for us, yay! I don’t have to blog that! (It’s that TV commercial about…oh I’ll just link it from YouTube later. Ameriquest commercial 1)

82 percent say that Web analytics is poorly understood and/or not used in their organization.

Behavior models matter. You need to think about all the steps. Understand how your site fits within the comsumer cycle. Don’t assume that you know the steps because it’s your site, you might be missing something. Doing a behavior model helps figure out how people are using the site.

Questions to ask: What are your Corporate Goals? How does your Web site support those goals? What are the Web site’s goals? Are they aligned with the Corporate Goals?

He has several examples of how to align goals, I’ll just pick one: Corporate goal is to increase revenue so the Web site goal would be to increase online sales.

KPI Assignment

Key Performance indicators break down goals into smaller organized targets. “KPIs are always rates, ratios, averages or percentages. Never raw numbers.” Show things in relation.

KPI examples for the increased site revenue goal

• Increase search engine ranking/presence
• Increased visitors
• Increase conversion rates

If things fall apart, you can go back to those three things and figure out what went wrong.

(another video. I promise, I’ll link these later. Ameriquest commercial 2)

Results may be different than they appear. Look at them by comparison. Is there a percent of increase or decline? What is it compared to this month, last month, etc.?

Do summary reports. Drill down into each KPI and figure out WHY things happened.

Analytics are important but knowing how to use them is more important. Use KPIs to track and report analytics.

(another video.)

I’m skipping Q&A, my battery is basically dead. Sorry, guys!

Susan Esparza is former managing editor at Bruce Clay Inc., and has written extensively for clients and internal publications. Along with Bruce Clay, she is co-author of the first edition of Search Engine Optimization All-in-One Desk Reference For Dummies.

See Susan's author page for links to connect on social media.