Internet Marketing
« Global Search for the B2B SEM | Main | Keynote Roundtable: Technical and Information Giants »
August 19, 2008
Measuring Success in a 2.0 World
First panel of the day and I'm elbow to elbow with people eager to see what's coming. Andy Beal is two seats to my right snapping pictures like a paparazzi.
Richard Zwicky (Enquisite) is moderating this session with panelists Jim Sterne (Web Analytics Association), Matt Bailey (SiteLogic), Avinash Kaushik (Google) and Marshall Sponder (Monster.com). This is going to be a great session so let's jump right into this, shall we?
Avinash Kaushik is up first. Thanks to twitter he's now going by @avinashkaushik. Hee. He shares some pictures of his kids. They're so freaking cute.
Why is 2.0 such a challenge? [Slow down, Avinash!]
Content Creation, Content Distribution Content Consumption. The current flow is broken. Right now the BBC creates the news, puts it on their site and then people read it. You can track that. But the problem is now that you've got people other than the BBC who are posting on the BBC and distributing their content and that guys is also posting on his site and distributing it through RSS and it goes everywhere and now you can't figure out where the content originated and you can't track anything.
It's different than it used to be but he thinks it's better now.
How to track this new world?
Multiplicity. You have to use many different tools to track each different distribution channel. You can't build a house with just one tool. You need many different tools.
Unique Measures for a New World: it's more than just time viewed and traffic and unique visitors, but you need more than that. What he thinks is great is RSS. It's permission marketing at the best. You can push content to people who want to see it. He tracks feedburner every single day, not really the actual number but the growth. That's a critical metric in the 2.0 world. Think of unique measures that are more relevant.
Unique Data Collection: There are new things out there that break the usual tracking. Gmail is one page view. That stymies a lot of analytics. Hovermaps where you don't get a click, still one page view. How do you track all that. You need to look for a different way to track and measure. It's events not page views.
Jim Sterne follows Avinash. He speaks slower. Yay!
Web 2.0 is now Web 2.0.1. Things are changing a little bit. You site is a window into visitor behavior. There's data everywhere. You can track almost anything anywhere. Videos can be tracked not only for views but for where the rewinds happen, when the viewer abandons the video, etc. Web metrics is really growing up. It helps you get into the hearts and minds of your
Search metrics is on the same path. Ranking to Traffic to Analysis to Dynamic Bidding to Predictive Buying.
The little hitch is the economy. People are cutting spending. They're focused on maximizing profit.
We need to understand not what brings leads, traffic, rankings. We need to understand what brings PROFITS. You need to connect everything or your budget will be cut.
Know what your goals are.
Short and sweet. Awesome.
Matt Bailey jumps up next. Hi Matt! He talks really fast so I'm sorry that you're going to miss this entertaining presentation.
Captain Kirk is analytics GENIUS. Green alien women were the key to be safe.
Apparently this is the first sexist analytics session we're going to have. Oh Matt.
Analytics 1.0 is pages of charts and graphs and it runs you down.
Anlytics 2.0 is about questions! Something logical, something creative, something software can't do. Asking how, why, what if? You need to look
Context, comparison, contrast. Ask WHY?
Take an assessment of the situation. 450 people, five year mission, 54 total deaths (that's our conversion rate.)
But that doesn't tell you anything about the mission itself. You need to segment.
Death rate by shirt color: Yellow shirts die 11% of the time, Blue 9% Red 79%
But that still doesn't give us context. We need to look deeper. Keep segmenting. Segment the segment. Onboard deaths versus planetside death rates. Now we're getting a little closer to where the conversions (ahem) are happening.
Bring in another segment. In the episodes where Kirk meets a woman, the death rate drops to 16 percent.
People are not cows, they don't move through the site in a herd. They come to your Web site for different reasons through different keywords. Which means you need a different conversion rate for every segment. You have to know what they're looking for. Context is everything.
Telling a story is the only way to compare and contrast.
1. Ask questions
2. Take action.
"Question-asking is the most significant tool human beings hands."
Marshall Sponder steps up to the podium. He gives us a little bit of his history. He's from an artistic background and he didn't bring any slides.
They're studying social media. Social media standards are coming later in the year from WAA.
Social networks get their traffic from social media not from search.
Web 2.0 is about empowering people to contribute. The way that you can empower people on Facebook, etc, is giving them tools to communicate.
Social media traffic is more directed than search traffic. Unlike the traditional idea that search traffic is the most directed but it's really social media. The bounce rate is high, but the long tail is where the traffic is.
Look at the data like an artist. Focus on the big pictures and the patterns. There's a ton of data coming at us all the time. Try to answer the question that sparked the report in the first place. Try to figure out what companies need and answer their questions.
You need to have something for people to do when they hit a site. It's not just getting them to the site, it doesn't just stop there. You have to keep going and lead them.
Search is part of marketing but social media isn't really part of anything and it's not really clear where it belongs. It's hard to qualify the ROI from that. At the end of the day the CMO is going to be the Web analyst. Measurements are going to be baked in at the very beginning. You won't roll out things that you can't measure.
Q&A
Social media traffic, how do you segment it?
Marshall: We look at the logs and see where people are coming from. I want Google analytics to take into account ComScore data.
Matt: Again it comes down to segment the segment. You can't treat all social media traffic the same. Go back and see 'why are people coming here'. See what was said on Twitter or on Facebook. Set up the context. Here's what they saw as they clicked that link, now you know what people were expecting to see when they clicked and how well it matched up with what they got.
Page views, are they dead?
Avinash: They'll continue to be relevant for a while, most of the Web is still flat, static pages. We need to measure more that that a page loaded in a browser. We want to measure interaction. Page views will be with us for a while but they're not what we're looking for.
Jim: What he said. What I want to know is why someone came to my Web site. Viewing a page was not their goal. I want to know what it is.
How do you engage around What If?
Matt: You're off charts and graphs as soon as you ask what if? You're into the realm of creativity. Now you're into event trapping. It's not just about how long. When Vista came out, the average page visit was high but do you think they were having a good experience? You need to think about what it means and come up with creative questions about 'what if' then think of ways to test it.
Avinash: Before, the problem is that things moved so slowly that you had to torture the data to get it to confess. Why do you have to do that? Why not just do it? "Will this convert? Let's look at the data, blah blah" Why? Just do it. Send out a test email. For a lot of questions, the Web reduces risk, you can just try it. You don't need to spend a lot of time data crunching.
Jim: Pick your bumper sticker: Don't Make Me Think. Always Be Testing. Those are the core tenets.
Marshall: You can do some what if's now but you'll need the high end tools.
How do you integrate all the different data? Are there tools to do that?
Avinash: All the vendors are moving in that direction but if they tell you that they can do it right now, they're probably lying. Now your best friend is probably Excel. He also likes Crystal Excelsis.
Marshall: The more important part is the analyst, not the tools. Spend the money on the analyst, that's where the value is. Data is just data.
Matt: That comes up frequently, why are all the tracking data different? The question isn't why are you more accurate, it's why are you less inaccurate than everyone else. Don't take the best numbers out of which ever tools, that's a no no. You're looking at the trends, not the hard numbers.
How do you measure lift from dynamic bidding?
Jim: Keep working at it and when you perfect it let me know and I'll write another book. [hee]
Avinash: It's difficult to model human behavior that well. Dynamic bidding works more efficiently for long tail term. Head terms are much harder to model.
Jim: The long tail is where your most valuable traffic is anyway. When they're specific they're buying.
How do you find the why?
Avinash: the only way to find the why behind the what is qualitative feedback. Even something as simple as a four question survey is helpful. "Why are you here?" "where you able to complete the task" "Where did you have trouble?" [that's only three. I don't know where the other question went.]
Marshall: Didn't Google Analytics integrate something with question data?
Jim: You need to find why they went there. Then how they feel. Then what did they do--what was the outcome. Finally, you need competitive intelligence.
Matt: That's where the convergence of usability and analytics come in to me. I get more out of observing them. We can see where the hesitation is and back it up with the data.
Microsoft intregrated spreadsheets, will Google?
Avinash: The Microsoft AdCenter plug in is great. I'm very fond of it. The reason you keep using excel is that the tools available aren't that great at doing the segmentation you need. Google docs are being integrated into G Analytics.
Matt: Excel is my most favorite tool. It's fantastic stuff. You can't present analytics in power point. Excel allows you to display the relationships.
Jim: Every major tool exports to Excel because that's where the power is.
Marshall: Everyone uses Excel.
So we have to ask the right questions and get the answer but now what? How do I convince someone who is a manager looking at the Omniture data who only sees search views are going down and doesn't care that the page views are going up?
Marshall: (For Monster:) Search traffic comes in peaks and valleys. It's driven by spend and inventory. But the traffic continues. You need to understand the relationship. Fine tune it. You'll have to move up the food chain.
Avinash: Blog post: "Seven Tips for creating a data driven boss". There's a great report in Google Analytics: Visitor Loyalty and Recency. Measures not only what happened in the visit but across the visits and the distance between the visits. Segment first, then go in and see where these people are and where they're coming from and how often they're coming from. You can also survey the customers.
Marshall: All web analytics is really proving hypotheses. You might need to do more. Do comparison. Figure out what the right question is.
Matt: I think you'll see that the two points of data you have isn't enough. You need more. You need to go in and do more. In order to clarify a chart, add more information.
Jim: It also depends on goal. If the goal is just 'bring more traffic to the Web site through search' that's an artificial goal. You need to figure out the business goal behind it.
Marshall: Search has to go all the way up the food chain. If you're doing optimization and measuring afterwards, you don't have enough control. You need to base your deliverable on things that you can control.
Posted at August 19, 2008 12:14 PM
View related entries in: Analytics, Liveblog, SEO, sessj2008
Stumble It!
Add to Delicious
