Actionable Organic Search Analytics
We’re back in the search track with Matt Bailey (SiteLogic) and Diane Hoag (IBM developerWorks). Matt is giving me a hard time for going to bed early last night and missing out on all the debauchery that went down after I left. Based on how tired Matt looks, methinks I made the right decision. :)
Mikes moderating and starts off by telling the story about his stick. If you know Mike, you know what I’m talking about. Basically he takes it everywhere and it’s been around the world. If you haven’t heard the story, ask Mike the next time you see him. He loves to tell it.
Up first is Matt Bailey.
Matt says conferences like this are the ‘dump truck conferences’. You come, the conference dumps all this information into your brain, and then you have to go back and apply it.
He says that Plato talked about three different groups. The first were the gods and they already knew everything. The second were the beasts. The beasts didn’t know anything. They didn’t ask questions. The third group was the people. People are the only ones who can ask questions. We’re the only ones with the fundamental ability to ask.
[I know I say this all the time, but I love listening to Matt Bailey speak. He’s so animated. He gets people excited!]
We can’t look at our dashboards of information and expect them to tell us what to do. They don’t function that way. We have to pull information out of the machine data and interpret it.
You start with the data – 34,000. There’s nothing you can do with just the number. You have to add a bit of information and context. There’s no such thing as complete and true accuracy. If your supervisor is asking for accuracy, you need to let them know it’s impossible. Instead, you have to look at the trends and gather the information that’s happening on your Web site. Adding context to the equation doesn’t make things actionable. What makes things actionable are when you can create a story. Visitors who searched for X stayed on the site for X minutes and looked for Y and converted at a rate of Z.
You want to get beyond knowledge. You want to get to understanding and that comes from people. People are the central focus for adding that understanding and figuring out what to do on your Web site. No program is going to give you understanding. It can only give you information and knowledge. Avinash said that analytics are 90 percent the person and 10 percent the tool. That needs to be your mantra.
Reporting or Analysis?
Reporting just gives you information like path views, path analysis, hits, monthly visitors, etc. That’s not actionable.
You can’t do analytics on an ad hoc basis. You have to have clearly defined goals. If you don’t have goals you don’t have measurements. Your goals need to be written down. If there are no goals, there are no insights.
Your goals can be anything. They can be downloads, page views, contact form leads, sales, etc. You have to ask question of the data. That’s how you’re going to improve those numbers.
He starts talking about segmentation. It’s an essential part about beginning to ask questions about your Web site. People are not cattle. They don’t come to your site in a herd and move from point A to point B. We don’t have a herd mentality. (Matt has clearly never attended high school) People come to your site with vastly different motives. You can’t treat them all as one group of visitors.
He brings up the famous Star Trek segmentation example. I’m going to copy and paste it from our coverage of SES NY to save my fingers a bit of retyping.
The Starship enterprise had a crew of 430 people. He knows this because startrek.org told him. There were 59 total deaths in the 5 year mission. That’s a 13.7 percent mortality rate (aka conversion rate). Of the 54 deaths, yellow shirts made up 10 percent, blue shirts made up 7.2 percent and red shirts made up 72.8 percent. That’s our data but we still have no action. We have knowledge, but no indicator on how we can improve it or make it worse.
Factors that lead to a red shirt death: If you beamed down with Captain Kirk and wore a red shirt, you died 57.5 percent of the time. This is the number one factor that leads to the death of a red shirt. OMG the giggles.
To increase the survival rate: You’ll see that if Captain Kirk meets an alien woman, the red shirt survival rate increases to 84 percent.
How often do these factors occur? Capt Kirk has a conquest rate of 30 percent. If you go to a land of [insert name of things that fight Captain Kirk. I’m not geeky enough for this.] and you’re a red shirt, you’ll probably die. If you go to a land of peaceful women, you’ll live 30 percent more of the time. That’s segmentation!
Say you have an electronics Web site. You have a visitor who comes and is trying to locate a digital camera. They’re looking for price, brand, size, battery life, etc. On the other side of the site, someone is looking for an MP3 player. You need to understand that you can’t classify these two people with a single conversion rate. It doesn’t tell the story. It actually ignores the story that you have two people looking for two different things on two different sides of the Web site. Each group needs a new conversion rate. People see your Web site differently based on what they were looking for.
Three C’s of Analytics: Context, Comparison, and Contrast.
Key Performance Indicators:
- Time on site
- Pages Viewed
- Web sites
- In-market links
- Social News
Anyone who clicks on a link about you from a blog has a high context for your site. They have a specific expectation and you need to know what that is. On topical sites, you have lower context and a lot of competition. You have to be unique to stand out.
Use your analytics to tell the story. Find out where the Red Shirts are on your Web site. Find out how you can improve your site for all the different segments. Add more context to the situation. And then do something.
Forrester says that bringing in an analytics person will result in a 900-1,200 percent increase in ROI. You can’t just collect the data; you have to do something with it.
I’m just going to say that if Matt Bailey was a priest, more people would go to church. He can get you excited about pretty much anything. Seriously.
Diane Hoag is up.
Diane works for IBM and says that in 2005, out of nowhere, their Google referrals dropped off. They started missing monthly targets in June. Analysis exposed search (Google) issues. They went to the Internet (as you do…) to see what was happening and found that in May Google had released the Bourbon algorithm update. It focused on sequence and types of redirects. The intent was to discourage URL hijacking. They don’t do any blackhat SEO techniques so they didn’t understand why their traffic dropped off. They did more analysis. They finally concluded that it was due from moving to different hosting environments, which led to long redirect paths.
What resulted was that their URLs had incredibly long redirect paths. Some were 7 or 8 redirects long. Sometimes they even circled back and went through the same server a second time. They needed to clean up their redirects.
Steps to Recovery:
- Established canonical URL: www.ibm.com/developerworks
- Internally: Changed links on their content to the canonical URL
- Externally: Asked marketers to use the canonical URL
- Cleaned up their redirects: Monitored progress with HTTP simulate textbed of 33 URLs
- Made DNS changes
- Eliminated unnecessary Meta refreshes: Developed a redirect application for Web editors to use.
- Established a unified proxy with ibm.com: Eliminated redirects. The canonical URL appears in the visitor’s browser – beneficial for social bookmarking.
She shows how well IBM has recovered and that they’re growing their visitors.
Take a look at your own infrastructure and redirects. No matter how bad your situation is, recovery is possible! Aw.
Mike Grehan takes a moment to plug Mike Moran’s new book and wants to make sure it gets mentioned in the liveblogging. See, Mike, I got it.
Question and Answer
To Diane: Did Google provide any assistance when you were in this train wreck?
They got a letter from Google because they had a personal contact there. But that was it.
To Diane: What analytics were you using when you discovered the problem?
They used SurfAid, it’s an IBM product. They’re transitioning to CoreMetrics right now. She pitches SurfAid a bit.
To Matt: Did you actually watch all those Star Trek episodes? (Hee!)
Matt says he watched about half. Sci-Fi has an episode-by-episode recap and they list Red Shirts. He got his “data” from there.
For Diane: After you consolidated those URL, did you notice any changes outside of Google?
Google makes up 90 percent of their search engine referrals and has forever. They didn’t notice any changes with the other engines.