10:13am Check out http://tr.im/mmsc4 for more information on all of this!
10:12am Two ideas for what the future might look like. 1. Media mix modeling. Run experiments. Target geo-locations representative of your overall market. Test various mixes. Then use the data to inform your spend.
2. Marginal attribution analysis. It’s very simple but takes a bit longer. Gives example of spending on Baidu and Bing. What happens if you add a third campaign – incrementally?
10:09am The whole system of assigning credit based on an arbitrary set of rules won’t make you great at web marketing. Ask a different question. What is the mix of media that produces the highest amount of ROI?
10:07am Advises MbD model. Last click 75% (but could be lower/higher depending on your company). We *know* the last click converted. Remaining 25%, deceasing function back 3-5 clicks. All within last 7-30 days.
Impulse – back 7 days
More impulse to less considered – 7-14 days back
You don’t have to use this model – give great thought about what’s happening in your campaigns.
10:05am Credit/decay custom. It’s a “less worse” model. Uses days to conversion model.
10:04am Giving equal weight to all clicks on a conversion path doesn’t work either, because you don’t really know which campaign really worked.
Then, there’s the “Make Crap Up” model. 20% for email – the woman who designed the plan said it’s because she has a large team working on email. That, of course, has nothing to do with what’s really creating conversions. Another 20% for Bing organic “because I really want Bing to win against Google.” Nice sentiment, but doesn’t reflect reality.
10:00am Giving all of the credit to the first click is “like giving all of the credit for marrying my wife to my first girlfriend.”
9:57am The “problem” of attribution. Don’t generalize. Look at your conversion numbers and see if it’s a problem for you. If most visitors are converting on the first 1-2 visits, then you don’t have the problem. If it’s not as clear, break it down by campaign. Narrow down the root of the problem instead of saying you have a general attribution problem and then worrying about all campaigns.
9:53am Shows use of Google search-based keyword tool, which can take a site and make keyword suggestions – and which page on your site they’re associated with. You can set CPC filters – to realize the long tail. There can be thousands of words for $0.10 a click. Take *that* data to your boss.
9:45am Gives example after example of keywords – some very long tail ones copied straight from the site – that big camera brand Kodak doesn’t rank or advertise with paid search for. Same for other brands including Orbitz, Cisco, and Victoria’s Secret.
People complain about the CPC of “digital camera” but there are thousands of long tail keywords on the cheap.
9:41am People in the long tail don’t know what they want. “If I can get in front of them first, I can convince them.”
9:38am The long tail. Kaushik got 88,000 visitors to his site a couple months ago. Roughly half came from search – via 26,000 keywords. He *only* writes about analytics. 7 times more than the keywords he was obsessed about. This is the story of the long tail.
9:26am Story: Kaushik tries to convince his wife he wants to blog. She tells him to go to bed after a long day and he says he needs to stay up and blog. (This sounds like conversations between and my software developer husband.)
First reason he uses to convince her is “I’m kind of a big deal.” And then, of course, shows her how people all over the world are reading his blog.
She says “Go to bed.”
He says the numbers come across as meaningless – and it’s the same at many companies.
Next reason – conversions.
“It’s hard for me to figure out the point of my blog. So, I created four goals.” (What’s up with the number four?)
So he shows his wife his success of his goals. “I’ve made $26,210 in a month.” And he has a 5.24% conversion rate.
She says, “Work harder.”
We obsess about the 2% of people at our site – but we should focus about the 98% not coming to the site.
9:25am Talks about excuses people have about their web strategies. Points out that the focus is on crap instead of opportunities.
9:23am “It’s very easy to see how you suck a lot [from patterns in your keywords].” LOL. hilarious. Because it’s true.
9:21am Shows tag cloud surrounding hotel chain, but it’s full of cities that people already know about. Tag cloud can show what’s missing from your strategy.
9:19am Tag clouds are a fantastic way of visualizing tons of data. Use wordle.
9:17am Keyword research – 340 keywords, make $5.01, but “These are keywords I should be making love to.”
“Reduce massive glob of data to something manageable.”
9:15am Kaushik says he’ll tell us four stories. He’s our sensei, our guru – and a well-dressed, super-smart, funny one at that. And he donates the profits of his book to charity.
9:13 am Would people pay for a book when they can get the information for free on a blog? “What I found out was that people pay what they can get for free all around the world.” (Displays pictures of his book in different languages.)
9:12am Avinash Kaushik, supreme analytics dude, takes the stage to tell us how it is.
9:06am Mike Grehan takes the stage to introduce the session. Briefing the crowd on what’s going on today.
8:57am Black Eyed Peas’ Tonight’s Gonna Be a Good Night song is playing. Most overplayed song of the universe, is attempting to warm the crowd up pre-keynote. You’re gonna have to play some Them Crooked Vultures, Paramore or – what I listened to this morning – The Temper Trap – to accomplish that with me.