Jay Goldman

• Avinash is the Evangelist for Google Analytics and writes the Occam’s Razor blog
• Web Analytics in An Hour a Day
  • Wiley called a few months into writing the blog to turn it into a book
  • “People like paying for free stuff”
  • Donate all the (meagre) proceeds to DWB and the Smile Train
  • About $60k to date
• Accountability
  • The web has infinite amounts of accountability
  • Didn’t learn much from President Bush, other than “faith-based initiatives”
  • Magazine ads are faith-based initiatives
    • Can’t measure actual impressions other than subscribers and news stand sales
  • Web advertising isn’t at all because you can measure it
    • Number of clicks
    • Relevance (shown along with search results for the term)
    • Measure effectiveness based on subsequent activities on the clicked-through site
• The web gives you more data than God intended
  • Very good at giving you “The What”
    • You can spend your life doing path analysis and end up with no real conclusion
  • We don’t spend as much time analyzing “The How Much” or “The Why”
    • Clickstream: The What
    • Multiple Outcomes Analysis: The How Much
    • Experimentation and Testing, Voice of Customer: The Why
    • Competitive Intelligence, Insights: The Goldmine!
  • Not in the business of generating pageviews — in the business of generating revenue for your business (Clarity of Purpose!)
• HITS: How Idiots Track Success
• Bounce Rate
  • Not an aggregate — an actual measure of behaviour
  • Translates into: “I came, I puked, I left”
  • Very easy to understand: 70% = suck. 30% = good.
  • 20% is a fantastic rate (you always get traffic that’s not relevant to you from search engines)
  • Only seen less than 20 on high-end video porn sites, and the best still had a 4% rate (you made it all the way here and left?!)
  • Look at traffic sources and click the bar chart to compare sources to bounce rate to find your BFFs
  • Magazines are in trouble because they follow the “Selfish Lover Strategy”
    • Their ads take precedence over your content
    • Not the New Yorker, which is almost all content
    • Bounce rates are reflective of this — more ads and nav, higher bounce rate
• We think our site traffic is all the same person coming, but it’s really a huge mess of different people with different goals
  • Segmentation and deep diving gives us the right view
  • Pageviews, etc. are all aggregated data, no segmented
  • How to improve the Reader’s Digest Advice and Know How page?
    • ~320 links on a single page
    • 10,158,033 pageviews, 9 average pageviews
    • Look at the Depth of Visit stats though — 33% of people only looked at the homepage
    • Go into segmentation and create a segment of people who’s depth was greater than four pages
    • Use that segment throughout to figure out who is driving those people to the site, which keywords, what they looked for
  • Analyze the amount of content on your site made up by each section then compare it to the total percentage of visits for it
    • Might discover that you’re spending a lot of time and money to produce content that people aren’t as interested in
• Ecommerce
  • Average conversion rate for top ecommerce websites was 1.72%
    • What happened to the other 98%?
    • They might be offline conversions, tire kickers, etc.
  • Measure the conversions on your site for actual revenue, but also everything else
    • These are ‘macro-conversions’
    • What else does your website do to drive sales?
      • Create profiles
      • Research
      • Coupons
      • These are ‘micro-conversions’
  • Example: Avinash’s own blog
    • Set four goals: all posts, about, speaking engagements, RSS subscribers
    • “If you read two posts I’ve written, you will be convinced of my greatness”, so driving people to the “All Posts” page means they will probably read two posts
    • The last is a macro-conversion, the other three are micro
    • RSS is the ultimate form of permission marketing — I can push “any damn thing I want” to you with your permission
      • Can measure the value of that by finding out how much it would cost to purchase them as leads
    • Speaking engagements value can be measured by asking organizers how much traffic was driven to them and who registered
    • Can now measure the value in ‘fake dollars’ every month
• How do you measure conversions on something like a drug website where you can’t actually sell the product?
  • You can measure things like watching videos, using calculators, surveys, etc.
• Facebook measuring success by the number of active users is stupid
  • They should measure engagement
  • Look at Visitor Loyalty
    • Reconstructed data: most users came once in last 30 days, but a crazy high percentage came between 9 and 200 times
  • Look at Visitor Recency
    • Reconstructed data: 66% of people were on the site between now and 0 days ago
  • Loyalty + Recency = Profits
• You should listen to the voice of your customers
  • How do you do it?
    • Not by looking at top content viewed — doesn’t show what they wanted but couldn’t find
    • There’s no pageview if I can’t find what I want
  • Surveys to the rescue!
    • Not like NPR: 35 question survey in a giant scrolling page
    • Greatest survey in the world has 3 questions:
      • Why are you here? -> Primary Purpose
      • Did you complete your task? -> Task Completion Rate
      • If you were not able to complete the task, why not? -> Segments of Discontent
      • Compare Primary Purpose and Task Completion to find low hanging fruit
      • Look at distribution of data to find out how many people fit into each category
        • Probably a small percentage (e.g.: mid-30s) who are there for your primary purpose
    • Perceptions, Montreal-company, created an exit survey with 4 questions to do this
    • Five Second Test: simple usability tests
    • Ethnio: remote conversations with people who are on your website right now
• Most websites suck today because they’re created by HIPPOs (Highest Paid Person’s Opinion)
• One of the greatest gifts the Internet gives us is the ability to learn to be wrong quickly
  • Magazines or television ads take a long, long time to produce the content and learn from it
• A/B testing is great
  • Takes six and a half minutes to launch a test in Google Web Optimizer
  • iPhone has shown us that you can sell an object of lust to a CEO and she will force her IT staff to support it so that hers works
    • BlackBerry Storm page is horrible and creates no lust
    • Palm Pre is all about creating lust
    • Do an A/B test! You’ll quickly see that you can measure passion
• Competitive Intelligence
  • Google Ad Planner lets you set up a very specific demographic then analyze which sites they go to
    • Can add sites those people have also visited
    • Site Profiles let you analyze the traffic and stats to specific domains
• Questions:
  • We do a lot of work in the public sector and there’s sometimes resistance to accountability. How can public sector organizations with no bottom line tied their site use analytics?
    • The Obama administration is the best thing that could have happened to the web from an analytics perspective. Sometimes it takes a generational shift to get the old people out and new people in. Worked with a security agency to measure how many people were downloading things and found that it was 19 clicks to get to a download. Blatantly embarrassing them is a great way to get them to make changes. This is an agency America doesn’t acknowledge exists but they have a website with a Kid’s Club! Their bounce rate was 96% when we started looking. Use tools like compete.com to look up share of traffic you get for top terms (e.g.: “enterprise software” for large software companies -> top entry got 40m visits, being 44th got 3,976).

Likely-related posts:

• WSS09: Jeremiah Oywang on the Future of the Social Web Notes from Jeremiah Oywang on the Future of Social Media...
• Social Media Strategies for Organizations My presentation from Web Strategy Summit, delivered May 4th 2009....