Deep Dive Into Analytics: When Bounce Rate No Longer Floats Your Boat
Jeffrey Eisenberg, managing partner, BryanEisenberg.com
I wonder how deep this dive will go. If you’re in the front row (that includes you!) prepare to be splashed. Jeffrey asks if everyone brought their snorkel!
David takes the podium first. He often get asked what is deep dive analytics?
Most tools offer data available on the surface. Deep analytics casts a wider net and looks at the Web as a whole.
What’s the process?
- Create a hypothesis
- Define your scope
- Pull data
- Follow the crumb trail
- Translate into action
Case Study: Geo Insights for PPC
Business question: How should I change my market budget allocation based on geographic performance?
Objective: Uncover high performance niche markets and use to manage PPC campaigns
Focus geography deep dive on:
- Keyword cluster performance
- Messaging by geography
- Target audience density
Notice that there’s not much activity in the top-right quadrant. That represents untapped opportunity. The top left quadrant is working well, so maximize spend and investigate other platforms that would apply. Remember that cluster analysis doesn’t apply across regions. That cluster was returned for Europe but was very different in the U.S.
So What is the Action?
For each geography/cluster combination:
- Focus on best-performing ad messaging
- Tailor landing pages to reflect the best messaging
- Geo-target high-performing areas and work to improve results
Applying these insights they saw:
- 20 percent more revenue with the same amount of spend
- Considerations: PPC campaign architecture is critical to applying insights
Conversion Funnel Analysis
A hotel site had a seven-page conversion funnel. At one point the user picks a package. They identified those highly converting packages and analyzed what was attractive about the package. They also looked at where people were leaving in the funnel. They saw a lot of people are moving backwards in the funnel to see a previous page. What was the difference between the pages? The page they were going back to had lots of images, so they added thumbnails to the page where people were ditching. They also added a large picture on the last page before they click and book.
Applying these insights they saw:
- 20-40 percent increase in continuation rate on optimized pages
- Considerations: Testing takes longer but can increase impact
Look out for:
- Sample size
- Major site changes
- Inaccurate/incomplete tracking
- Change in marketing spend
- Changes in search inventory
Other popular deep dive techniques:
- Geo segmentation
- High value visitors
- Content area
- Product group
- Search engine
- Asset interaction
- Movement over time
- Largest percent movers
- Pathing patterns in behavior
Jeff is up next and he’s a marketer. This will be his perspective of what’s important in Web analytics.
The Ugly Truth
In Web analytics, everyone has different goals/tools/technologies. There are at least two truths. Two ugly truths:
- Implementation is touch to get right and often messed up.
- You probably have bad breath in the morning.
That last one was a joke but not many laugh. Methinks it’s late in the day.
Top 5 Implementation Errors
- Campaign tracking: Make sure you’ve got the URLs correct
- Page tagging: Maybe something’s going wrong when it’s implemented
- Goals: Can everyone on the team agree? Dialog during implementation
- E-commerce implementation
- Data validation
5 of Jeff’s Favorite Things in Web Analytics
1. Balancing the team with conversions and assists. The last click is often counted as the conversion driver but earlier influencers should be attributed.
2. What is Google Search Funnel? It’s a brand new tool that’s going to show what happens before the last click. AdWords now stitches together what happened, working back from a conversion. It’s generated through Google AdWords conversion tracking and has nothing to do with Google Analytics.
3. The no brainers. Multivariate testing: through testing, you can show the revenue generated before and after. Search Terms report
4. Advanced table filters. Chain together multiple conditions and rules so that sorting is more advanced.
5. Analytics intelligence. Automatically detects and surfaces significant changes in your key performance metrics that you should be aware of.
Next Christopher takes the mic. “It’s not about the data. It’s never been about the data. It’s about what you do with the data.” – Josh James, founder of Omniture
First step: Web analytics and online advertising
Impression > Click > ??? > Conversion
- Understanding a site’s micro-conversions:
- Product views
- Cart adds
- Cart removals
- Cost per step metrics (cost per product view)
- Step to step conversion rates (click to cart ratio)
The funnel will help you hone in on what’s really happening on your site. How granular can you get with your integration?
Next step: “offline data” in online marketing
Traditional SEM lead generation:
- Cost per lead
- Estimating target CPL
The lead funnel KPIs are mapped to imported CRM data.
- CRM closed leads
- Call center orders
Integrate with ERP systems:
- Closed leads
- Cost per closed lead
- Down to the keyword!
Are you targeting the right metrics? Here’s a case study using Vintage Tub & Bath.
- Challenge 1: limited understanding of how online advertising spend impacted actual revenue
- Why? Acquisition vs. offline conversion: Customers start online and finish sale offline by phone
- Challenge 2: more time spent digging for data, less time acting on it
- Why? “Spreadsheets of death” scenario
- automated integration with call center and their web analytics and bid management systems
- Cost of goods sold (COGS) data feed into Web analytics and bid management system
- Automated bidding decisions based on better data
- Use your unique “deep metrics” in bid logic
- Easily switch metrics and goals over time
Does the Search Funnel tool carry over cross-channel? Is it post click or first impression?
Gillis: Right now it’s only Google.com search ads. It goes back 30 days, counting impressions.
What can we do with the new Search Funnel tool?
Gillis: Right now you’ll see that a certain keyword has a high conversion rate and you can follow that though to see what value you’re getting back. Now you can attribute value ot the keywords that came before that. You can experiment then to see if certain keywords are helping in an assist way. It’s a ton more data for you to look at when you measure the cost of a conversion. It eliminates the guesswork if a keyword had any affect on your campaign.
Knoch: Avinash explained at SES Chicago that most attribution models are junk. First click and last click isn’t all that matters. Search Funnel gives you a better idea of your conversion funnel.
How will Google allowing users to opt out of Google Analytics affect the data?
Gillis: We have a feeling it won’t affect the data because generally people aren’t going to opt out. But we did want to offer the option. There have already been ways that people can opt out through cookies and such.
Is the Search Funnel team communicating with Conversion Optimizer?
Gillis: It’s currently just works through AdWords, though you can import goals from Conversion Optimizer.