Decisions I’ve Made Using Analytics and/or Data

An IBM study showed that professionals that tended to make data based decisions were exponentially more successful than those that made decisions based on intuition. The following are a few successful decisions I’ve made in my career based on analytics. I included an unsuccessful decision I made based on intuition to support the data-based conclusion of IBM in the study. 

Image

Google Analytics, Social Monitoring Tools and Salesforce.com- The Web IQ Quiz (www.webiqquiz.com)

 Content/UX Design:

 After deciding a scoreboard would increase traffic on a quiz I was helping build for IQ Agency we decided scores should come in around 70% correct. This was an intuitive decision because data didn’t exist to solve the problem. To prove our intuitive decision we had a few dozen, exclusive ad agency professionals take the quiz and found the percent correct was far too low. We removed some difficult questions and made others easier. In a test with a new group we reached our 70% range we had hoped for. The appropriate design of difficulty contributed to the virality of the quiz, as many users would brag about their scores and challenge others. Had the elite users (also high influencers) scored less than 80% they would have likely been embarrassed to post their score which would have lost the project tens of thousands of page views.

Promo campaign/launch:

Allocating my efforts

In monitoring a variety of data I was able to make quick decisions as my marketing campaign unfolded day by day. In the first few days Google Analytics data showed sharpest increases coming from tweets posted by web celebrities I had pitched the project to which caused me to allocate more of my time to reaching celebrities. Within a few days of social monitoring on Social Mention, Vistamix and Radian6 I realized the quiz was spreading through celebrities organically, posted by dozens of celebrities (30k+ followers) a day. Google Analytics showed that sites such as 20kaido.com in Japan had picked up and featured the quiz on their publications as a result of celebrity posting on Twitter. Because of this I decided to allocate more efforts on media pitches to publications such as CMO.com and Egotists in multiple cities. When publications such as these did feature the quiz, I saw a huge jump in traffic. Eventually, even publication features became organic, from that point I started submitting the quiz to content sites such as StumbleUpon. Three weeks after launch, Twitter went from generating 80% of traffic to 2%. As that channel was saturated with posts regarding the quiz I used data to further drive traffic via new channels. I was also able to monitor traffic jumps based on times of the day to chose when to post new content and prize promotions with the Windows Phone 7’s we acquired through a sponsorship we pitched to Microsoft.  

Email Marketing using SalesForce.com Mass Email and other tools:

Through a connection from our CEO, we acquired an email list of 60,000 brand managers. Rather than trusting a subject line we thought would be engaging enough for high click through percentage, we decided to perform a test sample of 3,000, emailing three subject lines, one to each thousand. I say “we,” because I collaborated with IQ’s marketing director on this process. Subject lines were (1) ‘The 20-year-old you check out every day’ (2) ‘Celebrate the Web’s 20th Birthday’ and (3) ‘I thought you might find this interesting,’ listed in order of what we thought would be the most successful. Ironically, #3 showed 7% click-through with the best soft bounce rate. Good thing we hadn’t trusted our instincts! Average email yields in marketing are 3% or less. The most vague subject line we had come up with, which had originally be designed as a constant variable actually became our most successful subject line.  

Google Analytics Tools Utilized

Measure Traffic and Top Content

130,000 Unique page views in three weeks, avg. dwell time of 5 min, 20% bounce rate

Map Overlay, Goals and Funnels

Conversion rates by country and state

Mined data to build info-graphics to prolong life of the product story, newsworthiness 

Traffic Sources

5% Organic Google Search- SEO not a worthwhile investment

Highest traffic came for first two days from Twitter and email

Most significant traffic jumps initially were from celebrity Tweets like Robert Scoble

Then from publications- Egotists, CSS-Tricks, Kaido, CMO.com

Then from content sharing sites like StumbleUpon and Reddit

Intelligence

Quick summaries of change in traffic

Advanced Segments

People that find it through a certain channel from a certain region

*What I will use to understand differences in behaviors of global traffic to DAA products

Ticket Leap Event Marketing Analytics- Cache Valley Clash

Choosing advertising channels based on previous returns

After planning and promoting a sporting event to which 1,300 attended, I analyzed data from online ticket sales showing where customers heard about the event. I organized the lists in excel and using pie graphs, was able to determine that the most effective channels in order were (1) WOM Referral (2) Radio (3) Fliers (4) Facebook and (5) Posters among others. Then I divided the number of introductions per channel by dollars spent on those channels. Radio, at $1,300 was the most expensive. Without analyzing data I would have intuitively invested more money for the next event in radio, however, by determining reach per dollar I found that radio was actually one of the least efficient channels on which to advertise per dollar. There was no data to determine what drove WOM Referral, so I intuitively decided a combination of many channels was important to reach a variety of people. I invested more funds into print and Facebook advertising. 

Estimating residential painting projects- Bountiful Painting

Mistakenly continuing to out-price other bids

By lowering bids on residential paint estimates in Bountiful, Utah, I was able to triple my booking rate in comparison to the local industry standard. Rather than analyzing profits on past projects, I made an intuitive decision to under bid projects, assuming I would find ways to cut costs, labor, and somehow turn acceptable profits. As I continued to ignore data readily available to me, I continued in minimal to neutral profits for a few weeks. Eventually, I realized the pattern I had established and based on that data, chose to start bidding high prices for projects. My booking rate dramatically fell, but my profits tripled and I was able to recover my losses and still gain satisfactory profit. Had I analyzed data sooner, I likely would have tripled profits for my efforts much sooner

Advertisements

One thought on “Decisions I’ve Made Using Analytics and/or Data

  1. Ambrose says:

    Hi there to every body, it’s my first go to see of this webpage;
    this web site contains remarkable and in fact excellent material for readers.

Intelligent sincerity below

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: