Howard Greenstein is a marketing technology strategist and president of the Harbrooke Group.
For many marketers, Google Analytics is the primary insights package for the web. They track simple things like most popular content, which pages people stay on the longest and what sites are referring visitors. With a little more sophistication, you can track how users get to your site and what path they take to convert to a sale or sign up.
For our new Social Media Measurement Guide, we asked Google’s Analytics Evangelist Justin Cutroni to explain some important Google Analytics functions to generate hard data from your social interactions with customers.
Here, we look at attribution modeling: a very effective way to understand what actions in different networks affect your customer’s behavior.
Attribution modeling allows you to experiment and see which interaction points lead to behaviors like conversions and sales. It is one of the most powerful features in Google Analytics.
Cutroni notes that “Multi-Channel Funnels” are a great introduction into attribution modeling. Multi-Channel Funnels let you see the role that prior site referrals, searches and ads play in bringing a visitor towards conversion. They also let you see how much time has passed between the first interaction and a purchase or another decision.
Justin Cutroni
Funnels help you identify all the different marketing channels you’re using and what Cutroni calls “Exposers and Closers.” An exposer is more upper-funnel and generates a lot of traffic, but not a lot of conversions.
For example, exposers may see a YouTube video with your product or service and that may bring them to your website, but they may not buy at that moment. A closer is more bottom of the funnel, generating conver- sions and revenue, such as a coupon. Social can be an exposer or a closer depending on how it is used.
Tweeting “Check out our new offerings” vs. “20% off when you click now” can have different results. Create the custom URLs so you can track how those activities come through.
You can customize a channel in Google Analytics; for example, “Came from Social with a custom URL value of X.” Then, you can see if these activities are achieving their goals with attribution modeling.
With attribution modeling, you can understand the return you get from each channel. The function lets you assign different values to different aspects of your multi-channel funnel and then experiment with weighting the values.
This is an excerpt from our new Social Media Measurement guide. Click here to download!
Cutroni talks about a model that gives all the credit to the last click in a channel. This model cannot account for the other interactions someone may have had with your site, your email campaign or your social presences. A time decay model (where several clicks closest in time to the sale or conversion get the credit) might be more appropriate, but you won’t know it until you test it.
“When you’re getting into attribution modeling, start with a basic model. For most businesses in e-commerce, a decay model is a good one to use. Compare a decay model with the last click model (you can have columns in the report) to see metrics and look for places where the models show very different values for the channel.”
“For a last click model, you may see social having a low value, but in a decay model, a very high value. That tells you the channel may not be good at closing, but it does expose people to your offering.We’re trying to monetize the funnel and see how much each item is worth. Then, the whole idea is you need to change something. Identify a place where there may be more value, like in social, and change some- thing. Try investing more time or money there to see if you get more conversions.”
Attribution modeling works only when you vary the conditions and see where the value occurs.
Can you track social media activity back to your offline store?
Research shows that networks like Facebook and Pinterest are influencing consumer behavior in larger stores. According to a report in the Harvard Business Review, “Pinterest is an especially popular driver of in-store sales: 21 percent of the Pinterest users we surveyed said that they bought an item in-store after pinning, repinning, or ‘liking’ it, and 36 percent of users under 35 said they had done so.”
Further data from a more extensive study by the same authors shows that Facebook is most likely driving customers to purchase – one in three Facebook users has purchased something after sharing, liking or commenting. Overall, 40 percent of social media users have purchased an item after sharing or favoriting it on a network like Twitter, Pinterest or Facebook.
Click to get the guide (short form required)
This August, Twitter announced a partnership with DataLogix to measure the offline sales impact of promoted and organic Tweets sent by consumer packaged goods compa- nies.18 The report notes, “Users who engaged with a brand’s Promoted Tweets purchased more from that brand than a statistically identical control group, resulting in a 12 percent average sales lift.”
There are ways to track online actions to in-store purchasing. None are guaranteed, but here are a few suggestions.
- Track views on items in your online catalog to see if viewed items are not purchased online but are moving in significantly greater volume in-store.
- Offer online site viewers a way to “print a shopping list” to take in-store with a tracking or coupon code.
- Use loyalty cards or having customers sign up for an email list, to tie their online identity to in-store activity. This might require some initial manual work, but could yield valuable results.
For companies who have not yet done so, one of the most valuable quick and easy things to implement would be Google link tracking. After setting up Google Analytics, start using the program to track the custom links you’ve created. This should immediately help track content performance on different networks with an even finer grain.
For those businesses that are already tracking, take the time to create or modify your funnels or attribution models. Make and test different assumptions and see how the results affect your KPIs. While this requires more effort and research, it can truly start to impact your bottom line pretty quickly.
