IBM Greatly Exaggerates Twitter's Ineffectiveness

IBM’s 2012 Black Friday and Cyber Monday reports said Twitter drove 0.0% of online sales over the two big shopping days during the Thanksgiving weekend. Conversely, MarketingSherpa reported that paid search had an average eCommerce conversion rate of 3.5 percent.

So does that mean social media, from a pure marketing standpoint, is dead? Should brands kill their social media programs and double down on search?

Emphatically, no. The issue has nothing to do with Twitter’s effectiveness in the selling process. Rather, it’s a classic problem of attribution. Social media’s impact goes way beyond the nearly laughable contrast illustrated by the IBM and MarketingSherpa statistics.

Marketing attribution is a major topic of conversation and debate among marketers and analytics teams today for two reasons. No. 1: CMOs are demanding performance, and attribution is critical to getting a full view of this. No. 2: The process of attribution is becoming easier and easier.

Here are two scenarios to help illustrate the concept of attribution:

1. A visitor comes to your web site from a link on a partner site. Later they see your banner and click it. The next day they come in from that same partner site again. Finally, they click on a paid search ad and register for your eNewsletter. What drove that conversion? The partner? The banner ad? Paid search? Your eNewsletter?

2. A visitor clicks to your site from a Facebook post. Later the user does a search on your brand and visits your site. The next day he finally types in your web site address before downloading your white paper. Which platform gets the credit for the lead? The post, the search, the direct link, or the white paper?

Until recently, the last click was almost always given the credit for the sale or conversion. But that is not giving the other touchpoints the credit they deserve. To solve this problem, you need attribution modeling.

With attribution modeling, we can assign weights and values to each interaction before the conversion. Then the values can be used to determine relative ROI for each activity, helping you to optimize your marketing mix.

There are many different attribution models that may be applied to determine how much each activity contributed to the ultimate conversion. Some of these models include last-click (it still has a place), first click (where awareness began), even weighting (not playing favorites) or time decay (more recent activities contribute more). Weighting can be done by platform type or countless other models that might better fit your business.

Access to tools for attribution modeling is rapidly expanding, putting powerful tools in the hands of every marketer. Recently, Google announced that its attribution modeling tool, previously only available to paying Google Analytics Premium customers, is going to be released for all users. And Facebook is now allowing advertisers to use its own tracking tools to measure off-Facebook conversions from on-site Facebook ads. These are very positive moves by two big players that help us see more accurately what is driving conversions on websites.

Some challenges still remain -- attributing view through content or ads; multiple platforms taking too much credit for contributing to conversions; and attributing non-digital activity. Still, attribution can now play a real role, which is finally getting the attribution it deserves.

As for the IBM statistics indicating Twitter’s 0.0 percent contribution to Black Friday online sales, I’ll leave you with a Mark Twain quote: “Get your facts first, then you can distort them as you please.”

Scott Chapin is a senior vice president and lead analytics analyst with Cleveland-based Marcus Thomas LLC, a $125 million integrated marketing communications agency focused on audience insight, idea generation and alternative delivery.