Wondrous post over at one of the planet’s smartest blogs – Occam’s Razor – by Google’s Avinash Kaushik on how to, and how not to measure social media ROI (incremental profit minus costs incurred).
Worth reading – but in a nutshell, Avinash takes Facebook to task for misrepresenting ROI on a Burberry campaign.
Brand promotes new product on Facebook. Sales go up. Therefore Facebook marketing works. Not.
The error happens in confusing correlation with causation. Take this example. Man takes contraceptive pill. Man does not get pregnant. Therefore pill is effective in preventing male pregnancies. Nope.
Any number of uncontrolled variables could have accounted for Burberry’s rise in sales – a sales push, increased distribution and availability, an above-the-line campaign. Just because sales happened to go up around the time you were doing social media marketing does not mean your social media marketing is effective. It’s true that your risk of error can be lowered with social commerce because a digital trail exists linking social media to sales, but complex attribution models notwithstanding, you’ll always be at the mercy of intervening variables.
The simple solution is, of course – and as Avinash points out – to control for other variables when you are active in social media by setting up a control region or market where the rest of your sales and marketing mix remains identical, except for social media efforts. Contrast the sales from the campaign and control areas to identify the ROI of social media.
The post ends with a pertinent recommendation and an incisive insight – both of which we throughly endorse:
- Don’t have a social strategy: create products and services that compel social activity. Ultimately social is not just about how social your company is. It is about how many social ripples your products/services create.
- Social media success does not guarantee business success. Dippin’ Dots crossed five million Facebook Fans a couple of days before they filed for bankruptcy. Pepsi outperforms Coke on social media activity, but continues to lose share.
Calculating ROI for social media campaign is really an hardest task. There no method to determine the exact ROI for social media campaign.
Thanks Ed for the comment. I agree, but if social media is to survive it must prove that the incremental profit it generates exceeds the investment made. It’s basic business sense. ROI is not difficult to calculate – what is more difficult is getting to the numbers needed – investment made, and profit generated. Controlled campaigns are one of the simplest ways of doing this.
Paul, agreed. Many brands have a problem justifying budget for marketing that doesn’t lead to ROI. As you said, if social media wants to survive it has to show ROI.
“If social media is to survive it has to show ROI.” ?? That statement is almost as obviously bad as the joke about the pill preventing male pregnancy that opens this article.
Social media will continue whether or not the marketers can show ROI. It is the service providers making their living helping advertisers with social media that would be endangered.
Social media was around before the brands got so invested, and if the brands pull away their investment there will still be social media.
Great post.Any updates coming soon?
[…] and capturing value with social media is not how about how social your company is, but about how many social ripples your products/services create, then how do you create social […]
The company that I work for has made it our business for 3+ Years to deliver ROI to businesses about their social media efforts. We drive consumers from Facebook and into your stores and deliver back BIG DATA about those consumers from their campaign. I don’t want this to be a commercial but I know for a fact it is possible. We just ran a campaign with Lindt and it was a huge success.
Typically, viewers invest in subscriptions to popular channels for such content, which can be costly. However, this app provides access to all content for free, making it a cost-effective solution for entertainment needs.