Every time a case study boasts about the number of Likes and Tweets it earned, an angel loses its wings. Hate to say it, but the culture of social media metrics is broken, and it's time to fix that. There's a reason I don't track Facebook Likes, and I feel our other industry standard metrics are equally good at failing to understand why people use social networks.
We need to change what gets measured and why. It's time to stop managing the wrong things. Here's how.
1. Start Measuring the Impact of Impressions
We've already discussed just how surprisingly weak the link is between clicks and sales when compared to the link between impressions and sales, and how to measure it. We've also talked about how Facebook Exchange in the feed could be a great way to capitalize on this fact. But this subject is so important to me that I felt the need to repeat it yet again.
I'm not saying you should stop tracking clicks. It's just that you're already tracking them. You're probably even tracking impressions on social networks. What you most likely aren't doing, though, is tracking how impressions are impacting business.
I know it doesn't come easy. I know wrapping your head around a fuzzy concept like "correlation" is uncomfortable for some. I know that separating cause and effect doesn't come easy either. But when Facebook argues that 99 percent of the sales they are responsible for come from impressions, not clicks, it's difficult to justify the money you're leaving on the table by ignoring it.
The thing is, not all impressions are the same. It's not enough to just optimize for impressions. Some impressions will create more sales than others. Some will contribute more to customer retention than others. Some will generate more viral activity than others.
I recognize that you can't split test social media impressions the way you can a PPC campaign. If the words "regression analysis," "confidence interval," and "P-value" are foreign to you, you're not alone. But I assure you, there are statistical methods available to identify which types of impressions earn the most sales and retain the most customers.
Get a statistician onboard. It's worth the trouble.
And if correlations are too abstract and don't feel real enough for you, start surveying your customers. Yes, you'll be measuring what they say, not what they do, but it's better than nothing.
By the way, I'd advise taking a similar approach with brand mentions. Since it's impossible (as far as I know) to measure the number of impressions you're getting from each brand mention, this data is a weaker starting point than raw impression figures, but it's still worth measuring its impact on sales.
2. Start Measuring Retention
If social media is about building relationships, as is so often claimed, then retention should be top priority. You need to measure how your social media efforts are contributing to repeat visits and repeat sales, or you have no way of knowing that social media is having any effect on your relationship with your customers.
We've discussed before how to measure repeat visits from Facebook, at least in the context of repeat clicks from Facebook. But, as should be obvious from the section above, clicks aren't everything, and you need to go further in order to truly measure how many repeat visitors you're getting.
To do this properly, you really need to get first touch attribution set up in Google Analytics. Will Critchlow has already written a great article on the subject, so I won't go into the details here. The important thing to understand is that you want to know how many people who were referred from social networks came back, and how often.
Keep in mind that a large portion of your direct traffic may actually consist of referrals from social networks. My estimate, based on correlations, was that about half of the referrals from social networks actually came in as direct traffic. Again, this is why it's important to measure the impact of impressions. While it's impossible to attribute a specific direct visitor to social networks, you can estimate how much of your direct traffic is the result of social activity, and project how many repeat visits this leads to.
These repeat visits aren't the only way to measure retention, of course. Repeat sales, even if they're made offline, are incredibly important. The only way to get first touch attribution on a repeat offline sale is to survey your customers. You can incentivize this with rewards. It's best if you just take a small sample and offer a reward high enough that you get a good response rate. If too many of them don't respond, it can skew the results.
3. Only Measure What Matters
This may sound like a bizarre thing to say after arguing that you should hire a statistician and look for correlations in complex data sets, but I'm sort of with Tim Ferris on this one. As a general (and unscientific) rule, about 20 percent of the things we do are responsible for 80 percent of the benefits. Meanwhile, another 20 percent of the things we do are responsible for wasting about 80 percent of our time.
I'm of the opinion that the basic structure of measurement should be simple:
1. Define a very small set of metrics that matter
2. Conduct exploratory experiments to discover new metrics
3. Toss out new metrics unless you find a strong correlation with the metrics that matter
In other words, you should constantly test and try new metrics, but you shouldn't bother sticking with them unless you find out that they have a strong influence on your KPIs. Otherwise, your list of metrics will just continue to grow, and you'll eventually end up consuming too much time.
It's true that what gets measured gets managed. It's also true that if you manage too many things, you'll end up with a pile of red tape and not much else to show for it.
That's why I don't bother measuring Facebook Likes, retweets, or any of the other vanity metrics. I don't see a meaningful relationship with the metrics that matter.
What are you measuring?
Image credit: aussiegall