How does one measure Facebook?

When I first found myself beginning manage branded presences on Facebook, some of my immediate questions were:

• What do I need to evaluate for in order to measure success?
• What exactly am I evaluating?
• How am I going to do that?

The What’s:

In any marketing marketing effort you are categorically measuring for three things:

• Reach
• Engagement
• ROI

Something that I initially found interesting as I started to research social media analytics that new categories like Influence, Conversational Tone, and Viralability had entered the conversation. I’ll admit that I was initially distracted by the sheer sex appeal of these upstart categories but, after diving deeper into them, I found them to be shallow substitutes as a result of an inability to measure the primary three categories listed above.

After I reaffirmed what I was evaluating for, I needed to determine what I was evaluating. Another interesting observation I was able to make early on was that several analytics approaches placed Fan-Page centric stats from and center (page views, uniques, views by tab, etc.). Again I’ll admit that I wasted some time chasing this approach, but again it turned out to be a shallow substitute as a result of a deficiency in the practice area.

While page views, etc. may be KPIs for website analytics, they are simply inadequate for measuring a stream-based platform such as Facebook where the power of the platform lies nearly entirely in the News Feed. Case in point, what was the last time actually looked at your best friend’s Facebook profile? When was the last time you saw one of their posts? There you go- Facebook is all about individual posts going into the News Feed.

Summary: Facebook campaigns need to be evaluated for Reach, Engagement, and ROI on a per post basis.

The Hows:

Armed with a strong a reasoned perspective with what I wanted to measure, and why I wanted to measure it, I set out to evaluate existing Facebook analytics packages. After all, why reinvent the wheel if it doesn’t need to be?

The first solution I evaluated was Facebook’s own Insights…and I quickly moved on. While Insights has improved exponentially over the last twelve months, and remains the source of much of my raw data, it just never painted a clear enough picture on the level that I wanted to evaluate and report from.

From there I moved on to the big name providers in the space such as Context Optional, Vitrue Publisher, Involver AMP, and several others. Here’s the highlights of what I found:

• Most of these services were dependent on content being pushed through a custom publisher.
  • This was annoying because there was no simple mechanism for comparing content published with the custom publisher to content published through a fan page’s native functionality, much less a white label application developed by a different provider.
  • This became increasingly problematic when circumstances called for mobile updates from Facebook’s own app.
• Despite using a custom publisher, the analysis this services provided was only marginally better than Facebook’s own Insights dashboard.
  • I quickly developed a litmus testing for evaluating new services, the “30 Second Fan Growth/Page Views” test. It goes as follows: If within 30 seconds of logging into your service I see graphs showing basic fan growth and page views, my next click will be the logout button.
  • Most services I evaluated failed.
• Per Post metrics were either severely lacking or completely absent.
  • Simply put, I needed to know if Video X performed better than Poll Y or Event Z and I wasn’t getting it.
• There was hardly any cross-brand benchmarking.
  • It would seem like a pretty basic need; if I’m managing two fan pages (one w/ 50,000 fans & another w/ 500,000 fan), how can I compare their performances?

Summary: I wouldn’t say that I found cause to reinvent the wheel. Rather, I was forced to question whether or not the wheel had been invented yet.

Designing My Own:

So I developed my own approach that resolves the deficiencies that I cited above. At first it was a rather manual process of data scraping and entry but, as I my understanding of the platform and model have improved,  I have found opportunities for automation. I hope to share more in terms of specifics as I am able to but, for now, trust me when I say it is really cool.