Facebook can predict if you’re gay and like ice cream

Networked Insights saw declining mentions of iPhones by teens as a signal for Apple’s depressed earnings forecast

What women most want is ice cream, while men want cars says social analytics firm NetBase.

Facebook can predict if you’re gay or exactly where you’ll be in one year from now within 50 meters with high accuracy (Hey, do you like Spongebob and work out on Wednesday mornings at the gym)?

Moneyball for social media? Not so fast.

After all, it sounds reasonable:

Grab all the mentions of a company across all social channels, categorize the sentiment using the latest LSA techniques– then correlate to their stock price.

Predict how your posts will do or even suggest content tweaks based on what will happen in the future. React to tomorrow’s news today, proclaims one social startup.

BUT REALITY IS MUCH LESS SEXY

• A liquor company wants to know if they run Facebook ads, what the impact will be on incremental sales.
• An entertainment company with the most popular TV shows wants to measure whether social engagement drives more TV tune-in, which yields more ad dollars.
• A B2B marketing company wants to drive not just leads, but sales-qualified opportunities via Facebook.

Of these three cases, which do you believe has actually measured social ROI with hard dollars?

If you guessed the last one, you’re right.  It’s Marketo and they’ve traced revenue all the way back.

But the entertainment company can’t definitively tell yet. Nielsen has a product called Online Campaign Ratings (OCR) that will show unique and total reach across different demos, broken out by digital vs TV.  They’ll show efficiency in reaching your target– even performance by Facebook ad campaign (if you spend at least $100k a month).

Their Watch Effect product certainly measures correlations between activity in social and TV, and may start to demonstrate incremental lift within segments that we define.

The CPG company wants to know not just increased awareness of the product (it’s already well-known in the target market), but how much incremental trial comes from stronger promotion on Facebook against the right audiences.

There is the DataLogix and Facebook partnership where point of sale data is tied back to actual Facebook users. Where the audience is big enough, it shows correlation, but not causation.

The new partner category targets allow you to segment your existing audience by loyalty card and other POS/retail data.

BUT HERE’S THE BILLION DOLLAR QUESTION

What’s truly driving incremental sales?

Correlation is not causation:

• We know that a popular TV show will drive more social activity– so these will trend together.
• We know that your existing real-world customers are more likely to become fans– so we can’t say that a fan is worth X% more than non-fans. Yes, you’ll still see a ton of “research” on this.
• And there are still plenty of folks trying to buy fans and grow engagement, often with puzzling results.

But you and I know that if anyone can truly answer this ROI question, they control how all marketing budgets flow. If you can demonstrate clear profits– that you can put in one dollar and get back 20 dollars– who wouldn’t go all-in?

WHAT THE FUTURE REALLY HOLDS

It’s not about predicting the future– that’s a gimmick for news coverage.

It’s about measuring the hidden impact of social on ROI in other channels. Yes, the ROI of social doesn’t happen in social:

• When you build you presence on Facebook or wherever, you have greater awareness, which increase the CTR on your Google ads. The higher CTR, in turn, drives more brand searches (for your name) and increases your account’s Quality Score (influencing what you pay per click). Have you ever thought about what caused people to search for your company name on Google?
• When you run a custom audience on Facebook (uploading your email addresses like so), some people will click on the Facebook message, but more importantly, your email open and click rates will increase for folks who saw your message on Facebook. Do you know how to measure this? It’s easy.
• When you run Facebook ads targeting people by job title or even to get your messages into the newsfeed (60 minute tutorial here), people see it, but might not click on anything. You see a picture of ribeye steak from Texas Roadhouse in your newsfeed on your iphone, then decide to pull into the restaurant as you drive by. No last click, coupon, or trackable link to your sale.

The real meat of social ROI comes from tying together all your data sources and tracking back to revenue events.

So if you can’t measure the ROI, even via indirect lift testing, then it doesn’t exist in our book. But at the same time, you might actually be hitting a home run in social, but don’t notice it, because you’re not measuring the impact in your other channels.

Some stuff just isn’t measurable yet or ever will be (although the NSA may beg to differ), but shouldn’t you at least be collecting unified reports across all your marketing channels to get full credit for your efforts?