Remember when GM pulled their Facebook ads because they “didn’t work”? Yeah, about that.

From Bloomberg:

Facebook (FB) got a black eye last week when General Motors (GM) announced it would cease advertising on the platform, yanking $10 million in annual ad spending away from Mark Zuckerberg just days before the company’s IPO. The move caused some to wonder if Facebook’s business model is flawed. After all, the social network made 85 percent of its $3.7 billion in revenue last year from those little ad boxes on the side. If marketers stop buying them, Facebook’s stock price could get hit even harder and further than its 11 percent ding on May 21.

That’s wrong-headed, because no free market spends billions on a media format with zero return. What GM’s retreat really shows is the harsh reality that other brands must face: Making social-media communications work requires heavier lift than many organizations can muster.

Before we conclude that Facebook ads don’t work, first let’s put GM’s slap into context. The auto giant is emerging from a bailout crisis in a strong, lean mode, slashing costs to shore up its balance sheet. The company has reportedly targeted $2 billion in marketing expense reductions over the next five years. Cutting Facebook ads by $10 million puts GM only 0.5 percent of the way there. A few days after the Facebook news, GM announced it will not advertise in the 2013 Super Bowl.

GM has a robust presence on Facebook and in other social media, beyond paid advertising. When I riffed on Twitter, “Wow, GM yanks $10M from FB…” Mary Henige, GM’s director of social media, tweeted back: “We have more than 8mil friends on FB; not leaving them; engagement & content isn’t same as advertising.” Fair point, Mary.

But the truth is that Facebook ads work better for some businesses than others. GM did what any savvy marketer facing a budget squeeze does—it optimized away from underperforming media channels. Advertising, after all, is an investment. You need to put your funds against what works best.

Most Facebook ads are bought on a cost-per-click basis. This means the front-end cost of getting a potential consumer to respond is low, typically less than $2 per click. Each click on a Facebook ad puts a consumer on your product web site. If you then can get only 1 percent of those consumers who click on ads to “convert” and buy your product, you’ve achieved a $200 cost per sale. In essence, marketers try to buy customers at the lowest cost per sale possible. Paying $200 per new customer isn’t bad for many business models.

The challenge with Facebook, though, is that conversion rates can be very low in some product categories. Social media users are being social, after all. Unlike the pay-per-click ads that Google (GOOG) serves up only after consumers type in the names of products they are hunting, Facebook ads pop up while you’re bragging about your five-mile run. Curious tire-kickers might click on a GM Facebook ad to see the sexy Chevy Volt, but that doesn’t mean they want to buy one. If your conversion rate—the portion of people who eventually buy after clicking on your Facebook ad—falls from 1 percent to 0.1 percent, you’re now talking a $2,000 per-sale cost. That’s an expensive customer acquisition.

Keeping Facebook conversion rates up and customer acquisition costs down requires a constant battery of audience-targeting refinement, creative testing, and website “landing page” adjustments. Even if all that works perfectly, it’s often very difficult to track all the consumer respondents who come in via Facebook and end up on a sales floor. Salespeople notoriously steal credit for respondents whose interest was triggered by advertising; they often get higher commissions for leads they generate, vs. prospects handed to them by marketing. This challenge in marketing is called “attribution modeling,” and if a marketing model misallocates even one-third of Facebook respondents, the math on Facebook return on investment tips the wrong way.

The second challenge with Facebook is that brands struggle to reap real value from its social interactions, which often start with paid advertising. Facebook is famous for its “likes,” which supposedly open the door to a wonderful engagement between brands and consumers. If a consumer sees your Facebook ad, she might “like” your brand, allowing its content to pop up again later in her Facebook stream. The idea is that you move from being an old-fashioned, interruptive advertiser to become a real “friend” of the consumer, sharing brand stories in the middle of her Facebook page, right next to her college roommate’s cat photos.