GM Pulls Its Facebook Millions: The Social ROI Argument We Needed

Three days before Facebook’s initial public offering, the third-largest advertiser in America gave the business press a gift. General Motors announced it was withdrawing its roughly 10 million dollar Facebook advertising budget because, in its assessment, the paid ads were not selling cars.

The detail that elevates the story from gossip to case study: GM kept its Facebook pages, the free presence, into which it reportedly pours several times the withdrawn ad figure in content and agency support.

One company, in one week, publicly separated the question ‘does Facebook advertising work for us?’ from the question ‘should we be on Facebook?’, and answered them differently. That separation is the clarity the industry has needed for three years, delivered by a budget line instead of a conference panel. This post uses it to install the framework properly: social’s three value propositions, three ledgers, and the interview each must pass.

Why the blur happened in the first place

‘Social media’ arrived as one budget conversation because it was one cultural novelty. Organizations funded it the way they fund novelties. A single line item, vague goals, and metrics chosen because they were easy to get. Fans, likes, impressions.

The result, visible in nearly every social audit we run, is a category where three very different investments share one ledger. Each props up the others’ justifications. The ads are defended by the community’s warmth. The community is defended by the ads’ reach numbers. Nobody can say what any of it produces. GM’s move slices the blur, so let’s slice it properly.

The three ledgers, and the interview each must pass

• Paid: rented attention, judged like advertising. The interview is conversion math, the same one our downturn triage prescribed for every rented channel. What does a lead, gift, or sale from this spend cost against the other options, all-in? Facebook’s targeting data is genuinely unusual. Age, location, interests, connections. For some categories the math works. For others, users in personal-life mode simply do not respond to display-style units, which appears to be GM’s finding and matches the click-rate benchmarks circulating publicly. The point is not that Facebook ads fail. It is that they must pass the same interview as every ad, and ‘we should be visible where people are’ is not an answer. It is an evasion, not an answer.

• Owned: your pages, judged like service and community. The free presence does jobs advertising cannot. Answering questions in public. Serving current customers and members. Humanizing the institution. Hosting the community that already cares. Its metrics are engagement quality, response times, service outcomes, and referral traffic into property you own. Not fan counts. GM kept this layer, note, at real content cost, because the interview it passes is different: is this presence doing relationship work our other channels cannot do cheaper?

• Earned: amplification, judged like PR. When content spreads because people choose to share it, you are collecting distribution you did not buy. The interview: does our content actually travel, measured honestly, and does the travel produce anything downstream? Links, subscribers, inquiries. The social-signals thread our search coverage maintains gives earned sharing a modest additional payer, which changes no strategy but sweetens the honest ones.

Running the audit on your own program

We re-ran this exercise with two clients this week, and commend the mechanics. Pull every social cost into the open. Ad spend, tools, agency fees, and the staff hours that hide in ‘free.’ Assign each to a ledger. Then run the three interviews with real numbers.

The outcomes to expect, from experience. Some paid spend survives brilliantly: targeted campaigns with tagged links and measured conversions, often for events and recruitment. Some dies on contact with math: boosted posts bought for reach nobody can connect to anything.

The owned presence usually justifies itself at modest cost if it is genuinely staffed, and usually does not if it is a broadcast channel pretending to be a community. And the earned ledger is empty at most institutions, because nobody built content designed to travel. That is a finding, not a failure. Budgets move after this exercise. That is the point of ledgers.

Sector translations, since GM sells cars and you don’t

The ledgers transfer cleanly to our readership. The interview answers differ instructively by sector.

Higher education: paid social for recruitment often survives the interview. Targeting by age, location, and interests describes admissions marketing almost exactly. Tagged campaigns to program pages produce measurable inquiry math. The owned presence earns its keep in yield season, admitted students deciding, more than in prospecting, which should shape the content calendar.

Nonprofits: the earned ledger is the underused one. Mission stories travel in ways product marketing never will. The organizations that design content for sharing, the SOPA lesson from January at gentler stakes, collect reach their ad budgets could not buy. Paid social for donor acquisition, by contrast, interviews poorly at most gift sizes. Knowing that before the pilot is the pilot’s cheapest outcome.

B2B: the honest answer is often that the buyers are on the platform in personal-life mode, per our 2007 framework, and the budget belongs in channels where they are working. The exception is recruitment marketing, where the targeting logic mirrors higher ed’s.

Same ledgers, different verdicts, which is the framework working as intended. It produces decisions, not dogma.

The timing, and the larger permission

A word on the IPO-week theater, because the timing was not incidental. Facebook goes public this Friday at a valuation that prices in an advertising future the GM announcement publicly questions. The next year of coverage will be a referendum conducted in earnings calls.

Our readers should watch it the way we watch all platform economics. The ad products will iterate, the measurement will improve, and the interview answers may change. Ledgers get re-run. That is their virtue.

But the durable gift of this week is smaller and more useful than the referendum. An unimpeachably mainstream company demonstrated that ‘no’ is an available answer to social spend that cannot explain itself, while ‘yes’ remains available to the layers that can. Boards heard it. Use the permission.

Three ledgers, three interviews, annually, and your social program will be the rare one whose budget meeting runs on evidence instead of anxiety. The fear of missing out just lost its biggest customer in America. Your organization can afford to lose it too.