The top of the company seems like the obvious place from which to steer a company in a new direction: The boss declares and it is so. Right?

Not really. As counterintuitive as it may seem, the CEO can be the most hamstrung employee of all, bearing the weight of the entire organization and living in the crosshairs of impatient investors, skeptical media, customers who talk sustainability but act on price, and middle management waiting for the next leadership transition.

Shortly after Indra Nooyi became Chairman and CEO of PepsiCo in 2006, she pledged to lead the company by an initiative she called Performance With Purpose, with plans to gradually shift Pepsi’s portfolio toward a healthier mix of products. But as she recently told a conference of corporate-sustainability professionals, some investors reacted poorly, asking why the company couldn’t just focus on tried-and-true products such as chips and soda. Nooyi remembers their position: “But your key competitor is doing great because they're focused on soda. Why aren't you doing the same thing?”

Despite encountering resistance, Nooyi has so far stuck with her strategy, which is obviously the right one from the perspective of public health. (That being said, Marion Nestle, a professor of food studies at New York University, told The New Yorker, “The best thing PepsiCo could do for public health is go out of business.”)

Investors can constrain company leaders. Employees can also turn out to have opinions that differ from those of the CEO (surprise). One is reminded of the saying that managing is like standing in a cemetery: lots of people under you, but no one is listening.

Jeffrey Hollender was founder and CEO of Seventh Generation, the eco-friendly consumer-goods company. He told me about how his awakening to “systems thinking”—a discipline, popularized by thinkers such as Peter Senge, that advises a holistic and interconnected approach to solving problems —led him to institute a major program in the company, requiring staff to spend a few days a month in training and discussions aimed at transforming not only how the company did business but how staff thought about their role in the world.

Three years in, Hollender said he realized that not everyone was as excited about the topic as he was:

I asked people to raise their hand if they supported this process or if they didn’t, and the majority of the company was not on board. It was one of the worst days of my life, but we ended the whole process.

The good news is that at least we had a culture that allowed that kind of dissent. But it showed me that people are not going to do what I want them to do unless I engender equal commitment and passion on their part.

Those debunking the myth of the omnipotent CEO also fault risk-averse boards, unenlightened regulators, and the inaction and fear of CEO peers for limiting what a company leader can do.

But to absolve the CEO of responsibility and shift blame to those other parties evokes David Foster Wallace’s riff in The Pale King:

But the whole dark genius of corporations is that they allow for individual reward without individual obligation. The workers’ obligations are to the executives, and the executives’ obligations are to the CEO, and the CEO’s obligations are to the Board of Directors, and the Board’s obligations are to the stockholders, who are also the same customers the corporation will screw over at the very earliest opportunity in the name of profit, which profits are distributed as dividends to the very stockholders-slash-customers they’ve been fucking over in their own name. It’s like a fugue of evaded responsibility.

Jeffrey Hollender agrees:

When someone says, “I can’t do it because the market won’t let me,” or, “My board won’t let me,” that is largely a cop-out. What that means is that you have failed to convince your board that you have a good enough strategy that they should endorse, and that they should weather whatever short-term negatives might be faced because, in the long run, you are doing the right thing for your investors.

Indeed, Nooyi has said that the PepsiCo board has been key in helping her manage the challenges to her strategy.

It is not that CEOs shouldn’t take responsibility for their actual and potential power—it’s that they should realize its limits. Not only should they not expect everyone to blindly carry out their agendas: If they truly care about what they’re promoting, they should not want that to happen. If a CEO takes too much personal ownership of a topic, the next person to fill that seat might feel the need to throw it out to make his or her distinctive mark on the company.

I joined BP in 1999 because I admired its CEO at the time—John Browne, who not long before had set ambitious targets to reduce BP’s greenhouse-gas emissions, becoming the most outspoken head of a major energy company on the need to fight climate change (though the bar was low then and still is today).

During my nine-year run with BP, I saw Browne’s lofty rhetoric on climate change and human rights come to life. BP met those emissions-reduction targets ahead of schedule while I focused on the social side of sustainability, setting up partnerships and programs around big projects in Indonesia, China, and elsewhere aimed at improving local livelihoods.

Browne stepped down in ignominy in 2007 after an ex-boyfriend sold his kiss-and-tell story to a UK tabloid. Tony Hayward took over, announced his intentions to sell off BP’s renewables business and told a Stanford Business School audience that the company had “too many people that were working to save the world.” The rest is tragic history: eleven men killed in the 2010 Deepwater Horizon disaster; $40 billion and counting in related costs.

And yet. Straight through the Hayward era, BP carried on with the progressive, extraordinary, and, it must be said, expensive efforts that we set up in Indonesia: anti-malarial programs, educational and vocational training for local residents, economic-development programs, and even an external panel which still visits the site regularly, meets with stakeholders independently, and writes a public report submitted directly to the CEO.

I left BP in 2008 to work on a United Nations project on business and human rights, and I would be lying if I said I hadn’t been eager to get out from under the new regime. But I realize in retrospect that while Browne’s departure felt like a seismic shift at the time, in reality, supertanker companies can’t change that quickly. Hayward didn’t turn the entire company completely upside-down, and Browne’s BP was far from perfect. Yes, tone at the top matters, but the C-Suite is not necessarily where all of the action happens.

The world needs CEOs who are willing to speak up about the need for business to be more sustainable and back that up with action: Paul Polman, currently Unilever’s CEO, is the darling of the sustainability movement for doing just that. (Disclosure: Polman contributed a blurb for my book.) But as much as I admire Polman and am rooting for him to succeed, I will judge his legacy by how much of his sustainability program outlasts his tenure.

Because at the end of the day, what matters most is what happens on the ground, to the communities and environment in and around a company’s operations. And for that, the CEO’s office might be the last place to look.

One place that we should look to assess a company’s impacts on the world is in its supply chain. Which is where we’ll turn our attention to next in this series’ next installment.