(Happy New Year all…I forgot to post this one last Fall…)

Low-cost carbon footprinting is a Holy Grail for the sustainability world. But how do you measure your footprint at multiple levels — from products to business lines to the whole enterprise — quickly and cheaply? Over the last few years, PepsiCo has been working with partners at Columbia University to solve this interesting and complex business problem. The results of this partnership, what the team is calling a “Fast LCA” process, are emerging. And they’re encouraging.

To understand this initiative better, I recently spoke with two PepsiCo executives working on sustainability, Al Halvorsen and Robert ter Kuile, and the academic brain trust at Columbia led by adjunct professor Christoph Meinrenken. Here’s what I learned about three major issues:

1. Why do carbon footprints matter for your business?

Understanding your carbon footprint is a required skill of 21st-century business. Customers, consumers, employees, and investors (like the increasingly influential Carbon Disclosure Project, backed by institutions with $78 trillion in assets) want to know your contribution to — and actions to solve — this global challenge.

But it’s not just about reacting to pressure.Knowing your footprint helps you get proactive, spot risks and costs along your value chain, and identify opportunities to innovate. Getting smart about green data makes money. In essence, carbon is a proxy for energy cost and waste, and good carbon management is a proxy for good operational execution.

2. In layman’s terms, what have Columbia and PepsiCo accomplished, and how?

The detailed methodology behind this advancement is complicated: for the math and data wonks out there, see this short but dense article in the Journal of Industrial Ecology.

But for even layman like me, the problem is clear: To use carbon data to reduce costs and risks throughout the value chain, you need know the footprint of every single product that contributes significantly to your bottom line or brand. Conducting a detailed lifecycle assessment (LCA) is, to put it mildly, a resource-intensive exercise.

As Meinrenken and the Columbia team suggest in their Journal article, a full LCA for even a relatively straightforward consumer product like a can of soda would require data on

“…the masses of three packaging materials and five ingredients, transportation distances of all materials to the plant, amounts of four types of energy, transportation distances to stores, refrigeration times in stores and at home… and then all materials and activities have to be paired with respective EFs (carbon emission factors), bringing the total count of individual [data] inputs to approximately 100 for a single product alone.”

LCAs for an entire product portfolio would require thousands of often hard-to-get data points. It’s tough to justify this level of investment. PepsiCo’s ter Kuile put it succinctly: “there’s no way to look at all of our products at this level of detail in any reasonable time frame.”

So what has Columbia done? I’m not doing it justice fully, but it’s about algorithms and shortcuts. They start with internal operational data from existing SAP and Oracle databases – bills of materials (packaging, ingredients, and so on) on every single product, as well as shipping, energy, and water data for every plant. But instead of collecting an exact carbon emissions number from every supplier of those materials, they use statistically generated emissions factors (EFs), which provide good estimates on carbon for common inputs like sugar or corn. Modeling EFs is what saves the most time.

Other shortcuts draw assumptions on systemic issues like transportation distances, refrigeration time in transit or in the home, and recycling rates, all of which influence the footprint.

Then the model does something critical: it runs a sensitivity analysis to identify the inputs where variation could cause a meaningful change in the ultimate calculation. Thus the model helps managers zero in on data that’s worth spending more time to get right. Let’s say the model assumed that soda in France sits in the store refrigerator for two days instead of four. Does that number impact the total footprint very much? If so, managers can do more research and find better numbers (that is, more “primary” data).

(Note: for another interesting take on this process that likens the whole thing to a “Facebook-inspired carbon calculator,” see Allison Moodie’s piece on Greenbiz.com.)

Finally, the model makes assumptions about elements like packaging that may be common across many products. This is where it gets even more interesting for PepsiCo since it allows execs to explore “what if” scenarios. Which brings me to #3:

3. What’s the business value for PepsiCo and all companies with broad product portfolios?

As PepsiCo’s Halvorsen told me, “the real reason you do an LCA is improve the business… to put more efficient processes in place and innovate in the supply chain.”

To see how this works in practice, let’s go back a few years to the beginning of the PepsiCo/Columbia working relationship. The team produced a fascinating study on Tropicana orange juice, which concluded that the biggest contributor to the carbon footprint was not manufacturing or transportation, but natural gas-based fertilizer. For essentially no cost, PepsiCo could eliminate a third of Tropicana’s carbon footprint — and all the potential cost and risk associated with it — by switching to non-fossil-fuel-based fertilizer (their test farms are a few years into their experiment).

This exercise was so helpful, PepsiCo’s executives wanted to gather this level of strategic knowledge across the business for all products. To test Columbia’s new fast LCA model, they submitted data on two different parts of the business: the beverage business in China and the snack business in Brazil.

What makes this story interesting is what PepsiCo can do with the information at the product and business unit level — and it’s not to get an exact number of grams of carbon per bag of chips, which is fairly meaningless to consumers anyway. The real goal here is to pose “what ifs” and find the quickest, most profitable way to reduce impacts and improve efficiency.

These execs want to ask questions such as, “If we reduce packaging in one product, what does that do for other products that use the same packaging elements? What do we save in carbon, material, and money?” They’ve begun this process, but it’s still the early days. Over the next year, I hope to report on some operational changes that were made and measured.

A final thought on what’s required to make this happen: To avoid the old “garbage in, garbage out” problem, you need good data. PepsiCo knows a lot about its business — from the precise formulations of every product (to estimate supply chain impacts) to the exact production rates for each facility (to accurately allocate energy use for every product). In essence, the innovation here is combining really good, so-called “big data” with really good algorithms.

There’s a lot at stake here in dedicating scarce resources well. Getting carbon footprints right is a critical step on the path to healthy brands, higher profits, and a livable planet for all of us.

(This post first appeared at Harvard Business Online)

(Sign up for Andrew Winston’s blog, via RSS feed, or by email. Follow Andrew on Twitter@AndrewWinston)