PepsiCo launches new Facebook-inspired carbon calculator

For a company like PepsiCo, which oversees more than 20 brands and hundreds of different products around the world, calculating the carbon footprint of just one of its products can take weeks, and at a significant cost to the company. To save time and money, PepsiCo teamed up with researchers from Columbia University’s Earth Institute to create a tool that can measure the carbon footprint of thousands of products all at once.

The calculator, which lacks an official name, can calculate the carbon emissions of different materials and activities in a company’s supply chain and operations, and within minutes pinpoint which of these carries the largest carbon footprint.

“The objective was to give companies several capabilities at once with only a single effort,” said Christoph Meinrenken, the tool’s lead researcher and associate research scientist at the Earth Institute.

The calculator was developed to follow publicly known carbon footprinting standards such as the GHG Protocol Product Life Cycle Analysis (LCA) standard and PAS20:2011. The methodology and software helps businesses identify which materials or activities in their supply chain and operations have the biggest effect on the total carbon footprint of one of their products, product lines, brands or regions. The calculator also reveals the accuracy of this information and how this accuracy can be improved so a company can make better business decisions.

“We saw the opportunity to use our carbon/greenhouse gas analysis as a base for building a broader decision-making tool that could help us identify other efficiency opportunities throughout our supply chain, drive innovation and improve our overall operations,” said Robert terKuile, PepsiCo’s senior director of environmental sustainability.

The tool also provides certifiable product footprints to be used in ecolabeling and for environmental measuring groups such as The Sustainability Consortium and GoodGuide. This certification requires an intensive, bottom-up assessment of each product’s entire life cycle in order to provide the required microscopic level of detail and to be auditable outside the company, said Meinrenken.

The tool is not the first of its kind. Earlier this year, Danone announced it had developed a system, in partnership with SAP, that can calculate the carbon emissions of individual products. Meinrenken said the inner workings of the Danone tool hadn’t been made public, so it was hard to adequately compare the two. He said PepsiCo’s tool was developed before Danone unveiled its calculator.

The PepsiCo tool takes inspiration from sites like Facebook and Netflix, which mine huge swaths of data to figure out what users like. It analyzes data already stored in a company's database to infer information, like what materials are in a product and where they come from. This process saves a company time and money, said Meinrenken.
 
"This is just a general argument of being smart and efficient with companies' existing data to mine and 'milk' it if you will to learn additional things from the same data, rather than hiring additional staff and building up new data," he said.
 
Another similarity is the way websites like Facebook and Amazon use information about a customer to extrapolate to other customers they know very little about. These websites have an ever-expanding sample of known customers who each carry descriptive data such as age, gender and areas of interest. Along with this basic information, the websites also know to what extent their customers like certain things.
 
For example, in the case of Netflix, the company asks customers to rate various films, so it will know how much a customer likes a genre of film. A computer algorithm then uses this sample to estimate how much a customer who hasn't yet rated the movie will like it -- all the algorithm needs to do so is use the available descriptive data.
 
In the case of the PepsiCo tool, its sample is made up of materials that have their own descriptive data such as material type and origin. The algorithm, which has been trained on a sample with known emissions, can now estimate the emissions for a variety of materials that the company only has descriptive data for, but no known emissions' information. This means companies don't have to manually enter those emissions for each material, which would place a huge cost and time burden on a company, said Meinrenken.  
 
"No company in the world has yet been able to footprint all of its products," Meinrenken said. "In this sense, we see our software not as a tool to do company-wide product carbon footprinting faster and cheaper, but indeed to do it at all," he said.  

In the past, companies relied on teams of experts to manually measure emissions. These experts used sophisticated LCA software, but the method had its drawbacks. LCA is time-consuming and costly -- it still requires manual data collection such as phone calls and emails, and manual entry of the collected data into the system.

The new tool "dramatically reduces the time and effort as well as required LCA expertise that company employees or outside consultants have to spend on the tool and its associated databases before it can spit out any meaningful carbon results," said Pepsi's TerKuile.

It's too early to tell the total cost savings from using the calculator, he said, but he anticipates the savings will be significant. Prior to the launch of the calculator, measuring the carbon footprint of a single product -- known in the industry as a stock-keeping unit (SKU) -- at PepsiCo took more than 100 person-hours, and even that was a “somewhat optimistic benchmark,” Meinrenken said.

“You can imagine what this means if you have tens of thousands of SKUs,” he said.

The Earth Institute and PepsiCo initially aimed to standardize footprinting and measure individual SKUs, such as Tropicana orange juice, said Meinrenken. But in 2009, the focus shifted to entire product portfolios to minimize a company’s costs as much as possible.

Although Meinrenken said he couldn’t go into specifics, his team hopes to make the tool available to other companies to use in their carbon analysis. They also plan to test the methodology in other environmental areas, such as calculating water savings.  

Image of discussion with laptop by Wave Break Media available via Shutterstock.