Automating LCA for Multi-Ingredient Food Products

Modelling complex processed food products in a Life Cycle Assessment (LCA) is among the most complicated challenges in applied LCA. Consider a ready-made lasagne. The ingredient list could read something like this:

INGREDIENTS: Bechamel 31% [Water, Cream, Corn starch, Skimmed milk Powder, Mature cheddar, Salt, vegetable oil (sunflower, palm), Black pepper, Nutmeg, Preservatives], Water, Beef 24%, Pasta sheets 13% [Durum Wheat semolina, Water, Egg, Salt], Tomato sauce [Crushed peeled tomatoes, Tomato concentrate, Onions, Vegetable broth, vegetable oil, Corn starch, Oregano, basil, salt, pepper], Mozzarella cheese, Dijon mustard, garlic, salt, pepper.

There are over 35 distinct components once you unpack the nested sub-recipes, and each one carries its own supply chain, its own cultivation region, its own set of environmental variables. Now imagine conducting a full Life Cycle Assessment across that product manually. This can take four to six weeks per SKU. For a food brand managing hundreds or thousands of products, that timeline is untenable. By the time the report is complete, the recipe may have changed, a supplier may have switched, or a new database version may have superseded the data used. The LCA is already out of date before it reaches a decision-maker's desk. Ideally, LCA should be a live input to product design, not a report produced after the decisions have already been made.

The complexity creates a pain point on two fronts: the sheer volume of work, and the challenge of maintaining consistency across models. Many ingredients are sourced from multiple suppliers, each requiring its own dataset, transport assumptions, and logistics data. Layered on top of this is the full complexity of modeling every life cycle stage, a process that routinely takes many weeks to complete. It does not need to be this way. Much of this work can and should be automated, scaled from weeks down to minutes.

The Data Granularity Problem

Food cultivation practices can vary widely from location to location. Climatic conditions affect the need for irrigation and greenhouse coverage, the tomatoes in our lasagne example being a case in point. Regulatory frameworks vary by country and region, shaping farming practices further, and conditions can differ from farm to farm, for which primary data is ideally available. Regardless of whether data is primary or secondary, each ingredient requires a corresponding dataset to be selected in the model, a process that grows increasingly cumbersome as ingredient and supplier counts rise. Because this selection should be deterministic and must be repeated across many products, it is a natural candidate for automation.

To understand the scale of the challenge, consider just the tomatoes in the lasagne's sauce. Tomatoes sourced from Spain under open-field Mediterranean conditions carry a fundamentally different lifecycle inventory (LCI) than those grown in heated greenhouses in northern Europe. Differences in energy use, water consumption, and land impact can shift the Global Warming Potential (GWP) and other impact category contributions significantly. Apply that same variability logic to every ingredient in the list above; beef sourced from Argentina versus Ireland, generic palm oil from Malaysia versus a certified sustainable source, and each permutation represents a different dataset selection decision. In a portfolio of thousands of SKUs, those decisions compound rapidly.

This challenge reproduces itself across multiple LCAs when a producer uses nested LCA results from a supplier of intermediate products. A producer of a ready-made lasagne might, for example, source the ragù from a supplier and incorporate its corresponding LCA results into the lasagne's own assessment. When both producer and supplier use an automated system like Sustained, the producer can be confident that all suppliers are working on the same methodological basis.

A further complication arises when an ingredient simply isn't represented in standard LCI databases. Niche additives, novel functional ingredients, or highly regionalised processing steps may have no direct match in datasets like Ecoinvent or the EU's EF database. In these cases, practitioners must use proxy datasets - a judgment call that introduces uncertainty into the model. In an automated system, these gap-filling decisions are made once, validated centrally, and applied consistently across every product containing that ingredient. The cumulative uncertainty across 20 or more ingredients can be tracked systematically, rather than accumulating invisibly across disparate spreadsheet files.

Methodological Consistency

Efficiency is only half the battle; the other is consistency. In manually executed complex LCAs, the subjective choices and assumptions that are a necessary part of the process become difficult to track. When browsing a Life Cycle Inventory (LCI) database, choosing a proxy dataset involves a degree of practitioner judgment. While justifiable, that judgment can shift over time or vary across personnel.

In our lasagne example, a colleague might have used leek as a proxy for garlic when creating an LCA for a ready-made ravioli product. By the time a new practitioner runs the lasagne assessment, that colleague has left the company, a newer database version is available, and onion is selected instead, or perhaps the new database even contains a direct match for garlic. Such small differences accumulate.

One of the most consequential consistency challenges in food LCA is allocation: how environmental impacts are divided when a production process generates more than one output. Consider a by-product like citrus peel, which might be a waste stream from one product and a functional ingredient in another. Depending on whether mass allocation, economic allocation, or system expansion is applied, the attributed impact can vary by an order of magnitude. In a manually managed spreadsheet, there is no mechanism to ensure the same allocation rule is applied consistently to the same multi-output process across every product in the portfolio. The result is not just inconsistency, it is potential greenwashing liability, particularly as regulators under the EU Green Claims Directive scrutinise the methodological basis of environmental claims. See our earlier post on allocation for a deeper treatment of this topic.

This challenge is particularly acute in complex models with many ingredients, some of which feed into nested intermediate products. It is also common for intermediate products to operate under a different set of system boundaries, adding a further layer of consistency that must be verified. Information and context are easily lost when LCA results are passed up and down the supply chain via spreadsheets. If the bechamel sauce in our example is not produced in-house and the supplier provides only an aggregated set of environmental impact figures, the ability to verify consistency depends entirely on shared methodology. When both parties use Sustained, that consistency is assured.

Consistency challenges extend beyond dataset selection. They also apply to data gaps filled with calculations or estimates, something LCA practitioners must regularly address when primary data is not comprehensively available. Beyond judgment, there is the risk of simple error: applying a methodology's prescribed defaults incorrectly, for instance. In an automated system, such defaults are entered centrally and validated once, ensuring accurate application across all instances. This matters equally for LCAs conducted at different points in time. When methodologies are updated, or more current factors such as recycling rates become available, older LCAs must be brought into line to remain comparable. Doing this manually across complex models carries a high risk of missed updates. The same principle applies to allocation rules, waste treatment, and energy sources at the use stage. In an automated system, consistent application is built in.

Scalability & Modelling

When a producer considers replacing a supplier (or an ingredient altogether) that change can affect multiple points in the model simultaneously. Take a switch of vegetable oil supplier where the new supplier not only sources from a different geographical region but also uses a different oil type: soybean instead of palm, for example. Vegetable oil may appear as a main ingredient, as a processing input such as frying oil, and across multiple products in the portfolio. Modelling the full environmental impact of a company-wide switch of this kind is where automation becomes a decisive advantage.

In a spreadsheet-based workflow, that exercise would require a practitioner to manually locate every product containing palm oil, update each model individually, recalculate all downstream impacts, and reconcile the results, a process that could take days or weeks across a large portfolio. In an automated system like Sustained, the same change propagates instantly. A product developer or sustainability manager can swap an ingredient, adjust a sourcing region, or trial a reformulation and see the full environmental impact shift in real time, before any physical change is made to the recipe.

This dynamic capability also makes proper sensitivity analysis practical, a methodological requirement under ISO 14044 that is rarely feasible in manual workflows. By systematically varying key inputs such as ingredient sourcing, processing energy mix, and transport distances, and observing the effect on overall GWP, teams can identify which variables most drive a product's footprint and prioritise their reduction efforts accordingly. The model becomes a decision-support tool rather than a compliance exercise. See also our post on eco-design as the first step toward sustainable food production.

The scalability argument is equally compelling operationally. A manually managed LCA programme scales linearly with headcount - more products means more practitioners, more hours, more opportunity for inconsistency. An automated system scales differently: the methodology is defined once, validated once, and applied consistently whether you have one SKU or ten thousand. For a food brand with an extensive portfolio, this is the difference between LCA being a resource-intensive exercise limited to flagship products and LCA being an integrated capability available across the entire range.

Compliance & Standards Alignment

The efficiency gains described above extend to auditors and reviewers. When deterministic automation is applied, repeating elements need only be assessed once. Across a portfolio of thousands of products, a reviewer can assess the system's behaviour and then focus only on the manual changes and overrides made in each case, drastically reducing the workload. Mass balance verification across model nodes no longer requires individual checking; it is baked into the system from the beginning. Sustained has been built to conform with the EU's Product Environmental Footprint (PEF) framework from inception, and ISO 14040/44 compliance is similarly accessible at the system level.

Version control is a particular vulnerability in spreadsheet-based LCA that tends to go unnoticed until it becomes a problem. When a methodology is updated (for example, when the EU transitions between EF versions or when characterisation factors are revised) manually updating hundreds of spreadsheet models creates significant error risk. An automated system manages these updates centrally: when a methodology version changes, every connected product model is updated simultaneously and the change is logged with a timestamp, creating a complete audit trail. This is not merely an operational convenience; it is increasingly a regulatory requirement. Under the EU Green Claims Directive, companies making environmental claims must demonstrate the methodological basis of those claims and show that they are current. An automated system provides this documentation as a matter of course.

Report creation is another area where automation adds significant value. Rather than producing a set of numbers in a spreadsheet, LCA models in Sustained automatically generate the corresponding documentation. These reports serve reviewers directly and can be submitted to substantiate environmental claims, an increasingly formal requirement as governments introduce mandatory disclosure frameworks.

A recurring challenge in monitoring a product’s environmental impact over time is comparability. To compare a product’s current footprint against a calculation from a few years prior, you must account not only for recipe changes, but also for any methodological updates or improvements in primary data in the intervening period. Pre-existing LCAs must be updated accordingly. A centralised automated system handles this in seconds; a manual practitioner must revisit the entire model.

Business Impact

What does all this boil down to? The ease of creating, modelling and updating LCAs means they can be integrated directly into reformulation decisions, proactively, rather than reactively reevaluating impacts after a new product generation has been implemented, and for a fraction of the effort and cost. A company with thousands of products is no longer limited to LCAs for flagship products, the whole portfolio can now be evaluated. This enables not only better decisions that reduce environmental footprint across all scopes and lifecycle stages, but also to substantiate those improvements with credible evidence. As eco-design becomes a commercial imperative, reducing turnaround from weeks to minutes is not a marginal efficiency gain, it is the difference between sustainability being a reactive compliance function and a genuine driver of product strategy.

To summarise the business case:

• Speed: Minutes per SKU rather than four to six weeks. Teams can respond to sourcing changes or recipe updates without waiting months for a revised assessment.
• Time to market: Reformulation decisions can be made in real time, with environmental impact as a live input to product development rather than a post-hoc check.
• Compliance readiness: Audit-ready documentation and consistent methodology reduce exposure under the EU Green Claims Directive and equivalent frameworks.
• Defensible green claims: PEF and ISO 14040/44-aligned methodology provides the rigour needed to substantiate environmental claims publicly and under regulatory scrutiny.
• Cost efficiency: As portfolio scale increases, the per-SKU cost of automated LCA decreases, while manual approaches require proportionally more resources. 

The future of food LCA isn't just about having better data, it's about having systems that can keep pace with the speed of modern product development. For brands that want to reduce their footprint, prove it credibly, and build sustainability into how they work rather than treating it as a periodic compliance exercise, automation isn't a nice-to-have. It's the only approach that scales.

Frequently Asked Questions

Q: How does switching to an automated system change the product development process?

A: Automation transforms LCA from a compliance exercise into a live strategic input for product design. Rather than waiting weeks for a report after a recipe has been finalised, developers can swap ingredients, adjust sourcing regions, or trial reformulations and see the environmental impact shift in real time. This makes sustainability a tool for rapid, informed innovation, rather than a post-hoc check that arrives after the important design decisions have already been made.

Q: How does switching to an automated system affect the topic of primary and secondary data?

A: By lowering the cost and overhead of conducting LCAs, automated systems make primary data collection more viable as adoption spreads upstream through the supply chain. But even where suppliers still work manually, integrating the primary data they provide is far more seamless in an automated system. Rather than manually updating individual product models, a user can replace the secondary datasets mapped to a given ingredient with newly received primary data in a single operation, applying that update consistently across every product in which that ingredient appears.

Primary data improves accuracy and is required for higher-quality assessments, but secondary data remains essential for ingredients and processes where primary collection is not yet feasible. Read more about balancing secondary and primary data in LCA here. 

Q: How does automation specifically mitigate the risk of greenwashing?

A: Greenwashing does not always stem from deliberate misrepresentation. It can arise just as easily from unintentional methodological inconsistency, applying different allocation rules to the same multi-output process across different products, for instance. In spreadsheet-based or conventional LCA software workflows, these discrepancies are difficult to detect. An automated system enforces consistent logic across the entire portfolio and centralises methodological updates, ensuring every SKU is calculated on the same basis. The result is a transparent, audit-ready evidence trail that supports credible environmental claims and reduces regulatory exposure.