How to choose LCA software: eight evaluation criteria

The most common mistake in choosing LCA software is putting feature lists side by side. The lists look alike; the real differences emerge in use and usually concentrate in three places: your access to data, whether the output survives verification, and whether your team actually uses it.

1. Database access and licensing

The software itself is a calculation engine; the background data decides the result. Is the ecoinvent licence sold separately or included? Which version, and which system model? Are version updates covered? These questions drive most of the total cost and rarely appear on a feature list.

2. Standards output

If you will produce EPDs, the software has to generate the EN 15804+A2 indicator set by module — A1 through D, with the three separate GWP rows. Some tools produce results but leave the conversion to declaration format to you, which means days of manual formatting per EPD.

3. Verifiability

A verifier wants to follow the path from source data to result. Can the software show each inventory line’s source, its data quality score and the calculation steps? Is the model versioned — can you return to the inputs behind a result from six months ago? This materially affects verification time and therefore cost.

4. Scale and repetition

Managing a thirty-product portfolio is a different job from a tool designed for a single study. Without model sharing across variants, parametric recipes and bulk recalculation, every variant is modelled from scratch. With a large portfolio, this is where the real cost accumulates.

5. Multi-user work

LCA is rarely a one-person job: the plant supplies production data, the sustainability team builds the model, sales uses the result, a verifier looks in from outside. On a single-seat desktop tool that means file versions circulating by email. Role-based access and concurrent work stop being features and become requirements as the team grows.

6. Scenario and sensitivity analysis

An LCA’s value is not in producing one number but in comparing alternatives. If you cannot run the effect of a material change, an electricity source or an end-of-life scenario within minutes, the model is a reporting tool, not a decision tool.

7. Export and data ownership

Can you export your model and data in a standard format? Are machine-readable formats such as ILCD+EPD produced — you will need them for submissions to databases like ÖKOBAUDAT and for the Digital Product Passport. And a more basic question: if you leave the software, do years of accumulated data stay with you?

8. Total cost of ownership

• Software licence — per user or per organisation
• Database licence — often more expensive than the software
• Training and onboarding — how many weeks to the first productive result
• Verification time — a traceable model reduces verifier hours
• Consultancy dependence — is external help needed for every new product