🎯Accuracy Score

Life cycle assessment is a complicated concept that is often subject to uncertainty. Assumptions often have to be made behind the scenes, and data of varying quality is processed.

This is why Yook developed the accuracy score to create transparency on the quality of every product's carbon footprint. It indicates limitations due to data quality and availability. On top of the accuracy score, we determine a safety margin to avoid underestimating the carbon footprint.

The accuracy score is the average of three sub-scores: product data quality, background data reliability, and mapping precision.

The accuracy score is derived from a weighted average of three sub-scores, each assessing different data layers for every life cycle stage. This score is calculated for each product part and life cycle stage.

sub-score	criteria
Product data quality	Reflects the availability and quality of product data. Relevant product data points such as weight, material composition, location, production, and supply chain information are assigned relative importance depending on the product category. This enables targeted data collection and improvement where it matters most.
Background data reliability	Background data refer to emission factors or other information about a product's environmental impact This sub-scores measures the quality of background data connected with the product data (i.e., the quality of emission factors). Depending on whether suppliers' environmental impact information is available, background data can be primary or secondary. We evaluate the reliability and quality of such background data.
mapping precision	This reflects how precisely we could map a data point. It corresponds to the DQR's main areas of technical representativeness (the degree of similarity) and temporal and geographical representativeness (the degree of precision and similarity).

Calculation of the sub scores

For every sub score, we have a unique calculation process:

Product Data quality

As this reflects the product data availability and quality, we check whether all relevant product information needed for LCA is available. These include the following:

Material

• product weight

• material type

• material composition/ weights

• material origin

• material properties

Production

• process type

• material loss rate

• location

• energy consumption

• energy mix

Transport

• start and destination locations

• transport means

Packaging

• material type

• material weight

• material origin

• material properties

Background data reliability

Background data refer to emission factors or other information about a product's environmental impact. If no primary information (e.g., a supplier specific LCA or EPD) are available, we use secondary data from LCA databases such as Ecoinvent, Idemat, Agribalyse, etc.

For the reliability assessment we evaluate the source with respect to its general reliability, transparency, documentation and plausibility. Additionally we flag controversial materials by adding them to our "material red list".

Secondary background data

The secondary background data reliability is an average of

1. source_reliability: indicates the general trustworthiness of our EF database

2. activity_reliability: refers to the material being on the redlist of controversial materials

Primary background data

We use the PEF DQR rules to calculate the reliability of primary background data reliability. These include the following criteria:

• whether the study is externally verified

• whether the data refers to the same time period

• whether the activities refer to the same technology

• whether the activities reflect the same geography

Mapping precision

The mapping precision sub-score is aligned to the PEF DQR. This means, in order to have the DQR for PEF compliant studies, it can easily be extracted.

mapping precision

Three criteria for the mapping were assessed based on the ISO DQR system:

• Technological representativeness

• Geographical representativeness

• Time representativeness

what does the integrated accuracy score mean?

Next, in order to determine the overall accuracy score of the product's carbon footprint, we compute two weighted averages, each founded on a materiality assessment, specifically regarding the relative carbon impact. Hence, we proceed with the following three steps:

1. item accuracy score: equally weighted average of product data quality, background data reliability and mapping precision. Calculated per life cycle stage. This means, every part of a product gets its own accuracy score for materials, production, transport and packaging.

Imagine a product made of 3 kg plastic and 3 kg steel. Both raw materials are processed and then assembled together. We calculate the accuracy score for both materials separately, so that we end up having a value for each material and each life cycle stage of them. 

2. life cycle stage accuracy score: based on the relative impact of different materials/ production steps to the respective life-cycle stage, we weight the accuracy scores of each element to calculate a weighted average accuracy score per life cycle stage.

Due to their different carbon impacts, the plastic part contributes 30% and the steel part contributes 70% to the material carbon footprint. Therefore, 70% of the material accuracy score is made up of the steel accuracy score and 30% of the plastic accuracy score.

3. overall product accuracy score: as a last step we calculate the final accuracy score based on the relative impact of each life cycle stage to the final PCF.

Accuracy Score = 
    Material CF / PCF * Accuracy Score (Materials)
    + Production CF / PCF * Accuracy Score (Production) 
    + Transport CF / PCF * Accuracy Score (Transport)
    + Packaging CF / PCF * Accuracy Score (Packaging)

The weighting of the accuracy score ensures that the materials, production steps or parts of the product with the highest carbon impact count more than small parts that have almost no significance. It also helps with targeted data collection and improvement where it matters most.

The accuracy score is given on a scale of 1 to 100, where 1 is the lowest and 100 is the highest. An accuracy score of up to 90 can be achieved, as no model can ever represent the full reality.

We have developed a system to assign the scores to a qualitative description for accuracy score.

Accuracy	Score	Description
Low	1 - 20	Very inaccurate Spend-based and Average data No specific product primary data (e.g., material composition, weight) is available
Medium	21 - 40	Poor availability/ quality of primary data, Used heuristics and proxies or very low secondary data quality
Elevated	41 - 60	Primary data available for some datapoints (e.g., product weight, main materials) Good background data quality (This level is usually obtained after the first iteration of PCF calculations)
High	61-80	Detailed primary data availability (e.g., as above + material origins + specifications, production details) Good background data quality (This level is usually obtained with some data collection effort)
Very high	81-90	Very good availability and quality of primary data Supplier specific information on materials/ components. Good background data quality Precise mapping of product and background data.
-	91 - 100	Since no model can perfectly reflect reality and climate sciences are always subject to uncertainty, an accuracy score of 100 can never be reached.

Safety margin

The safety margin is a factor determined by the accuracy score by which the calculated product carbon footprint is multiplied in order to control for uncertainties and to avoid underestimating carbon emissions.

Accuracy score	Safety margin
1 - 20	1.5
21 - 40	1.4
41 - 60	1.3
61 - 80	1.2
81 - 90	1.1

Imagine we have a calculated PCF of 10 kg CO2e.
- an accuracy score up to 20 leads to a PCF including safety margin of 15 kg CO2e
and
- an accuracy score higher than 81 to a PCF including safety margin of 11 kg CO2e