The Review step is a quality-check stage that appears after your data has been imported.
Our AI-powered import pipeline automatically:
Extracts weights and units from your data and estimates item weights where necessary.
Matches each item to a food or drink item in our emissions database.
We may flag certain rows where our system is uncertain about the overall pack weight or cannot find a suitable food match. This could be due to ambiguity in your input data or where the food item is better modelled as a recipe.
This step helps improve the accuracy of your emissions calculations before generating reports.
While reviewing your data is strongly recommended, it is not mandatory before generating Purchase, Company Carbon Footprint (CCF), or Sales reports.
How the Review Process Works
We recommend the following workflow:
1. Open the Review Tab
Navigate to Food and Drink → Review to see all flagged items.
2. Check Each Flagged Item
Click into each item to review:
The selected food match
The item weight
The number of items per pack
If the information is correct, confirm the item using the green tick.
If changes are needed, edit the relevant fields and save.
3. Work Through the List Systematically
Move through the flagged items from top to bottom. If you leave the page, your flagged items will remain in the Review tab, so you can return to complete the process at any time.
4. Optionally Review Non-Flagged Items
The AI flags items that are likely to need attention, but it does not guarantee that every issue will be identified. You can review any item at any time by clicking into it under the food and drink tab and updating the fields manually.
Special Consideration for Sales Reports
For sales reports, you may notice more items flagged for non-ideal food matches.
This is both common and expected.
Why Does This Happen?
Sales data often includes:
Prepared dishes (e.g. “Chicken Caesar Salad”)
Composite menu items
Packaged products
By default, all food and drink items are initially matched to a basic food in our database (which contains approximately 3,000 food items). This database is most comprehensive for individual food products like fruit and vegetables, grain products, meat and dairy products.
While some dishes are included, the database is not designed to comprehensively cover multi-ingredient meals.
Prepared dishes and composite products are rarely made up of a single ingredient. As a result, the initial AI food match may not fully reflect the true composition of the item.
Improving Accuracy for Complex Items: Switching to a Recipe Model
For dishes or multi-ingredient products, emissions are usually more accurate when the item is modelled as a recipe rather than a basic food.
How to Do This
Open the flagged item.
Change the food match from Basic Food to Recipe.
Use the AI recipe generator (if needed) to create an ingredient list for the item.
Review and adjust the ingredients as required.
The recipe model allows emissions to be calculated based on individual ingredients and their proportions within the dish.
This provides a more realistic and representative emissions profile for menu items and packaged products.
Taking the time to review your data will significantly improve the quality, accuracy, and reliability of your emissions reporting.