A customer was asked by a retailer to develop a range of low-carbohydrate products with a maximum of 4.0% carbohydrate. We explained to the customer how to use the contributions function to view the contribution made to the carbohydrate value from each of the ingredients.
They then developed a series of products which all contained 3.8-3.9% carbohydrate, a very consistent set of results.
However, when the retailer sent the products to the lab, the carbohydrate values were between 3.4% and 4.9%.
After calling in for help, we checked the customer’s work and results and found everything to be authentic. From our experience of running laboratories, we were then able to explain that the discrepancy had occurred because carbohydrate testing in the lab can lack precision as it is calculated from other values. This means that if there is a small error of say, 0.3% in fat, 0.3% in protein, 0.3% in moisture, 0.5% in fibre and 0.1% in ash, the % carbohydrate could easily be over 1% out.
The customer passed our explanation on to the retailer, who accepted that lab analysis was not ideal in this case for a very low carbohydrate product.