EVIO Labs Website Search

Edibles Testing – Why are the results so variable?

EVIO Labs tests a wide range of products that go far beyond cannabis flower and oils. We get products as varied as elaborately decorated chocolate truffles to ice cream sandwiches and even granola. Tests results on these products have come back with widely varying potency results. This summer, EVIO Labs Regional Lab Director Ellen Parkin and Chemical Engineer Jeremy Campbell set out to identify the cause of these variations. Their results were presented to the at the Institute of Cannabis Research Conference in Pueblo, CO. Their presentation Sources of Uncertainty in Cannabinoid-Infused Products highlighted the many factors involved in analyzing different types of edibles, and identifying what could cause variability in test results. They found that testing uncertainty is less than 0.1%. However, variations that occur during the processing and manufacturing process can be much higher - thus it is important for producers to ensure they are making a highly homogeneous product prior to sending their edibles to test. A synopsis of their study is included here.​

Sources of Uncertainty in Cannabinoid-Infused Products
by Ellen Parkin, MS and Jeremy Campbell, MS

As states across the United States legalize recreational cannabis use, cannabinoid-infused edibles have increased in popularity. For many consumers, this is the first impression and experience of the cannabis industry. Edibles, however, have gained a reputation for being notoriously variable in their potency (think Joe Rogan’s story about gummy bears in his recent Netflix special). This situation has left edibles processors and analytical laboratories scrambling to find and address the sources of variability in edibles potency.

EVIO Labs is accredited by the state of Oregon to perform potency testing and we pride ourselves on providing consistent results of the highest quality. To address the mentioned edibles potency concerns, we started asking questions. Questions like: What are the sources of variability in edibles potency? Are these tales of edibles disasters an indicator of product variability or consumer inexperience? Can we, as laboratories, successfully measure all the active THC from an edible? Can we improve our testing techniques?

We know that the concentration of active THC in edibles is extremely low. In Oregon, the recreational limit for cannabinoid infused ingestibles is 5 mg of THC per package. Therefore, if the total weight of the ingestible is 15g (pretty standard for gummies and chocolates), the target THC concentration of THC within the product is 0.33 mg of THC per gram of product. We suggested that this low dosage may be in part to blame for variability in at provides the analytical lab with a very low tolerance for variability of results. To put this in perspective, a 15mg piece dosed with concentrate of average 70% THC would need 0.47 mg (0.00047 g) of concentrate for a dosage of 5 mg.

In the life cycle of testing, the plant material is tested for potency, which determines if it will be processed into cannabinoid extract. Once extracted, the concentrated cannabinoids can be processed into a format to introduce it to the edible process. Finally, the edible itself is tested. At each step, errors and non-homogeneity of the samples will affect the final product.


EVIO Labs examined 2 sets of sample variation: between units and within units. This allowed us to dive into the analytical uncertainty and sources of variability around edible extractions. To determine these sources of variability, we utilized 6 different edible systems (3 model, 3 commercial) and 4 solid-liquid extraction techniques. We determined the most appropriate extraction for each sample type and examined the inter and intra variability within the commercial sample sets.

We analyzed chocolate, gummies, and baked goods (cookies and brownies) as our commercial products. From sample to sample, the gummies were the most variable. The cookie was the least variable. We then examined each sample type in fractions. Overall, increasing the sample fraction (i.e. decreasing the sample size) increased the %RSD (Relative Standard Deviation) between the samples. All samples were analyzed in triplicate.


We developed model systems to specifically replicate fat, protein, and sugar systems that one might see in an edible matrix. Each was dosed in a controlled lab environment and expected recovery was calculated. The model systems were extracted using 4 different extraction methods (MeOH, Quechers, 1:3 H2O:MeOH, n-hexane) and analyzed with our validated EVIO cannabinoids method. Sugar showed great recovery across all 4 extraction types. Protein did well with all extraction types expect hexane. Fat had adequate recovery with MeOH and Quechers extractions, had excellent recovery with the hexane extraction and had terrible recovery with the 3:1 MeOH:H2O extraction.


Finally, we move on to our examination of uncertainty. Initially, variability within the unit was included – however laboratories can take steps to decrease that through homogenizing the sample or extracting it entirely. Recovery values from the extraction process are < 10% in each calculation, and analytical uncertainty is < 0.1%. Overall, the biggest contributors to sources of uncertainty in edibles are the production methods and the variability unit to unit.