Cannabis testing only works if the number on the certificate is the number the instrument produced. That sounds obvious. It has also been one of the sector’s most persistent problems, and it is worth understanding why — because the causes are structural rather than a matter of a few bad operators.
The Incentive Problem at the Centre of It
In most industries, the party being assessed does not choose and pay the assessor. In cannabis, it does. Producers select the laboratory, pay the laboratory, and can move to a different laboratory if they dislike the outcome.
That arrangement creates pressure in exactly one direction. A laboratory that reports higher potency figures is more attractive to clients whose products compete on the THC number. A laboratory with a stricter contaminant posture loses work to one with a looser interpretation. Neither dynamic requires anyone to falsify anything — it only requires clients to be free to shop, and results to differ enough that shopping pays.
This is usually called lab shopping, and it has been documented across multiple mature markets. The consequences land on people who never see the certificate: patients paying for potency that is not there, and consumers assuming a contaminant panel was as rigorous as it looked.
What Potency Inflation Actually Looks Like
Reported THC figures can drift upward through several mechanisms, only some of which are deliberate.
Sampling. If the producer selects the material submitted, the sample can be the best material rather than representative material. This is why regulated markets increasingly require laboratory or third-party sampling.
Method choice. Different extraction solvents, dilution schemes and calibration approaches yield different recoveries. Choices that are individually defensible can, in combination, sit consistently at the top of the plausible range.
Integration decisions. Chromatographic peaks require judgement about where they start and end. Small, systematic choices in how peaks are integrated move results measurably.
Outright misreporting. Rarest and most serious, and the reason regulators run blind proficiency programmes.
Most inflation is not fraud. It is accumulated small choices in an environment where one direction is rewarded. That makes it harder to detect and harder to fix.
What Data Integrity Means in Practice
Data integrity is the principle that a result can be traced back to the measurement that produced it, without gaps and without undocumented changes. It rests on a handful of concrete controls.
- Audit trails. Instrument and information systems record who did what and when. Changes to data are logged with a reason, not overwritten.
- Independent review. Results are checked by someone other than the analyst who produced them, before release.
- Controls in every run. Calibration standards, blanks and spiked samples confirm the method is behaving. If controls fail, results are not reported.
- No post-hoc editing. A released certificate is amended through a documented revision, not quietly replaced.
- Retention. Raw instrument data and sample retains are kept long enough that a result can be re-examined.
None of this is exotic. It is standard practice in pharmaceutical, environmental and food laboratories, and it is what ISO/IEC 17025 accreditation assesses.
Client Data Is a Separate Obligation
Laboratories accumulate commercially sensitive information: yields, potency trends, failure rates, batch volumes, product development work that has not launched. Some of it would be valuable to a competitor and damaging in the wrong hands.
The obligations that follow are ordinary but easy to neglect — restricting internal access to what a role requires, not discussing one client’s results with another, being explicit in contracts about what is shared with regulators and what is not, and treating aggregated or anonymised industry data as something clients should knowingly consent to rather than something the laboratory simply does.
There is a real tension here. Aggregate testing data is genuinely useful — for research, for spotting contamination patterns across a region, for regulators trying to understand where failures cluster. That value does not override the individual client’s expectation of confidentiality; it has to be negotiated rather than assumed.
Reliability Is Not Only About Honesty
An entirely honest laboratory can still produce unreliable results. Under-validated methods, instruments outside calibration, poorly trained analysts, samples handled badly on intake — none of these involve bad intent, and all of them produce numbers that do not reflect the material.
This is why proficiency testing matters so much. Blind samples with known values are the only mechanism that reveals whether a laboratory’s results are accurate, as opposed to internally consistent. Consistency without accuracy is a laboratory reliably reporting the same wrong answer. More on where methods genuinely struggle in the limits of current cannabis testing methods.
What Actually Reduces the Problem
Several measures have made a measurable difference in markets that adopted them.
Independent sampling. Removing sample selection from the producer eliminates the easiest route to a flattering result.
Blind proficiency testing run by the regulator. Samples submitted without the laboratory knowing they are being assessed. Voluntary programmes are useful; mandatory blind programmes are what catch systematic drift.
Standardised methods. Where regulators prescribe methods rather than outcomes, the room for defensible-but-convenient choices narrows. Progress here is covered in standardisation and transparent reporting.
Published enforcement. When suspensions and corrective actions are public, buyers can factor them in. Enforcement conducted privately does not change purchasing behaviour.
Buyers who ask. Dispensaries and processors that request accreditation scope, detection limits and proficiency results shift the demand side. As long as the market rewards only the highest number, the pressure remains.
What Operators Can Do
If you are selecting a laboratory, the ethical posture is assessable before you commit. Ask about proficiency testing participation and results. Ask who performs data review. Ask what happens when a result is disputed. Ask whether raw data is retained and for how long.
Be wary of any laboratory that markets on outcomes rather than capability. And be wary of your own reasoning if you find yourself moving laboratories after an unwelcome result — that is the exact behaviour the incentive problem depends on.
The full selection checklist is in how to choose a cannabis testing laboratory. The short version: a laboratory that is comfortable being audited is the one whose numbers are worth having.