The Sampling Chain: One Lab, One Sample, and the End of Lab-Shopping (Proposed DCC Track-and-Trace Rules, 4 CCR §§ 15304–15710)
Status: Part of the DCC’s proposed track-and-trace overhaul, DCC-2026-02-R (noticed June 5, 2026) — proposed amendments to the testing and sampling sections (§§ 15304–15710). Not yet adopted law. Public comment is open. [1]
Of the six changes in the proposed track-and-trace overhaul, this is the one that decides whether a Certificate of Analysis (COA) actually means what everyone downstream assumes it means. For California cannabis manufacturers and distributors, the rules are tightening around a question the system has left open: how many times, and at how many laboratories, a batch can be sampled before a result becomes official.
Where Track-and-Trace Stands Today
The mechanics of collecting a sample are already disciplined, and they are not what is changing. Under § 15304, after a distributor takes physical possession of a batch, a laboratory employee comes to the distributor’s licensed premises to select a representative sample. Under § 15305, the distributor must be physically present to observe it, the collection must be video-recorded (and the footage kept for 90 days), and both the distributor and the laboratory employee sign a chain-of-custody form. [3]
What today’s rules leave open is how many laboratories and how many samples. The word “a” laboratory in § 15304 has been read as a floor, not a ceiling — and the CCTT system does not limit how many test-sample packages can be created from a single batch. By the Department’s own account, some distributors exploit that: they have multiple laboratories sample the same batch, collect informal results, select the most favorable, have that laboratory upload its result, and void the other manifests so the unfavorable results never post. [2] The industry calls it “lab shopping,” and it pits laboratories against each other to return the most favorable numbers — rewarding potency inflation and disadvantaging the labs that play it straight.
What the Proposed DCC Rule Changes
The draft closes the loophole at four points. [1]
One lab, one sample (§ 15304). A distributor must arrange for “only one” licensed laboratory to obtain “one” sample of a batch, and the CCTT system will block creating more than one test-sample package — except for a Department-approved retest under § 15705(g). [1]
Recorded before the lab leaves (§ 15305, new subsection (f)). The distributor must record the sampling in track-and-trace before the laboratory employee leaves the premises. Today, § 15049(b)(1) allows 24 hours — enough time for a round-the-clock lab to feed back a preliminary result before the sampling is even logged. Closing that window removes the head start lab-shopping depends on. [2]
Sampling restrictions (§ 15305.1). The section formerly titled “Resampling” becomes “Sampling Restrictions”: once a representative sample has been taken, the distributor may not create a second test-sample package, or arrange for another laboratory to sample or re-sample the same batch, unless § 15705(g) is met. [1]
A tighter gate at the lab (§ 15710). The laboratory employee who receives the sample must sign the chain of custody, and the laboratory may not analyze the sample — and the batch may not be released for retail sale — if the COC is missing, the tamper-evident seal was broken before the sample arrived, or there is evidence of commingling, contamination, or degradation. [1]
The § 15705(g) exception is narrow by design: it lets a distributor who requests and obtains Department approval have a batch re-sampled by another laboratory only when the first laboratory cannot competently complete the testing. [2]
Comparing the Track-and-Trace Workflows
| Workflow element | Current | Proposed DCC-2026-02-R |
|---|---|---|
| Labs & samples per batch | “A” lab / “a” sample, read as a floor; CCTT allows multiple test-sample packages. | “Only one” lab, “one” sample; CCTT blocks extra packages (except § 15705(g)). |
| Recording the sampling | Up to 24 hours (§ 15049(b)(1)). | Before the lab employee leaves the premises (§ 15305(f)). |
| Re-sampling | “Resampling.” | “Sampling Restrictions” — no second package, no other-lab re-sample, except § 15705(g). |
| Collection itself | Lab employee on-site, distributor present, recorded 90 days, both sign COC. | Unchanged. |
| Lab receipt | Accept with a COC. | Receiving lab employee signs COC; no analysis and no retail release if COC missing, seal broken, or contamination. |
Regulations Are Optimizers in Disguise
A regulation is a forcing function. The old story is that more sampling discipline slows the business down. The Freed Up frame is that rules reveal what the business was already missing.
When a result can be shopped, every COA carries a quiet asterisk. When it can’t, a passing COA means one thing. Closing the lab-shopping loophole doesn’t just satisfy the regulator — it protects the value of a clean batch and the brand attached to it.
A result you can’t shop is a result the market can finally trust.
For the operator, one documented sampling event per batch is fundamentally cheaper than the re-tests, reconciliation, and disputes that a loose chain invites. Fixing the audit risk and fixing the unit economics are the same project.
The Standing Practice for Clean Testing
The operators who will barely feel this proposed shift are the ones already running a single, clean sampling event and logging it on the spot.
The standing practice is simple: arrange one accredited laboratory per batch, have your distributor present for the collection, record the sampling in track-and-trace before the laboratory employee leaves, keep the signed chain-of-custody form with the batch record, and never build a financial plan around a second bite at the result. An operation already keeping the clean batch records we mapped in No. 01 and the inspection-ready standing file of No. 03 has most of this muscle already.
Weighing In on the Public Record
Because this is still a proposed rule, the final framework is not yet set — we do not know what the final adopted text will look like. If recording the sampling before the laboratory employee leaves is impractical for the way your batches actually get tested, or if the § 15705(g) retest path is too narrow for a genuine first-lab failure, that is precisely the operating reality the public record is built to capture. Written comment is open through July 20, 2026, with a hearing on July 21. [1]