Potency Isn’t Just a COA: Using In-House Testing to Control Distillation Cuts and Drying Endpoints

In regulated production, the Certificate of Analysis (COA) is the scoreboard—but it’s rarely the steering wheel. By the time a third-party lab report arrives, the batch has already been distilled, dried, packed, or shipped. If the potency is off-target, you’re left with expensive options: rework, blending, discounting, or scrapping.

The operators running your evaporators, wiped-film systems, and vacuum ovens need feedback during the run—fast enough to guide cut points, decide whether a fraction is worth reprocessing, and stop drying at the right endpoint before you burn time (and yield).

That’s where in-house potency for process control earns its keep.

Below is a production-optimization approach to in-house potency testing—designed to reduce rework and improve yield—plus a practical workflow (sampling plan, chain-of-custody lite, acceptance criteria, and feedback loops) you can implement without turning your facility into a full analytical lab.

---

Why “potency as a control signal” changes the economics

When potency is treated as a periodic compliance test, teams tend to over-process “just to be safe.” The result is predictable:

• Late distillation cuts to avoid cross-contamination → lower throughput and more degradation risk.
• Over-drying to avoid residual solvent failures → longer cycle times and more loss of volatiles.
• Excessive re-runs because operators can’t quantify whether a fraction is salvageable.

Using quick in-house potency reads as an in-process control enables decisions that are both faster and more consistent:

• Confirm when the “heads” fraction has cleared and it’s safe to move into the main fraction.
• Identify when the “tails” are creeping in and the economics of the run change.
• Decide whether a questionable fraction should be re-run immediately or staged for blending.
• Stop vacuum drying when the curve flattens (instead of “one more hour”).

The biggest ROI usually comes from rework avoidance and cycle-time compression, not from replacing third-party labs.

---

Method harmonization: comparability beats perfection

A common objection is: “In-house results don’t match the outside lab.” That’s often true, and it’s also often fixable.

The goal for process control is comparability over time—within your site, across shifts, and eventually across multiple sites—so operators can rely on trends and triggers.

Two important industry anchors support that direction:

• AOAC’s Cannabis Analytical Science Program (CASP) has been driving method performance requirements and standardization conversations in the cannabis testing ecosystem—pushing the industry toward comparable results and proficiency testing expectations rather than ad hoc methods. Source: https://www.aoac.org/scientific-solutions/casp/

• NIST Reference Material 8210 Hemp Plant (RM 8210) provides cannabinoid values (and other analytes) in a controlled matrix to help labs with method validation/verification and routine QC benchmarking. This is exactly the type of reference material that improves long-term comparability and helps you detect drift. Source: https://www.nist.gov/news-events/news/2024/07/rm-measuring-cannabinoids-and-toxic-elements-hemp

For production teams, the takeaway is practical: build an internal potency program that is repeatable, trended, and anchored to QC checks. If you do that, your in-house data becomes trustworthy enough to guide operations—even if it’s not the same as every external COA.

---

The workcell mindset: treat potency like a utility

The most successful deployments don’t put the analyzer “in the lab.” They create a testing workcell that sits close to production, with clear ownership and simple rules.

A good workcell has:

• Defined sample points (where/when samples are pulled)
• Defined sample prep (how samples are diluted/handled)
• Defined acceptance criteria (what results trigger action)
• Defined feedback loops (who gets notified and how the process changes)
• Defined QC checks (how you know the analyzer is still telling the truth)

This is exactly where a compact, guided workflow analyzer can shine.

---

Product plug: a practical analyzer for in-house potency workflow

If your goal is fast potency reads to guide production decisions, the Orange Photonics LightLab 3 is purpose-built for that type of deployment.

Recommended gear: https://www.urthandfyre.com/equipment-listings/orange-photonics-lightlab-3-cannabis-analyzer---potency-testing-lab-

Urth & Fyre’s listing includes key details and photos, and we can help you design the surrounding workcell—training, sampling SOPs, QC cadence, and ongoing service/calibration connections—so the analyzer stays trustworthy and useful.

(Manufacturer background: LightLab 3 is a fit-for-purpose HPLC-based analyzer designed to simplify potency testing workflows for non-analytical operators. More info: https://orangephotonics.com/lightlab-3/)

---

Where in-house potency testing impacts production the most

1) Distillation cut control (short-path / wiped-film)

Distillation decisions are often made with indirect signals:

• Temperature
• Vacuum level
• Color/clarity
• Flow rate
• Time-on-cut

These signals matter, but they can drift with feed variability. A quick potency read provides an additional signal to prevent two common losses:

• Cut contamination: potency/spec drift from heads/tails mixing into the main fraction.
• Over-processing: extending the run because “it might still be good,” when potency data shows diminishing returns.

Operational pattern to implement:

• Pull small, consistent samples from each fraction window (heads, main, tails).
• Measure potency quickly.
• Trend the values run-to-run.
• Use pre-defined triggers to switch cuts.

The objective isn’t to “certify” the product in-house. It’s to decide when to move the valve.

2) Re-run decisions (economic triage)

Reprocessing can be necessary, but it should be intentional.

In-house potency can support a simple triage model:

• Green: meets internal spec → release to next step.
• Yellow: borderline → hold for blend strategy or targeted re-run.
• Red: clearly off-spec → re-run immediately or reject.

When you don’t have fast data, “yellow” gets treated as “red,” which creates hidden capacity loss.

3) Vacuum oven drying endpoints (stop when the curve flattens)

Vacuum drying is often run on “fixed time recipes,” which are vulnerable to variability in:

• Solvent load
• Tray loading depth
• Material viscosity
• Shelf spacing and thermal gradients
• Pump performance and leaks

Instead of time-only endpoints, use a blended endpoint strategy:

• Time minimum (don’t open early)
• Weight stabilization (mass change per hour)
• Potency stabilization (potency trending stops improving)

Potency changes can signal whether you’re still removing volatiles/solvent or just burning hours.

If you’re running larger vacuum ovens, pay attention to uniform heating and stable vacuum integrity—both have outsized impact on drying repeatability and “false endpoints.”

If you’re building out post-processing capacity, you can also explore Urth & Fyre’s extraction and post-processing equipment category pages for complementary gear: https://www.urthandfyre.com

---

A production-ready workflow design (SOP-level)

Below is a practical framework you can implement in 2–6 weeks depending on staffing and training.

Step 1: Define the purpose and scope (what decisions will potency drive?)

Write down the decisions you want to make faster. Examples:

• Switch from heads to main fraction when in-house potency is within X% of target.
• Stop collecting main fraction when potency drops below Y%.
• Re-run only if potency is within a salvage band (e.g., between A and B).
• Stop vacuum drying when potency improvement over 2 hours is below Z.

Keep it tight. If you try to do everything, you’ll do nothing.

Step 2: Sampling plan (simple, repeatable, defensible)

A strong sampling plan focuses on consistency more than complexity.

Define sample points by process:

• Feed: one sample per feed tank/drum (start + mid if large).
• Fractions: one sample at the start of each cut window, then every N minutes or every N liters.
• Drying: one sample at baseline (pre-oven), then at fixed intervals (e.g., 2-hour marks) until stabilization.

Define sample size and container:

• Use a consistent mass/volume (e.g., 0.2–1.0 g) and consistent vials.
• Label every vial with: date/time, batch, unit operation, operator initials.

Define sample handling time:

• Maximum time from pull-to-test (e.g., 30 minutes) to reduce drift and confusion.

Step 3: Chain-of-custody lite (enough traceability to trust the data)

You don’t need full eQMS to run strong in-process controls.

Implement a chain-of-custody lite log:

• Sample ID (unique)
• Batch/lot
• Process step and location
• Time pulled / time tested
• Operator who pulled / operator who tested
• Any deviations (hot sample, wrong container, spill, re-dilution)

This solves a real production problem: it prevents “mystery numbers” that no one can reproduce.

Step 4: Acceptance criteria (internal specs for control, not for labeling)

Use internal specs that are tight enough to guide decisions but realistic for in-process variability.

Examples:

• Cut switch trigger: main fraction begins when potency is within ±X% of the historical mean for “good main.”
• Tails trigger: stop main fraction when potency drops by Y% from peak or falls below a floor.
• Drying stop: stop when potency changes less than Z% across two consecutive sample points.

These should be based on your own historical data. Start with conservative bands, then tighten after 10–30 batches.

Step 5: Feedback loops (make the data operational)

Results that live on a clipboard don’t control anything.

Define who gets notified and what they do:

• If a fraction potency is off → operator switches cut / supervisor confirms.
• If a drying trend stalls → operator ends cycle / QA confirms hold status.
• If analyzer QC fails → stop using results for decisions until resolved.

Practical implementation:

• Set up a shared channel (Teams/Slack) with a standard message template.
• Post sample ID + result + decision recommendation.
• Train operators to request confirmation when results are borderline.

---

Keeping the analyzer trustworthy: QC cadence that production teams can live with

A process-control program fails when the analyzer drifts and nobody notices.

Build a lightweight QC routine:

• Daily start-up check: run a known control material or check standard; verify it lands inside a defined window.
• Mid-shift check (high-throughput sites): catch drift early.
• After maintenance: re-verify before releasing results.
• Trend QC: plot control results over time and define “action limits.”

Tie your QC approach to harmonization anchors where possible (e.g., reference materials like NIST RM 8210, and method standardization efforts like AOAC CASP). Even if you don’t use the exact same methods as a third-party lab, your program becomes more stable and auditable when it’s benchmarked.

---

ROI and implementation timelines (what most facilities see)

Every site is different, but realistic operational gains from in-house potency for process control commonly come from:

• Reduced rework: fewer unnecessary re-runs because borderline material is identified and routed intentionally.
• Improved yield protection: fewer overcooked cuts and fewer over-dried batches.
• Shorter cycle times: drying and distillation runs end when data says “done,” not when the schedule says “maybe.”

A typical rollout:

• Week 1: define decisions, sampling points, and acceptance criteria draft.
• Week 2: set up the workcell, train two “super users,” and run parallel testing vs. your current approach.
• Weeks 3–4: tighten SOPs, implement chain-of-custody lite, start trending.
• Weeks 5–6: lock in triggers for cut points and drying endpoints; add QC cadence and service plan.

The key is to treat this as a production system, not a lab project.

---

Common failure modes (and how to avoid them)

Failure mode 1: “The numbers don’t match the COA, so we ignore them.”

Fix: define the goal as internal comparability and decision support, then anchor with QC and trending.

Failure mode 2: Sampling is inconsistent.

Fix: standardize sample size, location, timing, and labeling. Audit sampling for one week.

Failure mode 3: Results don’t reach the operator fast enough.

Fix: place the workcell near production, simplify prep, and define notification pathways.

Failure mode 4: No one owns the analyzer.

Fix: assign a primary owner (operations or QA) plus a backup, and schedule QC checks.

---

How Urth & Fyre helps you operationalize in-house potency

Buying an analyzer is the easy part. Making it a reliable production control system is where teams win or stall.

Urth & Fyre can support:

• Workcell design: layout, utilities, consumables, and contamination control
• SOP development: sampling plan, chain-of-custody lite, acceptance criteria, deviation handling
• Operator training: super-user model, competency checks, shift handoff routines
• Calibration/service connections: keeping the analyzer stable and minimizing downtime
• Process optimization: translating potency trends into cut-point logic, re-run rules, and drying endpoints

If you’re ready to turn potency into a control signal (not just a COA), explore current equipment listings and consulting support at https://www.urthandfyre.com.