Potency, LOD & assay adjustmentscomputed at the scale, signed at the kiosk, printed on the eBMR.
V5 Ultimate stacks up to 11 LIMS-driven factor types — assay, potency, purity, moisture, LOD, water, activity, concentration, dilution, overage and prior-step yield — into a deterministic adjusted dispense target. Version-pinned to the formula revision. E-signature-gated. Rendered verbatim on the batch record.
The scale-side kiosk IS the potency calculation.
Watch a real dispense: CoA parsed, factors stacked, target computed, weight bounded, signatures sealed. The math is visible; the chemist never types it.
Four factors.One order. Fixed in code.
A spreadsheet lets an operator reorder SBF, basis, PF and LOD and get a different number every time. V5 fixes the order in the recipe engine — the same four factors, the same order, every batch, every site. If a factor is inactive it resolves to 1.000 and is still rendered on the audit trail. Never missing. Never inferred.
The whole calculation, in six lines the operator can read.
No hidden math. No collapsed spreadsheet. Every intermediate value is stored against the batch and printed on the eBMR — the same arithmetic the operator saw, the QA reviewer sees, and the FDA investigator reconstructs.
Deterministic. Order fixed in code. Version-pinned to the formula revision. Replayable from the eBMR ten years later.
- On the kiosk before any material moves
- On the printed reference label at the dispense booth
- On the accepted-weight confirmation screen
- On the batch record (eBMR/eDHR) at review
- In the QA reviewer's exception queue if any factor is missing
If the item master doesn't have LOD attribute typed for this ingredient, V5 refuses the dispense with a routed exception — it does not silently substitute 1.000 and continue.
Every factor is a first-class attribute, not a spreadsheet cell.
Each factor lives on the lot with method, analyst, instrument, CoA reference and e-signature. Recipes declare which factors are active per ingredient; the MES enforces the stack; the eBMR renders every value — including the ones that resolved to 1.
% active vs label claim — most common potency driver in pharma & nutra.
Biological / cannabinoid potency (mg/g, IU/mg) applied to APIs, oils, extracts.
Chromatographic purity used for chiral, isomeric or reference-standard adjustments.
Karl Fischer water content for hygroscopic actives and excipients.
USP <731> loss-on-drying for solids where water + volatiles matter.
Distinct from LOD when the method separates bound vs free water.
Enzyme / probiotic activity (U/mg, CFU/g) for biologicals and live-culture inputs.
For liquid actives (mg/mL) so mass adjusts against measured strength.
Blend / stock dilution factor for pre-standardised intermediates.
Approved overage (%) baked into the formula, applied deterministically.
Prior-step in-process yield factor pulled from the preceding batch record.
CoA → LIMS → MES → kiosk → eBMR — one signed chain.
LIMS captures the CoA
Supplier CoA (paper, PDF, or portal upload) is OCR'd or keyed into LIMS. Assay, potency, moisture, LOD and any other factor are stored against the incoming lot with method, analyst, instrument and e-signature.
MES stacks the factors
At WO release, MES reads every factor flagged 'active' for that ingredient and computes an adjusted target: target = spec × Π f(factor). Deterministic, version-pinned to the formula revision, stored with the batch — replayable years later.
Kiosk shows the adjusted number
The operator sees the ORIGINAL spec, the LIMS lot(s) selected, each active factor with its value, and the ADJUSTED dispense target — all before any material moves. The scale is bounded to the adjusted target ± tolerance, not the raw spec.
Two-sig approval + immutable log
For regulated formulas, the adjusted target requires a second signature (QA or supervisor). Every factor, every value, every signer, every timestamp is captured in the 21 CFR Part 11 audit trail. eBMR/eDHR prints the calculation as evidence.
Audit-ready record
The batch record contains: spec, lot(s), CoA reference, each factor and its value, combined factor, adjusted target, actual weight, delta, tolerance band, operator, verifier, timestamps, hash chain. Reviewers reconstruct the math without leaving the record.
Reversal and re-work
If a late assay result invalidates a dispense, V5 doesn't scrap silently — it opens a signed reversal, holds the affected batch, and routes CAPA. The original event stays in the audit trail; the correction is a new signed event.
The adjusted target is the single source of truth — everything binds to it.
LIMS, scales, ERP, doc control, maintenance and the eBMR read from or write to the same batch record in real time. No spreadsheets in the middle. No parallel copies to reconcile.
Assay, potency, moisture, LOD posted per lot with method, analyst, instrument, e-sig. CoA image attached.
Mettler, Sartorius, A&D — BT, USB HID, serial, Ethernet/IP. USP<41> minimum-weight rule enforced at binding.
SBF, MW ratios, salt stoichiometry, basis defaults typed on each material — versioned, not free-text.
Consumption posted against adjusted target; yield variance closes to ERP at batch close — not on paper.
Effective MMR revision fetched at runtime — operators never dispense against yesterday's recipe.
Out-of-cal balance or overdue PM blocks the dispense at the kiosk — refuses to write to the regulated step.
The edge cases real dispensing lives in.
The happy path is easy. What matters is how the system behaves when a CoA is late, a lot runs out, LIMS is down, or a recipe rev lands mid-campaign. V5 treats each as a first-class workflow.
CoA arrives late (after dispense)
Batch is held automatically. If the late assay invalidates the adjusted target, V5 opens a signed reversal, routes to QA, and blocks release until the disposition is signed. The original event stays; the correction is a new event.
Two lots to make one dispense
Operator splits across two approved lots; V5 computes per-lot adjusted contribution and rolls them into one blended target. Both CoAs, both factor stacks, both weights bind to the same step.
LIMS unreachable at dispense
Kiosk executes against the last-known-good CoA snapshot cached at WO release. If no snapshot exists the step is blocked, not defaulted. On reconnect, values reconcile; a mismatch triggers a signed exception.
Stacked corrections on a blend
For a pre-blended intermediate, the dilution factor cascades on top of assay + LOD of the base API. V5 records the full lineage — every factor from every parent lot appears on the eBMR, no black boxes.
Recipe changes mid-campaign
The batch is bound to the MMR revision at WO release. Later MMR changes don't rewrite an in-flight batch. Next batch uses the new revision; the previous batch replays exactly as it ran.
Alternate factor definitions per market
USP LOD, EP water, JP moisture — some products dispense with a different factor definition per market. V5 lets the recipe select which factor to consume; the same lot can back both without a parallel item master.
The numbers that show up on the quarterly review.
Pilot ranges from V5 Ultimate customers. Tied to what the potency engine actually controls — not marketing air.
Questions QA leaders ask before they buy.
Keep digging on the potency & dispense stack.
Step-by-step numerical walkthrough of the same metoprolol dispense, top to bottom.
ReadWhen to use which — and how V5 handles both without a parallel item master.
ReadWhere SBF lives on the item master and how V5 versions MW ratios.
ReadThe deterministic way to bake approved overage into the recipe.
ReadThe full kiosk workflow the potency engine plugs into.
ReadHow V5 scales a recipe up or down without losing factor fidelity.
ReadThe kiosk-first execution layer this all runs on.
ReadWhere the four-factor stack is rendered on the batch record.
ReadStop re-calculating in Excel.
Included in every V5 Ultimate tier from Express upward. See the four-factor stack on a live workspace with your own SOP in under 20 minutes.