Unit price is the wrong denominator, and the error compounds

Last quarter I sat with a lab manager who was proud of a $0.31 per vial price on 10 mL clear screw-thread vials. Her cost per reported result was $4.87. The gap was not freight or duty. It was 14% reruns on a method running 60 samples a day, a 3% septum-coring reject rate on the autosampler, and 40 minutes of analyst time per batch spent re-capping and re-racking. Once you do the vial supplier cost per test calculation properly, the unit price stops being the interesting number. The interesting number is what each reported result costs you after the vials have done their job or failed to.

I have run this calculation for about forty labs over the last six years, and the pattern is depressingly consistent. The lab that negotiates hardest on unit price usually has the highest per-test cost, because the cheapest vial is the one that reruns. The lab that pays $0.28 and never thinks about it usually has a per-test cost under $2. The difference is not skill. It is which denominator the buyer was looking at when the purchase order was signed.

Procurement teams are trained to compare line items. A vial is a line item. A reported result is not. The moment you move the denominator from “vial purchased” to “result reported,” every hidden cost in the workflow attaches itself to the vial, because the vial is the only consumable that touches the sample at every stage from receipt to injection.

Here is the arithmetic I use. Build a per-test cost from four blocks: consumable cost per injection, rerun cost per injection, reject cost per injection, and labor cost per injection. The first block is the only one a quote controls. The other three are set by the vial’s physical behavior on your instrument, and they are usually two to four times larger than the first.

The formula, in its plainest form:

• Cost per reported result = (vial unit price ÷ usable vials per lot) + (rerun rate × cost of one rerun) + (reject rate × cost of one reject) + (prep minutes per vial ÷ 60 × loaded labor rate)

“Usable vials per lot” is not the count on the carton. It is the count after you subtract vials that fail incoming inspection, vials that jam the tray, and vials you discard because the cap would not seat. On a good lot that is 98 to 99%. On a bad lot it can be 88%. That single correction moves unit price by more than most negotiated discounts. On a 5,000-vial lot, an 88% usable rate means you effectively paid 13.6% more per usable vial than the invoice says, and that is before a single injection.

The compounding effect is what most buyers miss. A vial that reruns at 10% does not just cost you the rerun. It costs you the instrument time that the rerun displaced, which delays the next batch, which pushes the analyst into overtime, which shows up in a labor line three cost centers away. I have seen a $0.09 unit price increase eliminate a $14,000 annual overtime line. Nobody in procurement connected the two, because the two numbers lived in different spreadsheets.

The two grades that actually show up in a quote, and what separates them

When a vial supplier quotes you two prices, they are almost never quoting the same product with a margin difference. They are quoting two manufacturing routes. I call them Grade A (certified, lot-traceable, dimensional tolerance held) and Grade B (commercial, certificate of conformance only). The physical differences are small on paper and enormous in the autosampler.

The Grade B vial looks like a 50% saving. It is a 91% cost increase per reported result. The table is not hypothetical; the labor and rerun figures are the ones I use for a mid-size pharmaceutical QC lab running a single HPLC per analyst.

Run the table with your own numbers and the gap widens or narrows but rarely closes. A lab with a $45 per hour loaded rate and a 15-minute method gets a smaller absolute gap but a similar ratio. A lab running LC-MS with a 30-minute method and a $12 per hour instrument charge sees the gap nearly double, because the rerun cost block grows and the unit price block does not.

Where the rerun cost comes from

A rerun is not just another injection. On a 20-minute HPLC method with a 10-minute equilibration and a 5-minute blank, one rerun consumes 35 to 45 minutes of instrument time and 4 to 6 minutes of analyst time. Loaded instrument time at $12 to $18 per hour and analyst time at $65 per hour gives you a rerun cost of roughly $6 to $8. If the rerun is triggered by a ghost peak from a septum extractable, you also lose the sample if the vial was the only one drawn. That is a resample, and a resample can cost a full day.

On GC-MS the number is worse. A 35-minute temperature program with a 10-minute cool-down and a 5-minute bake-out is 50 minutes of instrument time. A rerun at that scale costs $9 to $11 once you load the instrument charge. If the rerun is caused by a septum bleed peak that coelutes with a target analyte, the analyst may not even catch it on the first pass, and the bad result reaches the data system. That is not a rerun. That is a reportable error, and the cost is a deviation, a CAPA, and a client conversation.

The rerun rate is the single largest lever in the whole calculation. A 1% swing in rerun rate on a 60-injection batch at $7 per rerun is $4.20 per batch, which sounds trivial until you multiply by 250 batches a year and realize it is $1,050 for a 1% swing. A 10% swing is $10,500. That is the number worth negotiating against, not the $0.14 unit price gap.

How rejects and prep time enter the same equation

Rejects are different from reruns. A rerun is a failed injection that you repeat. A reject is a vial you cannot even load, or a result you throw out before it reaches the data system. Septum coring is the classic cause. A 23G needle punching a 50 Shore A septum with poor cross-link density will core on the third or fourth puncture, and the silicone fragment lands in the sample. You see it as a spike in the chromatogram at a retention time that does not match any standard, or as a blocked needle on the next injection.

The blocked needle is the expensive version. A needle that cores on vial 40 of a 60-vial batch does not just lose vial 40. It loses the rest of the batch while the analyst pulls the needle, inspects it, replaces it, and re-runs the calibration curve because the needle change is a system suitability event. That is a 90-minute recovery and a $180 to $250 cost for one cored septum. A 3% coring rate on 250 batches a year is roughly 450 events, and even if only one in five blocks the needle, that is 90 recovery events and $18,000 to $22,000 per year from a defect that costs the supplier nothing to fix.

Prep time is the quiet one. Uncapping, visually inspecting for particulates, recapping, and racking a 2 mL vial takes 18 seconds on a well-molded Grade A lot and 26 seconds on a Grade B lot where the caps do not seat cleanly. On a 60-vial batch that is 8 minutes versus 26 minutes. Across 250 batches a year, that is 75 hours of analyst time for the same work. At $65 per hour, $4,875 of labor that never appears on a vial quote.

If you are running LC-MS and your prep includes a filtration step, the filter choice compounds the same way. A HPLC filter that sheds particles or binds your analyte adds reruns at the same rate as a bad septum. When I audit a lab’s per-test cost, I check the filter and the vial together, because they fail in the same place: the sample path. A PTFE membrane with a 0.45 µm pore and a polypropylene housing is the default for aqueous and most organic mobile phases, but if the housing is not specified you can get extractables from the housing itself that show up as ghost peaks at 3 to 6 minutes. I have seen a lab chase a septum problem for two weeks when the culprit was a filter housing lot. The PTFE membrane filter specification matters as much as the vial specification when you are calculating per-test cost, and the two belong in the same audit.

Two field cases where the per-test number told the truth

Case 1: the 14% rerun rate that was a durometer problem

Symptom. A contract lab running EPA method 8270 on a GC-MS saw rerun rates climb from 2% to 14% over three lots. The unit price had dropped 22% when they switched suppliers six months earlier. Nobody connected the two.

Measurement. I pulled the injection logs and plotted rerun rate against lot number. The rate tracked lot, not analyst, not instrument, not method. I then measured septum durometer on retained samples from each lot with a handheld Shore A gauge. The bad lots read 54 to 58 Shore A against a spec of 45 ± 3. The harder septum did not seal cleanly against the crimp, and the inlet pressure pushed sample vapor past the seal on injection. The result was a low-area peak with a drifting retention time, which the analyst flagged as a rerun.

Fix. Moved to a Grade A lot with a reported durometer of 45 ± 3 and a lot-specific extractable screen. Rerun rate returned to 1.8% within one lot. The unit price went up $0.09. The per-test cost fell $2.10. If you are running GC methods, the Agilent GC consumables reference is worth keeping open when you specify septa, because the durometer and bleed specs are published per part number.

The follow-on lesson was that the contract lab had no incoming inspection for durometer. They had a visual check and a certificate on file. Adding a $180 handheld gauge and a 10-vial sample per lot cost them $2,200 a year in analyst time and caught the next bad lot before it reached the instrument. That is the cheapest insurance in the whole workflow.

Case 2: the 3% reject rate that was a neck tolerance issue

Symptom. A pharma QC lab running a stability study on an autosampler with a fixed gripper saw a 3% “vial not detected” reject rate. The autosampler had been validated. The method had not changed. The vials had.

Measurement. I measured neck ID on 50 vials from the current lot with a pin gauge. The spec was 6.00 ± 0.15 mm. The lot ran 6.32 to 6.41 mm. The gripper jaws could not close to the reference diameter, so the vial slipped during transfer and the instrument flagged it as absent. The supplier’s certificate of conformance said “conforms to drawing.” The drawing did not state a neck ID tolerance.

Fix. Rewrote the purchase spec to state neck ID at 6.00 ± 0.15 mm and required a first-article dimensional report with the shipment. Reject rate dropped to 0.2%. The spec change cost nothing. The reject rate had been costing the lab roughly $1,900 per year in instrument time and analyst time alone, before counting the delayed stability timepoints.

The stability angle is the one that changes behavior. A 3% reject rate on a timepoint batch of 120 vials means 3 to 4 vials fail on the day of the timepoint. If the sample cannot be re-drawn because the timepoint has passed, the timepoint is lost, and a lost stability timepoint can push a filing back a month. The dollar value of that is not in the per-test formula, but it belongs in the decision. I put a $500 per lost timepoint placeholder on the worksheet and let the lab manager adjust it. Nobody has ever argued it was too high.

A decision checklist for the per-test comparison

Before you compare two quotes, run this list. If a supplier cannot answer items 3 through 6 with a number, you cannot do the per-test calculation and the lower unit price is an illusion.

• 1. Usable vials per lot. Ask for the AQL and defect class on the last three lots. If they cannot produce it, assume 96% usable.
• 2. Rerun rate on your method. Pull the last 500 injections from your CDS and count reruns by lot. Do not use a supplier-supplied average.
• 3. Septum durometer and puncture rating. Demand a Shore A value and a puncture count to coring at your needle gauge (23G or 22G).
• 4. Neck ID and height tolerance. State them on the drawing. ±0.15 mm on neck ID is achievable; ±0.40 mm is what you get if you do not ask.
• 5. Extractable screen. For LC-MS, request a lot-specific extractable profile at your mobile phase and temperature. For GC, request a bleed profile.
• 6. Prep time per vial. Time 20 vials uncap to rack with a stopwatch. Multiply by your loaded labor rate.
• 7. Resample cost. Assign a dollar value to a lost sample. For a stability study it is often $200 to $500 per lost timepoint.
• 8. Needle recovery cost. Estimate the annual cost of cored-septum needle blocks. If you have never measured it, assume one per 200 injections and a $200 recovery cost.

Once you have those eight numbers, the comparison table above fills itself in. The supplier with the higher unit price wins on six of the eight rows in most labs I audit. The two rows where the cheap vial can win are unit price and, occasionally, incoming inspection time, because a cheap vial that you do not inspect looks cheaper until the reruns start.

What the per-test number does not capture, and how to bound it

Two costs sit outside the formula and both are real. The first is method validation impact. If a vial change forces a partial revalidation, that is a fixed cost of $8,000 to $25,000 depending on the method, and it belongs in the decision but not in the per-test number. Amortize it over the expected life of the method, typically three to five years. The second is regulatory exposure. If a lot fails an audit because the extractable profile was not documented, the cost is not measurable in the same units. I bound it by requiring the extractable screen on every lot for regulated work and accepting the unit price premium as insurance.

There is also a floor on prep time you cannot buy your way out of. Even the best vial requires an analyst to uncap, inspect, and rack. The realistic floor is 15 to 18 seconds per vial for a manual workflow. If a supplier claims their vial eliminates prep time, they are selling you a workflow change, not a vial. That is a different purchase, and it belongs in a different budget line.

For labs running environmental methods, the container requirements in EPA SW-846 set a floor on what the vial must do regardless of price. A vial that meets the method on paper but cores on the fourth puncture does not meet it in practice. The per-test calculation is how you catch that gap. The Thermo Fisher chromatography and sample handling library and the Restek technical resources both publish septum and vial compatibility data that will save you a week of troubleshooting if you read them before you switch lots. The lab chromatography reference material on injection technique is also worth a read, because half the reruns I audit turn out to be a technique problem that a better vial only masks.

One more bound worth setting: the cost of a single lot failure. If a lot fails incoming inspection and you have to return it, the cost is not the refund. It is the two to five days of waiting for a replacement lot, the rescheduled batch, and the analyst time spent documenting the return. I put a $1,500 placeholder on a lot failure and add it to the annual cost of any supplier with more than one failure per year. That single line has ended more supplier relationships than any price negotiation.

The number to put in front of procurement

When I present this to a procurement manager, I present one number: cost per reported result, calculated on the last 500 injections, with the rerun and reject rates measured from the CDS, not estimated. That number is the only one that survives contact with the lab. Unit price is a line item. Cost per reported result is a decision. The decision rule is simple: if the per-test cost of the cheaper vial is higher than the per-test cost of the certified vial, the cheaper vial is not cheaper, and no volume of negotiation changes that. Buy the grade that holds the rerun rate under 2% and the reject rate under 0.5% on your instrument, and let the unit price be whatever it needs to be to get there.

References & further reading

Related resources from our lab consumables network:

• Vial Supplier at vialsupplier.com
• Lab Chromatography at labchromatography.com
• Hplc Filter at hplcfilters.com
• Ptfe Membrane Filter at ptfemembranefilter.com

External references and standards:

• Agilent chromatography consumables and columns
• Thermo Fisher chromatography and sample handling
• U.S. EPA SW-846 test methods for waste and materials
• Restek technical resources for chromatography