Compare two lab deals without getting lost in the rebate

Reading time: about 10 minutes

The situation. One supplier rep leaves you a proposal and another leaves a different one. One is "12+2 plus 8% off"; the other "17% flat with a 3% rebate above €6,000". They are not comparable by eye, and both look good.

This case is pure arithmetic, which is what makes it perfect for AI: no clinical judgement, no personal data, and a verifiable answer. The only thing you have to get right is describing both offers without leaving anything out, because what you do not mention does not make it into the maths.

What you need to hand: Both proposals in front of you, with their full terms, and an honest estimate of how much you will genuinely buy in a year. That last number is the one that decides, and it is usually the inflated one.

Step by step

Write out both offers in full, including the awkward bits

What almost everybody leaves out: the payment terms, the minimum order, whether the rebate is tiered or from the first euro, whether free goods count towards the rebate, and whether you must buy the full range.

Those five details move the result more than the big percentage on the proposal’s first line. 17% at 30 days with a €2,000 minimum can end up worse than 12% at 90 days with no minimum.

Ask for the real unit-cost table

The figure that makes two offers comparable is one: what each unit costs you on the shelf. Everything else — percentages, free goods, rebates — are different routes to that number.

And ask for it at three volumes, not one: what you buy today, what the rep says you will buy, and one below. That is where the comparison usually flips.

Compare two supplier purchase terms for a pharmacy. I want the REAL
COST PER UNIT in each case.

OFFER A:
- List price €8.40 per unit
- Discount 8%
- Free goods 12+2 (for every 12 bought, 2 free)
- Rebate: none
- Payment at 60 days. Minimum order none.

OFFER B:
- List price €8.40 per unit
- Discount 17% flat
- Free goods none
- Rebate <<3% on the total, only above €6,000 a year, credited the
  following year>>
- Payment at 30 days. Minimum order €400.

CALCULATE IT for THREE annual volumes: €4,000, €6,500 and €9,000 of
gross purchases.

Give me a table with: volume, units I actually receive (free goods
included), real cash outlay, COST PER UNIT, and which one wins.

At the end, tell me in one sentence at what volume the winner changes.

Show your working. I do not want just the result.

Check one whole row by hand

Take the middle row and redo the maths on your phone calculator. It is four operations and it is what separates an informed decision from a decision with extra steps.

The model’s typical error here is in the free goods: sometimes it spreads them over the total and sometimes it does not, and that moves unit cost by well over 10%. If your check does not match, that is almost certainly where.

And if it does not match, do not argue: say "the unit cost of offer A at €6,500 comes out at X for me, check it again". It fixes it without fuss and tells you where it went wrong.
Add the figure the rep did not: what you actually sell

Every proposal is built on the volume that suits whoever brought it. The question that flips the comparison is yours: how much of this did I actually sell in the last twelve months?

If you did the dead-stock case you already have that figure. If not, get it now: last year’s sales ranking for that supplier.
```
Here is a figure I had not given you: last year I bought €4,100
from this supplier and sold 85% of what I bought.

With that:
1. Redo the comparison at MY real volume, not the volume needed to hit
   the rebate.
2. Tell me how much MORE I would have to buy to reach offer B’s rebate,
   and how much cash that means up front.
3. If reaching the rebate means buying more than I sell, say so plainly
   and work out what share of that discount I am paying for myself in
   parked stock.
```
Question 3 is what this whole case is for. A rebate that forces you to buy 40% more than you sell is not a discount: it is dead stock with a percentage on top. And it is the calculation nobody does in front of the rep.
Ask for the three questions for the next meeting

Close the case with something you take with you. Not a position or a script — you are not going in to fight — but the three concrete things you need to ask for the proposal to become genuinely comparable.

Ask for it like this: "give me the 3 questions I need to put to the rep to close the gaps in this proposal, ordered by how much each changes the result". They are usually the minimum order, whether free goods count towards the rebate, and what happens if you fall short.

A full example, and what is wrong with it

The two offers from step 2 — 12+2 with 8% against 17% flat with a rebate — at the three volumes. This is the table it returned.

What it replies

At €4,000: A → 517 units, outlay €3,680, €7.12/unit · B → 476 units, outlay €3,320, €6.97/unit. B wins.

At €6,500: A → 841 units, outlay €5,980, €7.11/unit · B → 774 units, outlay €5,395 (with rebate), €6.97/unit. B wins.

At €9,000: A → 1,164 units, outlay €8,280, €7.11/unit · B → 1,071 units, outlay €7,470, €6.97/unit. B wins.

"B wins at all three volumes. The rebate does not change the winner, it only widens the gap above €6,000."

And now, what is wrong with that reply:

Redo the middle row and it does not add up. €6,500 of gross purchases at €8.40 is 774 units paid for. With 12+2 you also get 774 ÷ 12 × 2 = 129 free, so 903 units, not 841. The outlay is right (€6,500 less 8% is €5,980), but the units are wrong — and the real unit cost is €6.62, not €7.11.
With that, the winner changes. A wins, not B. €6.62 against €6.97. And the report said "B wins at all three volumes" with complete confidence. It is exactly the failure step 3 warns about — mishandled free goods — and here you see why it matters: it does not shift the result slightly, it reverses it.
A’s unit cost is identical across all three rows, and that is the tell. €7.12 / €7.11 / €7.11. Tiered free goods do NOT give a flat unit cost: every time you cross a multiple of 12 the cost steps down. A flat figure means it treated them as one more percentage discount, which is precisely what they are not. A suspiciously stable number is as telling as a strange one.
And it applied B’s rebate at €6,500, when the threshold is €6,000. That is correct. What it does not say — and it is what actually decides — is that the rebate is credited the following year: this year’s outlay is €5,395 only in the accounts, not in the bank. It was in the brief and it did not make it into the table.

One row checked by hand, four operations, and the winner was the other one. That is the whole lesson here and the reason step 3 is not optional.

And when you tell it — "A’s units at €6,500 come out at 903 for me" — it corrects instantly, without arguing, and explains where it went wrong. The model does not defend its mistakes; the problem is that it does not detect them either.

And if your pharmacy is not like that

If the offer includes display material or training. Put a value on it yourself before it enters the calculation, and be honest: a display unit you have no room for is worth zero, and training nobody will attend likewise. If you ask the model to value it, it will use a market price — which is what it is worth to the supplier, not to you.

If there are three offers rather than two. It works the same and it is worth asking for one more column: "from what volume does each one win". With three, each usually wins in a band, and knowing where the boundaries are is worth more than knowing which wins at today’s volume — because your volume changes and the boundaries do not.

If the supplier ties the offer to the full range. Then the comparison is not about price: it is about how much dead stock you buy to get the discount. Do the previous case first on that range’s items you already hold, and bring the result in here as a cost. The full range nearly always looks worse than it seemed, and with a number beside it you can say so in the meeting.

If it is a renewal rather than a new offer. Compare against what you have, not against the list price. It sounds obvious and almost nobody does it: the renewal is read alongside last year’s and what matters is what changed. Paste both and ask only for the differences — sometimes it is the payment terms and not the percentage.

When it does not work first time

It tells me the two offers are almost identical.

Then the decision is not about price, and it is worth saying so: decide on payment terms, service reliability, or who actually fixes a problem. A tie on unit cost is useful information, not a failure — it frees you from haggling over cents and lets you look at what does differ.

I do not know what volume to enter because I do not know what I will buy.

Use last year’s. It is the only figure that is not wishful thinking, and if the rep says this year you will buy more, ask why — that answer is information and the figure in their proposal is not. Step 4 exists precisely for this.

The proposal has terms I do not understand.

Paste it as it is and first ask "translate this into plain language, do not calculate anything yet". Understanding what is being offered is a separate step from comparing it, and mixing them is how you end up comparing two things that were not what they looked like.

The numbers check out but I do not trust the whole thing.

Do the extreme-case test: ask for the unit cost at an absurd volume, say €100. If €100 gives the same unit cost as €9,000, it is treating the free goods as a percentage and the whole table is wrong. It is faster than checking a row and catches the same error.

Before you use what it gave you

You redid one whole table row by hand.
Free goods are counted as units received, not as a discount.
The comparison is done at your real volume, not the rebate’s.
You know how much cash you would have to front to reach the rebate.
Payment terms appear somewhere in the reasoning.

It spreads free goods incorrectly. This is the specific error to look for. A mishandled 12+2 shifts unit cost by over 10%, and the result still looks like a tidy table. That is why you check one whole row.

It assumes you will hit the rebate. Give it the rep’s volume and it will answer about the rep’s volume. The deciding figure is your last twelve months, and it only enters the maths if you put it there.

It knows nothing about the supplier or your history. It does not know whether that rep meets delivery dates, whether there were stock-outs last year, or whether the full range includes three lines you will never sell. You put that in the prompt as conditions or it is not there.

It recommends the "best value" offer without asking anything. Left to itself it closes the report with a recommendation. And it is made only from what you gave it: it does not know whether that supplier delivers late, whether the other one sorted a recall for you last year, or whether your stockroom is full of their range. The arithmetic is its job; the decision has four more inputs and they are all yours.

Now, with your own. Take the last proposal you accepted and run it through this with your real volume from last year. It is uncomfortable and it is the most profitable exercise in this block: either you confirm you got it right, or you find out exactly what not doing the maths cost you.

← All practical cases Back to the school →

Compare two lab deals without getting lost in the rebate

Step by step

Write out both offers in full, including the awkward bits

Ask for the real unit-cost table

Check one whole row by hand

Add the figure the rep did not: what you actually sell

Ask for the three questions for the next meeting

A full example, and what is wrong with it

And if your pharmacy is not like that

When it does not work first time

Before you use what it gave you