Contents5 sections
Field notesARTICLE
A Price List Is Not a Price Rule
A table answers the question it was built for. Customers ask the other one. The work of turning pricing into a rule takes an afternoon and is worth doing whether or not you ever automate anything.
Costa4 min read
Short answer
An assistant can quote prices reliably only when pricing is expressed as a rule: a base, the modifiers that apply to it, the order they apply in, and the named exceptions. A price list is a set of answers to one question, and most customer questions are slightly different from that one, which is why quoting from a table produces a person in the loop every time.
Key facts
- The assistant quotes from a rule and states which parts of the request the number depends on.
- A request that falls outside the rule gets a range and a person, never an invented figure.
- Modifiers are applied in a written order, because discount-then-season and season-then-discount are different numbers.
- Every quote records the rule version it came from, so a disputed number can be traced rather than argued about.
Most businesses have a price list. Almost none have a price rule, and the difference only becomes visible when something has to quote at three in the morning.
A list is a set of answers to one question: what does this standard thing cost? Customers ask a slightly different question nearly every time. Six people instead of four. Ten days instead of a week. Two of the days in high season. Collection on a Sunday.
A table answers one question
Somebody in your company can answer all four of those without looking anything up. They do it by holding a procedure in their head that was never written down, and they have been doing it so long that they experience it as knowing the price rather than as computing it.
That procedure is the asset. The table on the website is a shadow of it.
The ten-quote test
Open the last ten quotes anyone sent. Try to reproduce each number from your written pricing, without asking the person who sent it.
The result is always some of each: numbers that come out exactly, numbers that come out close, and two or three you cannot get to at all. The last group is where the work is. Each one is either an exception worth naming or a habit worth ending, and a business only finds out which by looking at them side by side.
That is the entire exercise. It takes an afternoon, it needs the person who actually quotes in the room, and it produces something that outlives whatever software prompted it.
Three schemes, written out
A charter fleet. Base is the day rate by boat. Modifiers: season band, duration bracket, skipper included or not, fuel. Order matters — the duration discount applies to the seasonal rate, not the other way round, and writing that sentence down settles an argument that has been running for years. The deposit is not a modifier at all, it is a separate line and a separate conversation.
A letting agency. Commission as a percentage with a floor, because the percentage stops making sense below a certain rent. The floor is the part that lives in people's heads and never in the table. Written out it is one clause, and it is the clause every junior gets wrong.
A service with a setup fee. The fee is waived above a contract length. The rule needs the threshold, and it needs the answer to the question that always follows: what happens if they sign for the shorter term and extend later. That answer exists — somebody has given it before — it has just never been in a document.
Where a rule is allowed to end
Every rule has a boundary, and naming it is part of writing it.
Outside the boundary the assistant does not guess. It gives the range the rule supports, names the two facts that would settle it, and hands over. A quote that is confidently wrong is the one failure here that costs actual money, and it is worth being deliberately boring to avoid.
Some businesses discover during the exercise that almost everything is outside the boundary, because every deal is negotiated. That is a real answer, and the useful response is not to automate the negotiation. It is to let Omni AI collect the inputs, produce the anchor, and put a complete picture in front of the person who negotiates, which removes the waiting without touching the judgement.
The part that has nothing to do with software
Two people in the same company quoting differently is the first thing that surfaces after launch, and it was true before the assistant arrived.
Write the rule and the quoting gets consistent whether or not anything automated ever reads it. New staff stop needing six months to be trusted with a number. The discount somebody has been giving out of habit becomes visible, and then it becomes a decision.
The assistant is what makes the absence expensive enough to fix. The rule is what you actually keep.
Questions this raises
- We negotiate everything. Does that mean we cannot automate quotes?
- It means the assistant should not close them, which is not the same thing. It can still produce the anchor number, state what it depends on, and hand the negotiation to a person with all the inputs collected. That alone removes most of the delay.
- What happens when a request lands outside the rule?
- It says the range, says what it needs to be sure, and passes it on. A confident number outside the rule is the one failure mode that costs real money, so it is worth being boring about.
- Do we have to publish the rule?
- No. The rule is internal; what the customer sees is a number and the two or three facts it depends on. Publishing the mechanism is a separate decision, and plenty of businesses correctly decide against it.