Coffee customer service - allergen and dietary-restriction response scripts - read the ingredient facts, hand health judgments back, with separate lines for pregnancy and palpitations

You are a customer-service trainer for a coffee shop.
Turn the three question types below into counter-ready
scripts: for each, give the verbatim line (spoken, short
sentences), a "banned phrases" block right under it (with
why each phrase fails), and a second-round answer for when
the customer pushes back.

Master rule (print at the top of the script sheet; it
governs all scenarios):
Read facts only, make no health judgments. Facts come
from exactly three places: the shop's ingredient card,
the supplier's allergen information, and how the drink
is built (which drinks contain milk, which use oat milk,
whether a syrup carries a nut flavor note). If none of
the three sources answers it, say "let me check the
actual packaging and get back to you" - and then
actually go check.

[Type 1: Allergen questions - "does this have
lactose / gluten / nuts?"]
Say this:
"Let me read you the ingredient list: this one's milk is
[ ], the syrup is [ ], and per the supplier's allergen
labeling it contains [ ]. The thing you asked about -
[lactose/gluten/nuts] - is / is not on that list. Want
me to photograph the ingredient list for you?"
Banned:
- "There shouldn't be any" - "shouldn't" means nobody
  checked, and in an allergen context that's the same
  as no
- "It's just a trace, you'll be fine" - the store does
  not decide what dose triggers a reaction
Pushed with "so does it or doesn't it":
"The label says [ ]; what I can promise is that I read
you what's labeled. If you're sensitive even to trace
cross-contact, I'd point you to [plain black coffee /
no-syrup drinks] as the lower-risk option - the call is
yours."
[Fill in the shop's actual cross-contact practice:
shared grinder, shared milk pitcher, rinse-and-swap
policy - write what you really do, not the ideal
version]

[Type 2: Caffeine - "can I drink this pregnant / can
my kid / I can't sleep at night"]
Say this:
"That's not something I can decide for you - for health
questions, please go by your doctor's advice. The facts
I can give you: this drink is [single/double] shot, and
we can make it with our decaf beans, which tastes
slightly different. Want me to switch it to decaf?"
Banned:
- "A little is fine when pregnant" - that's medical
  advice
- "Ours isn't that strong" - a comparison with no
  basis behind it
Pushed with "so decaf has zero caffeine?":
"Decaf is a big reduction, not zero - I don't have a
number I can read out to you with any authority behind
it. If that amount matters to you, I'd check with your
doctor before deciding."

[Type 3: Feeling unwell - "my heart is racing / I feel
bad after drinking it"]
Say this (intake, no verdict on cause):
"Understood - let me note this down: you had [drink/cup
size] at [time], and what you're feeling now is [ ].
Can I take a contact? We'll get back to you after we've
checked. If you're feeling really unwell right now,
please see a doctor first - don't wait on us."
Banned:
- "That can't be it - our coffee is fine" - denial
  before any checking
- "It must be the coffee, we'll refund you" - full
  admission before any checking
- "Might just be your constitution" - throwing the
  attribution back at the customer
After logging:
"I've logged your feedback today; the manager will
reach you within [24] hours. On refund or compensation -
that answer comes after we've checked, so I won't
promise something off the top of my head."

Formatting rules:
- Keep the scripts spoken and short, at the density of
  reading them mid-shift at the counter
- Every [ ] is a fact slot filled by the shop after
  checking - annotate each placeholder "fill after
  checking supplier ingredient sheets"
- Keep the banned-phrases block, each entry with half a
  line of why
- No fourth scenario, no "friendly reminder" tail

Why it is written this way

Every script here stands on one master rule: the store reads facts, health judgments belong to the customer and the doctor. Allergen and caffeine questions go wrong once and it's an incident, and the root of nearly every incident is a well-meaning judgment call by the counter - "there shouldn't be any" sounds like help to the person saying it and sounds like confirmation to the person with the allergy. So each verbatim line contains only checkable sources (the ingredient card, the supplier's labeling, the build of the drink); when there's nothing to read, go check the package. A slower sentence beats an unsupported certainty.

The caffeine type is structured as "refuse the judgment + offer what we can do": a bare "I can't say" reads as being brushed off, so it's immediately followed by the one thing the store genuinely controls - the decaf swap. The follow-up "so decaf is zero caffeine, right?" gets its own line because it's the likeliest trap: "mostly removed, not zero" is common knowledge, but a specific number is usually something the store has no citable basis for - and that's exactly where a percentage gets invented on the spot, hence "if I have no basis to read out, I say so."

The three banned phrases in type 3 come from real failure patterns: denial ("can't be us"), total admission ("must be the coffee"), and deflection ("your constitution") are all the same error - locking in causation before checking anything. The intake script does three things only: log the facts, take a contact, get medical help moving if needed. The compensation line is deliberately split into a second stage that waits for the investigation, because refund-first trains a stream of mercenary complaints that drown the real feedback.

Variables