AI Assistant Pricing Replies in Cold Email: What It Should Do
An AI assistant handling cold email replies should not generate a price on its own, even if it can guess one from context. The safe pattern is to acknowledge the question, ask one qualifying detail if needed, and either point to a booking link or flag the lead for a human to quote directly. A pricing question is a strong buying signal, and the goal of the reply is to keep the conversation moving toward a call, not to close the deal in text.
When a prospect asks for pricing, an AI assistant should never improvise a number. It should acknowledge, qualify, and route to a human or booking link.
A reply that asks "what does this cost" is one of the highest-value signals a cold outreach campaign produces. The prospect has moved from curious to evaluating. That is exactly why it is the wrong moment for an assistant to improvise an answer — a made-up or inconsistent number in writing is hard to walk back, and it can commit you to a deal you did not intend to offer.
Why improvising pricing is risky
Most small businesses do not have a single list price. Scope varies by location, volume, urgency, or what the prospect already has in place. An assistant reading a reply thread does not know your current capacity, your margin on a rushed job, or whether this particular prospect is a good fit at your standard rate. If it fills that gap with a plausible-sounding number pulled from your website or a past email, that number becomes a quote whether you meant it to be one or not.
The fix is not to ban the assistant from discussing price at all — that makes replies feel evasive. The fix is to give it a narrow, explicit set of moves it is allowed to make when pricing comes up, and nothing outside that set.
A safe reply pattern
- Acknowledge the question directly. Don't dodge it with a generic "thanks for reaching out" — that reads as ignoring what they asked.
- Ask the one detail that changes the number, if you have one. For example, square footage, number of seats, or current provider. This is only appropriate if that detail genuinely changes your pricing and you've told the assistant so.
- Route, don't quote. Point to a booking link, offer a short call, or say a specific person will follow up with exact numbers. This keeps the human in the loop for the actual commitment.
- Flag the lead as high-intent. A pricing question should move the lead out of the general sequence and into a watched state, regardless of what happens next.
What the assistant needs from you first
An assistant can only follow this pattern reliably if you've given it boundaries, not just a general instruction to "be helpful." At minimum it needs to know:
| It needs to know | Because |
|---|---|
| Whether you have published, fixed pricing | Determines if quoting a number is ever appropriate |
| What detail(s) actually change the price | Avoids asking irrelevant qualifying questions |
| Where to send the prospect instead of quoting | A booking link or named contact, not a guess |
| How to flag the lead afterward | So a pricing reply doesn't just sit in the general sequence |
Without this, two things tend to happen. Either the assistant gets overly cautious and sends a reply that dodges the question, which frustrates a prospect who asked a direct one — or it gets overly confident and states a number as if it were authorized to, which it was not.
Treat it as a branch, not a special case
It helps to think of a pricing question as one of a small number of reply types your sequence needs to handle explicitly — alongside "not interested," "wrong person, here's who to contact," and "send me more info." Each of these should have its own defined handling, not a single catch-all "respond politely" instruction. Pricing deserves its own branch because getting it wrong has real cost: either a quote you didn't intend, or a stalled conversation with someone who was ready to talk.
Keep a human in the loop for the actual number
Even with a good reply pattern, the person who sets final pricing should stay the one who says the actual figure, at least until you've seen enough pricing replies to trust a narrower, pre-approved range. That's a judgment call based on how standardized your offer is, not something an assistant should decide on its own the first time it sees the question.
What to take away
- Never let an assistant generate a price from inference — give it a fixed, narrow set of allowed responses instead.
- Acknowledge the question, ask only the qualifying detail that changes price, then route to a human or booking link.
- Treat a pricing question as a high-intent signal that should pull the lead out of the general sequence.
- Define pricing replies as their own explicit branch in your playbook, not part of a general "be polite" instruction.
- Keep final number-setting with a person until you've seen enough real replies to trust a pre-approved range.