Keep the enquiry
Save the original message and contact details with a stable reference.
Your business is closed. Your enquiry form is still working.
An evening enquiry does not always need an evening sales conversation. It needs a clear acknowledgement, a believable next step and someone responsible for following through. AI can prepare the morning’s work; a simple approved message may handle the first response.
A receipt now. A useful answer next.
Separate a fixed acknowledgement from a reply that makes commitments.
Send the receipt
Approved text can send automatically after capture succeeds.
Help the owner reply
AI can summarise and draft. The owner checks prices, dates and promises.
Keep one conversation
Record the answer and next action. Update reminders when things change.
Separate the receipt from the answer
An acknowledgement says the message arrived. A substantive reply answers the request, gives advice or makes a commitment. Treating these as separate jobs lets you be responsive without pretending somebody is working overnight.
A fixed acknowledgement, approved in advance, can be sent automatically. An AI-written proposal or a promise about availability needs a different level of checking. Decide that boundary when designing the workflow.
| Message | Suggested handling | Important limit |
|---|---|---|
| We received your enquiry | Approved automatic receipt | Send it only after the enquiry is successfully recorded. |
| Here are our published opening hours | Approved text from a maintained source | Keep holiday exceptions current. |
| Here is a draft answer to your question | AI prepares; the assigned person checks | Do not treat guessed details as customer facts. |
| We can start on Tuesday for this price | Confirmed business decision | Use actual scheduling and pricing information. |
An acknowledgement worth sending
Illustrative template for a business whose team next responds on the following working day. Replace the response window with one your business can consistently meet:
“Thanks for your enquiry — it has reached our team. We’ll review it on our next working day and reply by 12 noon. If you would like to add anything, reply to this email and we’ll keep it with your request.”
Use that wording only if replies really join the same conversation. If your form is not monitored for emergencies or urgent support, put that information and the appropriate contact route near the form as well as in the receipt.
The morning queue is the part that makes the promise real
Illustrative example: a garden maintenance business receives a request for a one-off tidy-up. The prospect gives an area and preferred week but no property size. By morning, the owner needs the original message, a short summary and the missing question — not a second inbox to patrol.
A useful enquiry record contains the source, received time, contact details, requested work, assigned owner and next action. If AI extracts a detail, let the reviewer trace it back to the original message.
- Record the enquiry and issue the approved receipt after capture succeeds.
- Assign an owner using an agreed rule. Unmatched enquiries go to a monitored fallback queue.
- Prepare a short summary and, if useful, a draft reply asking for missing details.
- Let the owner confirm the substantive reply and update the next action.
- Keep replies, changes and closure attached to the same enquiry so old reminders do not keep firing.
Use AI to reduce preparation, not manufacture certainty
The valuable AI step may be as small as extracting the requested service and drafting one relevant follow-up question. Give it the current service information and tell it which details remain unknown.
For example, “Could you share the approximate garden size?” follows from the missing information. “Our team can come next Thursday” requires verified availability. Polished wording does not make the second statement true.
n8n supports human review of selected AI tool calls. That makes selective approval possible in a supported implementation; it does not require a person to approve every fixed receipt. NIST’s generative AI guidance also highlights the risk of plausible but incorrect output.
n8n: human review for selected toolsNIST: generated-content risks
Test the cases that usually fall between the cracks
Use sample enquiries before connecting a real customer channel. These are suggested acceptance tests, not reported results.
| Test input or event | Expected behaviour |
|---|---|
| The same form arrives twice | One coherent enquiry, with a deliberate duplicate policy. |
| A customer replies before the morning review | The reviewer sees the latest message; an outdated draft is not sent. |
| No service or location matches a rule | The enquiry remains visible and gets a fallback owner. |
| The CRM connection fails | The original message is retained and someone is notified to recover it. |
| Tomorrow is a holiday | The receipt does not promise an unstaffed response window. |
| The AI cannot understand the request | The original message reaches a person without an invented interpretation. |
Start with one channel and an honest response window
Connect the enquiry source your team already uses most. A working form-to-owner handoff is a better first release than five channels with different rules and nobody responsible for exceptions.
Review what happened: were messages captured, did owners respond within the promised window, and did the draft save useful preparation? Those observations tell you what to improve next. They do not need a speculative revenue forecast.
Sources
Make the first response easier to follow through on
Show us where your enquiries arrive and who answers them. We can help connect the receipt, the owner and the next action.
Discuss your enquiry workflowUpdated