Case pattern: a service business that automated intake
A composite pattern showing how a small service business automates enquiry intake, what it takes, what it returns, and where it nearly went wrong.
The short answer
This is a composite pattern rather than a named client, drawn from how this project consistently runs in small service businesses. The shape: enquiries arrive through a website form, a general inbox, a phone line and occasionally social messages. They are answered by whoever notices, often hours later and sometimes not at all, with no consistent set of questions, so the first useful conversation happens days after contact. The intervention consolidates all four channels into one flow. Every enquiry receives an immediate acknowledgement, a short set of qualifying questions appropriate to what they asked about, and either a booked call slot from a real calendar or a routed record with the owner notified. AI reads free-text enquiries and drafts the reply; rules handle the calendar and the routing; a person takes over the conversation. The gain that matters is time to first response, from hours or days to minutes, and that alone moves conversion more than any later refinement. The build is a few weeks and the owner is the office manager.
The before and after
| Element | Before | After |
|---|---|---|
| Channels | Four, handled separately | One flow, four entry points |
| Time to first response | Hours in the day, overnight for evenings and weekends | Under a minute, always |
| Qualification | Whatever the responder thought to ask | A short consistent set, adapted to the enquiry type |
| Record | Sometimes a CRM entry, sometimes an email thread | A structured record every time, with the original message attached |
| Booking | A round of emails to find a time | A slot offered from the real calendar |
| Ownership | Whoever replied | Assigned by rule, notified with one-tap claim |
| Visibility | None; nobody knew how many enquiries were missed | Weekly numbers by channel, response time and conversion |
| Out-of-hours | Nothing until morning | Acknowledged and qualified; call booked |
How it was built
- Measured the baseline for two weeks: enquiries per channel, response times, how many went unanswered.
- Consolidated the channels into one inbox and one CRM, with the phone line taking messages into the same flow.
- Wrote the qualification questions with the people who usually answer enquiries.
- Built the flow: acknowledgement, questions, calendar slots, structured record, owner notification.
- Added AI to read free-text enquiries, classify them and draft the first reply for the rules to send.
- Tested on a week of real past enquiries before going live.
- Went live on one channel first, then the rest.
- Reviewed weekly for the first month, and this is where the correction happened.
What the business noticed
Enquiries that used to arrive overnight were acknowledged and often booked before anyone opened the office. The team stopped rediscovering old emails with unanswered enquiries. Everyone was asked the same questions, so the first call started from a known position. The office manager could see how many enquiries came from each channel and what happened to them, which changed where the business spent its marketing money. And the total build was smaller than the first quotation for the ambitious version that was considered and rejected.
What this means for you
If enquiries reach your business through several channels and are answered inconsistently, the intake pattern applies: consolidate the channels, acknowledge and qualify immediately, offer real calendar slots, create a structured record and notify an owner, with AI reading free-text messages and rules handling the calendar. Measure time to first response before you start, keep the qualification short and test completion, and expect the speed of response to produce most of the gain.
Frequently asked questions
Is this a real client?
It is a composite pattern drawn from the way this kind of project consistently runs, with no client identified and no figures presented as one business's results. We publish patterns rather than case studies with numbers because the numbers depend heavily on the starting point, and a pattern is more useful for deciding whether your business has the same shape.
What made it work?
Three things. The process was already understood, so it could be described precisely. Time to first response was measured before anything was built, so the improvement was visible. And the office manager wanted it, which meant exceptions were handled and the flow was corrected quickly when the qualification questions proved too long.
What would have made it fail?
Building the clever version first: an assistant that conversed freely, answered questions about services and tried to book work, on top of four unconsolidated channels. That version is harder, riskier and would have delayed the simple gain that produced almost all of the value, which was answering everyone within a minute with a relevant next step.