Multi-location home services · the notes field nobody reads

A customer types “water everywhere, urgent” into your booking form. It books a slot three days out anyway.

A booking widget’s notes field is free text, the same raw material a call or chat already gets read for urgency. Nothing parses the notes field the same way — it files as an attachment on whatever slot the customer happened to click.

The desire, and the field that gets stored but never read

Nobody searching “online booking software” wants a calendar widget. They want a customer who types “emergency” into a form to be treated like it is one, not scheduled into the same queue as a routine maintenance request.

1. 1. The capability — language-based urgency parsing. Already built and running on this site’s call and chat pages: read what a customer says in their own words and classify how urgent it is. This is the well-known half.
2. 2. The unpointed field — the booking widget’s notes box. Free text, filled in by the customer at the exact moment they are booking, and read by nobody until a person opens the ticket — sometimes after the slot has already confirmed for three days out.
3. 3. The close — point the same capability at it. A note containing flooding, no-heat or safety language flags the booking for review before it confirms, the same way an urgent call already gets escalated instead of following the default flow.

The gap is not a missing capability. It is a capability aimed at two inputs and never pointed at the third one sitting in the same stack.

Which platform this runs on

• Field-service software, by scale — the platform that owns the booking widget and its notes field.
• AI phone answering: what it replaces, what it operates — where this same urgency-parsing capability was first built for this site’s tracks.
• AI chatbot for ecommerce: the signal Klaviyo never hears — the same capability, a different unread input.

Some links on the recommendation page are partner links: the vendor pays us if you sign up, your price does not change, and each button says which.

How to build it yourself, end to end

Five steps. Each names the vendor touchpoint from the picks above.

1. 1. Read the notes field before the booking confirms, not after. Parsing it once the appointment is already set defeats the purpose — the check has to run in the booking flow itself.
2. 2. Build a short list of real urgency signals, not a keyword blocklist. Flooding, no heat in freezing weather, gas smell, no power — a handful of genuinely urgent categories, not a broad net that flags half the notes field as an emergency.
3. 3. Hold the booking for review, do not auto-reschedule it. Flag it for a person to confirm the real slot; an AI automatically moving a confirmed appointment without human sign-off is a different, riskier failure mode.
4. 4. Show the customer their note was seen. A confirmation that acknowledges the specific thing they typed reads as attentive; a generic confirmation right after they wrote “urgent” reads as ignored.
5. 5. Track how often a flagged note actually needed the earlier slot. If flagged bookings routinely did not need to move, the urgency list in step 2 is too broad; tighten it against what the data shows.

Step 3 is the one to hold firm on — the whole point of this system is catching what a widget missed, not replacing the judgment of the person who confirms the slot.

Or have the missed-call half built and run

The same urgency-aware thinking, applied to the calls that never reached a booking form at all — built and operated, paid only as a share of the revenue it recovers.