Water Damage Restoration Leads After a Flood: Answer and Book With AI

Answer every flood and burst-pipe call within seconds, ask six fixed intake questions, and book a technician arrival window before the caller hangs up. An AI receptionist can take the calls your team cannot reach, but a person owns dispatch. Speed matters: public-health guidance says to dry a wet home within 24 to 48 hours.

Treat the AI as the person who picks up when your crew cannot. It greets the caller, asks the intake questions, offers a booking window your dispatcher has already approved and sends a confirmation text. It does not diagnose, quote a final price or promise a truck by a time nobody has checked. Where that line sits, and whether the first four minutes of every call are identical, decide if callers feel looked after or fobbed off.

Follow me on Instagram@sagnikteaches

Why the first call decides the job, and how fast is fast enough

Water damage is one of the few home emergencies with a clock built into the job. Public-health guidance on flood cleanup tells households to clean and dry their home and everything in it within 24 to 48 hours if they can, and warns that things left wet for more than two days usually go mouldy. The same guidance warns that flood water can carry sewage, chemicals and germs, and that the safest course is to stay out of it. A caller with water across the kitchen floor will not know the detail, but they know it is urgent and they will keep dialling until somebody answers.

Connect on LinkedInSagnik Bhattacharya

So plan on the assumption that the caller is ringing more than one firm. The firm that picks up, asks sensible questions and names an arrival window usually earns the first visit, and the first visit is where most jobs are won. I cannot give you a statistic for how much faster answering lifts your booking rate, and I would be wary of any vendor who quotes one without showing its source, because it depends on your market and the weather. You can measure it on your own phone log, which the next section shows. Until then, these are the targets I would set for a small crew. They are my suggestions, not industry benchmarks.

Subscribe on YouTube@sagnikteaches

Not every lead phones. Web forms and text messages deserve the same discipline, and the tutorial on replying to trade enquiries in 60 seconds covers that side. On a flood night, though, the phone is where the money rings.

Audit last month's missed calls before buying anything

Export your phone log for the last busy week or the last storm. Most business phone systems can download call history as a spreadsheet, and a carrier's online account usually can too. Add four columns: answered live, voicemail, rang out, and outcome (booked, quoted but not booked, existing customer, wrong number or spam). Then pull the average value of your last 20 completed jobs from your job software. That is all the data you need.

Here is how the log looks when it is filled in. The rows are illustrative, but the patterns are the ones to hunt for: a late-night call that rang out, a voicemail returned too slowly, and a daytime caller who was plainly ringing round.

Now the main worked example, again with illustrative figures. Take a four-van restoration firm that logs the 72 hours after a night of heavy rain. Swap in your own numbers; the sums are what matter.

Set that against what cover costs. On the AI plans priced below, 19 extra calls cost between $0 and $150 for the month, and a human answering service starts at $300. A single $4,200 job pays for 28 months of a $150 plan. If you want the same sum with your own call volumes and booking rate, the AI receptionist ROI calculator tutorial sets it out as a worksheet.

Six intake questions, asked in this order

The order matters because each answer decides what you do next, and the first one can end the intake entirely. Whether a person or an AI asks them, the wording should stay the same on every call, so that your dispatcher reads identical fields every time.

1. Is anyone in danger right now? If the caller mentions sparks, a gas smell, a sagging ceiling or sewage rising through drains, the script stops, tells them to leave the area and call the emergency services, and transfers to your on-call manager. Nothing else is worth asking first.
2. Where is the water coming from, and has it been stopped? The source decides the crew and the kit: a clean supply line, an appliance, sewage, or rain and flood water from outside. Water that is still flowing and cannot be shut off is tier 1.
3. When did it start, and which rooms or floors are wet? Time since the start and the extent of spread set the urgency. An upstairs leak usually means wet ceilings below, which callers forget to mention unless asked.
4. What is the property, and who is in it? House, flat, shop or office; which materials are wet (carpet, hardwood, plasterboard ceilings); and whether anyone relies on powered medical equipment, or is a baby or an elderly relative. That last part can lift a call a tier.
5. Has the insurer been told, and who has authority to approve work? Owner, tenant or manager for the owner. A tenant can book an assessment, but a landlord or managing agent may need to approve the work.
6. What is the best number to text, how do we get in, and can you send six photos? This closes the call with the details the crew needs and opens the photo list covered later.

Here is the instruction block I would paste into an AI receptionist's script or prompt field. Services word the field differently, so adapt the shape rather than copying it literally.

You are the after-hours phone assistant for [company], a water damage restoration firm.
Open every call with: "Thanks for calling [company]. I'm the automated assistant. Is anyone in danger right now?"
If the caller mentions sparks, a gas smell, a sagging ceiling or sewage rising indoors: tell them to leave the area and call the emergency services, then transfer to [on-call manager] and text the summary.
Otherwise ask, one question at a time and in this order:
1. Where is the water coming from, and has it been stopped?
2. When did it start, and which rooms or floors are wet?
3. What kind of property is it, and does anyone rely on powered medical equipment?
4. Have you told your insurer, and do you own, rent or manage the property?
5. What is the best number to text, how do we get in, and can you send six photos?
Rules: never diagnose, never quote a price, never promise an arrival time. Offer only the windows returned by the booking tool.
If the caller asks for a person, or the call is not going well after two attempts, transfer to [on-call manager].
End by repeating the address and the window, and say a text confirmation is on its way.

And here is the summary that should reach the on-call manager within a minute of the call ending. Every line is a field you can ask your receptionist service to fill.

NEW JOB REQUEST - TIER 2
Caller: [first name], mobile [number], texting OK
Address: [street], side gate, code in confirmation text
Source: upstairs bathroom supply line, water stopped at 22:40
Extent: bathroom floor, hallway carpet, kitchen ceiling stain below
Occupants: two adults, one child, nobody in danger, no powered medical equipment
Insurance: not yet notified; owner-occupier
Photos: caller will send six after the call
Window offered: 07:00 to 09:00 (approved slot)
Flags: none

Priority tiers for the night the phones do not stop

A storm can put 25 calls into two hours, and the worst thing a crew can do is take them in the order they arrived. Tiers let the receptionist and the on-call manager sort the queue the same way. The AI proposes a tier from the answers above; a person confirms every tier 1.

When in doubt, the rule is to move a call up a tier, never down. For a closer look at how an AI sorts urgent plumbing calls, which share a lot of this logic, read how AI can triage emergency plumbing calls out of hours.

During a genuine surge, honesty about the queue keeps callers on the line. Have the AI say, for tier 2 and 3 callers, "Our crews are fully committed tonight; the earliest assessment I can book is [window]. Would you like that, or a call when a crew frees up?" I would rather a caller hear a long wait stated plainly than discover one for themselves an hour later.

Overflow answering: three ways to cover calls you cannot take

Overflow means only the calls nobody picks up in time. That is a narrower job than answering everything or covering nights, and it is usually the cheapest place to start. The comparison of all calls, after hours or overflow for a plumbing firm explains the three modes; the table here prices the cover itself for a restoration firm. Figures are list prices in USD from the vendors' own pages as of late October 2026, and the calls are assumed to be real, answered ones.

Read the small print before you pick. Smith.ai counts every call its AI answers against your quota by default, filters known spam before it answers, and lets you remove up to 10% of calls from a bill for selected spam. Its plans also list monthly allowances for test calls (25 on Free) and simulated calls (50 on Free), though the page does not say how they are billed. Its own AI receptionist pricing page carries the current tiers. Quo is a phone system first: a Quo seat costs $15 a month on annual billing for Starter ($19 monthly) or $23 for Business ($33 monthly), and call transfers to a human, which a restoration line needs, sit on Business and above. Quo's pricing page shows the Sona tiers. The human service is far dearer per call, but a trained person is likely to handle a frightened caller better than a script can, so some firms keep it only for tier 1 overflow.

Setting up overflow takes an afternoon:

1. Choose how long a call rings before it moves on, for example four rings or 20 seconds.
2. Ask your carrier or phone provider for forwarding when busy or unanswered, sometimes called conditional forwarding, and point it at the service's number.
3. Paste the script, your approved arrival windows and your hours.
4. Name the transfer contacts: the on-call manager first, a second person if the first does not answer.
5. Connect the calendar the service books into, or the form it writes to. Smith.ai's page lists scheduling through Calendly and native integrations including Zapier.
6. Run the rehearsal calls in the last section, then read every transcript daily for the first two weeks.

One caution on recording: AI receptionists typically record or transcribe calls. Tell callers, and ask a qualified adviser what notices your own market requires.

What the AI may promise, and what only a person can

This is where most set-ups go wrong, and the fix is a short list. Put it at the top of the script and test it hard.

• The AI may say: that it is automated; that a technician will assess scope on site; the arrival windows returned by your booking tool; that a manager will call within the stated time for tier 1; your assessment fee, if you charge one; what to do while waiting, using wording your safety lead has approved.
• The AI must never say: that a truck will arrive at a specific time unless the slot came from the board; a price for the job; what the insurer will cover; whether a material can be saved; anything about the cause of the leak or who is responsible.

Picture a restoration firm that told its receptionist to reassure callers with "a technician will be with you within the hour". On a storm night the nearest crew was two hours out. The caller, reassured, stopped ringing round. After 70 minutes with no van they rang a second firm, which offered a 45-minute window and arrived first. The first crew drove to a house that no longer needed them, and the owner posted a one-star review about a promise nobody had checked. The fix was a rule: the AI offers only windows from the dispatcher's board, and if the board is empty it says "a manager will call you within 10 minutes with a time" and creates a tier 1 task.

A second guard is a hard stop on anything resembling advice about cover or liability. Callers ask "will my insurance pay for this?" because they are worried, not because they want an answer from a phone assistant. The AI should reply that the insurer decides cover, offer to note the claim number, and carry on with the intake.

The photo list an adjuster can use, sent while the crew drives

Photos serve two readers: the insurer's adjuster and your own scope. Restoration firms commonly photograph in a fixed order so a stranger can follow the story from cause to condition. That order, as published by restoration companies, runs from the source of the water to wide shots of each room, then close-ups, contents and moisture readings. Treat the list below as common practice, and let the insurer's own instructions override it: ask for the claim number on the call, and ask the claims line what format it wants.

The AI's job is the first and cheapest part: while the crew drives, it texts the customer a short request for the first three or four photos, so that scope is half-formed before the van arrives. Never ask a customer to take photos from anywhere unsafe, such as standing in water or near wet electrics.

After the visit, the technician's raw material is a folder of images and a few spoken or typed notes. A chat assistant can turn that into a tidy log. Here is a filled-in example for an upstairs bathroom supply-line burst, followed by the prompt that produced it.

Job WD-1041 | Loss: bathroom supply line burst | Date of loss: [date]
Room: Bathroom (upstairs)
  IMG_0408 - source: failed compression fitting under basin, wide
  IMG_0409 - source: fitting close-up, split visible
Room: Hallway
  IMG_0411 - wide from landing, carpet saturated to skirting
Room: Kitchen (below bathroom)
  IMG_0414 - ceiling stain approx 1.2 m x 0.8 m, tape in frame
  IMG_0415 - moisture meter, ceiling 38%, tag KITCHEN-C1

Here are my photo filenames and the technician's notes for one water damage job. Write a room-by-room photo log: filename, one plain caption, no measurements unless they appear in the notes. If a note is ambiguous or a room is missing a photo, list it as a question at the end instead of guessing.

Check three things in the output before it goes anywhere: that each image sits under the correct room, that no measurement appears which the notes did not contain, and that the questions list picks up any gaps. Use the assistant for captions only. Never let an AI app enhance, retouch or filter a claim photograph, and keep the originals untouched. Photos taken inside homes also show families and belongings, so upload them only to a business-grade account where the provider says it does not train on your content by default, and check the privacy setting on whatever you use. For the moisture-log side, the moisture log and scope notes checklist goes into the records a drying job needs.

Texts that keep the caller from ringing the next firm

A voicemail greeting and an AI answer set very different expectations in the first ten seconds. Compare them.

Before: "You've reached [company]. We can't take your call right now. Please leave a message and we'll get back to you."
After: "Thanks for calling [company]. I'm the automated assistant. Is anyone in danger right now? ... Thank you. I have your address and the window of 07:00 to 09:00 booked. A text with what to do while you wait is on its way, and a manager will call if anything changes."

The text itself does the rest of the work. Write three and keep each short: the confirmation, the while-you-wait message, and the photo request. Use straight apostrophes and no emoji, because a single curly apostrophe or emoji can cut an SMS segment from 160 characters to 70 and split a short message in two.

Confirmation: [Company]: booked for 07:00-09:00 at [address]. A technician will call 20 min before. Reply HELP to change this.
While you wait: If you can reach the main water shut-off without standing in water, turn it off. Keep people and pets out of wet rooms. Do not use electrical items in wet areas. If you see sparks or smell gas, leave and call the emergency services.
Photos: Please text 3 photos if it is safe: the source of the water, each wet room from the doorway, and any damaged furniture. Do not throw anything away yet.

Have your safety lead, and if possible your insurer or broker, sign off the while-you-wait wording before it goes live. The advice must be conservative: switch off the water if you can do so without standing in it, stay away from wet electrics, avoid contact with floodwater or sewage, and leave if in doubt. Keep it one message so people read it.

Test it with ten rehearsal calls before the storm

Do not wait for a real flood to find out what the script does with an angry caller. Ring your own service ten times from a mobile, using different voices, and tick each result. Smith.ai's pricing page lists test and simulated calls as their own allowances, but it does not say whether they count against the bill, so ask your provider before you start rehearsing.

All ten must pass before you forward a single live call. After go-live, review each week on five measures: share of calls answered, minutes from call to booked window, tier 1 transfers that a person answered within 10 minutes, how often callers asked for a human, and complaints. Compare answered share and bookings with the baseline from your audit. If the AI-answered calls book at a much lower rate than your live-answered ones, read the transcripts of the lost calls first; the cause is often a clumsy question or a window that was never offered, rather than the technology itself.

Questions restoration owners ask before switching on overflow answering

Should the AI receptionist tell callers it is an AI?

Yes, in the first sentence and in plain words: 'I'm the automated assistant for [company].' Callers who feel tricked ring the next firm, and transparency rules for AI that talks to people have applied in the EU since 2 August 2026, which matters if you have customers there. Ask your adviser what your own setup needs, and add a recording notice if the service records or transcribes calls.

Can the AI quote a price for water extraction?

Not a fixed one. Scope depends on the type of water, the materials and the moisture readings, which only a technician sees on site. Let the AI state your assessment or call-out fee if you charge one, say that a technician confirms the scope on arrival, and book the visit. A figure given from a phone description is one you may be held to later.

What if the caller has no insurance or has not called their insurer?

Keep going. Record it as 'not yet notified' or 'no cover', still offer the earliest window, and let the technician explain the work and payment terms on site. Ask neutrally ('Have you spoken to your insurer yet?') so callers do not feel screened, and keep the AI away from any comment on whether a claim will be paid. Cover decisions belong to the insurer.

Will overflow forwarding work with my current phone number?

Usually. Ask your carrier or phone provider for forwarding when busy or unanswered, then point it at the number the answering service gives you. Quo works differently because it is the phone system, so you move or forward your number to it. Whichever you choose, call from a mobile and check that the caller's own number reaches the service, otherwise the confirmation text goes nowhere.

Further reads

• AI Receptionist for Plumbers: Cost, Burst-Pipe Rules and Booking Limits — The burst-pipe call from the plumber's side: costs and booking limits.
• AI Receptionist for Roofing Companies: Storm-Day Calls and Claim Questions — Another storm-day surge, with claim questions to script.
• Setting Up an AI Answering Service That Never Misses an Emergency — Escalation rules for any emergency line, not only water.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Smith.ai AI receptionist and live-receptionist pricing pages and Quo pricing page (read 29 October 2026); public-health flood-cleanup guidance on drying times, mould and contaminated flood water; restoration-company guidance on documenting water damage for insurance claims; series fact sheet for the EU transparency date.