AI Receptionist for Junk Removal: Costs, Photo Requests and Booking

An AI receptionist for a junk removal company costs about $65 to $215 a month at 100 calls on the plans vendors publish, and free tiers cover the first 10 to 25 calls. It books by asking about the load, access and timing, then texts the caller a request for photos, so a person can confirm the price.

The cheapest plan flips with call length, because vendors count different things: calls, minutes, unique customers or conversations. Junk removal calls run long, since callers describe piles, stairs and awkward items, and they repeat when photos arrive. So price the plan on your own call log, send photos by text so calls stay short, and start with the missed and after-hours calls only.

Follow me on Instagram@sagnikteaches

What 100 calls cost on six plans, and what each one counts

The table prices six receptionists from their own pages at three volumes: 30 calls a month (missed and after-hours calls only), 100 calls (every call) and 220 calls (a spring clear-out month). It assumes 3-minute calls for the minute-billed plans, 1.5 calls per unique customer for Goodcall (its own page equates 100 customers with about 150 calls), and annual seat prices for Quo. Figures are list prices in USD and cover the receptionist only.

Connect on LinkedInSagnik Bhattacharya

Three caveats on that table. Quo's seat price is $15 on annual billing and $19 monthly, and its call transfers are listed on the Business and Scale seats, not Starter. Jobber's figure excludes the Jobber plan the add-on sits on, and the Plus plan includes the Receptionist. Rosie's 250-minute Professional plan lists no calendar booking, and 100 calls of 3 minutes is 300 minutes, so the 100-call row uses Scale. For 100 calls the spread is roughly $64 to $213, and for 220 calls roughly $114 to $431, so the plan you choose matters as much as whether you buy one.

Subscribe on YouTube@sagnikteaches

The 100-call comparison for contractors prices the same idea for other trades, and the Quo Sona setup tutorial shows the cheapest route above in detail.

How each plan books a job, and where photos can go

Booking and photos are the two things junk removal callers expect, so here is what each vendor's own pages say. The pattern to notice is the last column.

None of the pages I read says the receptionist itself receives photos. So build the photo step from ordinary text messages: the AI sends the request, and the photos arrive somewhere a person watches. That is a design choice more than a limitation, and it works well when the text is clear.

The bill beyond the plan: texts, photos, calendar and your time

The plan price is one line. These are the others, with what I could verify and what I couldn't.

A two-truck firm's month, worked from calls to booked jobs

The firm here is an illustrative two-truck operation that gets 120 calls a month. Thirty arrive when both crews are loading or the office is shut. Of those 30, 17 are genuine new pickup enquiries; the rest are repeat dials, suppliers and spam. Of the 17, 11 send photos after a text request, and 7 accept the price they are sent. An average job bills $350 and leaves $190 after crew pay, fuel and disposal fees.

Seven booked jobs at $190 is $1,330 in gross profit from calls that would otherwise have rung out. Three scenarios follow.

• Cheap: the AI covers only the 30 missed and after-hours calls. Software cost runs from $15 on Smith.ai's free tier to $79 on Goodcall, so the month nets about $1,250 to $1,315. Every one of the seven jobs is new money.
• Mid: the AI answers all 120 calls. At 120 calls the plans in the table cost roughly $80 to $265. The extra 90 calls were answered anyway, so the gain is office time: 90 calls at 3 minutes is 4.5 hours, worth about $90 at $20 an hour. That covers the cheapest plans and falls well short of the dearer ones, so it pays only if the freed hours go to chasing photo replies or quoting faster.
• High: a spring month with 220 calls. Cost runs from $114 on Quo to $431 on Smith.ai, so choose a plan that bills spikes gently. In the table, Quo's per-call tiers and Goodcall's per-customer billing handle 220 calls most cheaply, while Dialzara's per-minute rate and Smith.ai's per-call rate cost two to four times as much.

The lesson is the order of operations. Begin with missed and after-hours calls, prove the seven jobs are real, and widen coverage only when the numbers justify it. The tutorial on quoting junk removal from photos covers how a person (or AI) turns the pictures into a truck-load price.

Nine intake questions, in the order a caller can answer them

Each question adds about 20 seconds on a call, so nine cost roughly three minutes. Ask these in this order, because the early ones are easy and the later ones need thought.

1. Name and best mobile number. The number is where the photo request goes, so have the AI read it back digit by digit.
2. Address for the pickup. This checks your service area before anything else is spent.
3. What is being removed. Offer categories: furniture, mattresses, appliances, boxes and bags, garden waste, building rubbish, or a whole room or garage.
4. Roughly how much. Large items and bags or boxes. Record the raw counts; only convert to a truck-load fraction if you have written the conversion down.
5. Where it is. Kerb or driveway, garage, ground floor, upstairs or basement, and how many flights.
6. Whether the truck can park near the door. Long carries, gates and steep drives change crew time.
7. Anything heavy or unusual. Appliances with a cooling system, TVs and monitors, tyres, paint or chemicals, gas bottles, soil, concrete or roofing, or anything medical. Any yes sends the job to a person with no price given.
8. When it must be gone, and whether there is a hard deadline. This feeds the same-day rules below.
9. Home, rental or business, and whether someone will be there. Property managers often want a callback, and outside loading changes who needs to be present.

Photos answer questions 4, 5 and 6 better than callers can. Drop those three from the call, keep the photo request in their place, and a 3-minute call becomes about 2 minutes 20 seconds. On a minute-billed plan at 100 calls that saves roughly 70 minutes a month, around $30 at Dialzara's $0.45 overage rate. Whether the short call is worth it depends on how many callers then send the photos, which is the next step.

The text-for-photos step: wording, routing and a test

The receptionist's script should end the intake with a sentence like this: "I'm texting you a link now. Please reply with three photos: everything in one shot, close-ups of anything big or unusual, and the way out, including stairs and where the truck would park. We'll come back with a price during opening hours." Then it waits until the text lands while the caller is still on the line, which Quo's Sona and Rosie's Scale plan both describe being able to do.

The text itself needs to stand alone, because callers read it later and out of context. A filled-in version, with a placeholder name:

Hi [first name], it's [business name] about your pickup at [address].
To price it, please reply with 3 photos:
1) everything in one shot
2) close-ups of big or unusual items
3) the route out (stairs, doors, where we'd park)
We'll reply with a price by [time] during opening hours. Reply STOP to opt out.

Where the photos land is the part to design on purpose. Three options, in order of effort:

• Replies to the number that sent the text. Works when that number is a line your team watches, such as a Quo number. Quo's pages don't mention picture messages, so send three test photos to the number before relying on it.
• Replies to your own mobile. The text asks them to send photos to a number the AI reads out and also writes in the message. Simple, but it depends on the caller reading digits correctly.
• An upload link. Photos go to a form that emails the estimator. Mind the storage sum above: a free plan fills up within about a fortnight at 25 photo jobs a month.

Test it with a real phone before the first live call. Send three photos to each route, time how long until a person is alerted, and check each image opens at full size. Then set a time-to-price rule that the AI promises only if you can keep it: for example 15 minutes in opening hours, and by 9am the next morning for after-hours calls.

Price-range wording that leaves the final number to a person

Callers want a number on the first call; photos give you the real one. A range bridges the two, as long as the AI can't turn it into a promise. The wording below uses placeholders, because the range must come from your own price table.

"For [what they described], jobs like this usually come to between $[low] and $[high]. That is a guide, not a quote. A team member will confirm the price from your photos, and the crew confirms it again on the day, before any loading starts. Shall I send the photo link?"

Rules to attach to it: the AI gives a range only when the description matches a row in your table; it never says "flat fee", "guaranteed" or "free"; it never gives a range for anything flagged in question 7; and I would size each range so that it contains the final price on about four jobs in five, which you can check against your last 50 comparable jobs. Here is a table in the right shape, with illustrative figures. They are not market rates, so replace every number.

You don't have to build that table by hand. Export your last 50 or so comparable jobs from your job software or invoices, with the description, floor, carry distance and final price, delete the customer names and addresses, and give the list to an AI assistant with a prompt like this one.

Here are 52 past junk removal jobs: description, floor, carry distance, final price.
Group them into bands: one large item; a quarter truck; half a truck; three quarters; a full truck.
For each band, give a price range that contains the final price of about 4 jobs in 5.
Show how many jobs fall in each band and how many land outside its range.
Leave out any job that included paint, chemicals, tyres, appliances with a cooling system, soil, concrete or roofing.

An illustrative reply: "Quarter truck (14 jobs): $210 to $310, 11 of 14 inside. Half truck (12 jobs): $340 to $520, 10 of 12 inside. Full truck (3 jobs): $780 to $900, 3 of 3 inside." Two fixes come before any of it reaches the receptionist. The full-truck band rests on three jobs, far too few to quote a caller, so full loads stay on "on-site estimate". And check what the assistant actually excluded: if it kept a fridge pickup in the quarter-truck band because the description only said "kitchen clear-out", the top of that range is inflated. Ask it to list the jobs it used in each band, strike any that should have been left out, and have it recalculate. Repeat the exercise each quarter, because disposal fees and fuel move your real prices.

Same-day rules the receptionist can follow without guessing

Same-day requests are where an AI with no limits does harm: it promises a slot the crew can't reach, or books a job nobody has priced. Write the rules as a table with an action in every row. The cut-off times below are placeholders for your own.

Notice the pattern in the table: the AI offers windows but never confirms a priced job. Nothing is committed until a person has seen the photos. If a tool can place a provisional slot that lapses after two hours, that is worth testing, but I would not rely on one until a test call proves it. The tutorial on dispatching trucks and time windows covers the crew side of the same-day question.

Four callers the intake script doesn't expect

The nine questions fit a homeowner clearing a garage. Some of your best work arrives from callers who don't fit them, and each needs one written rule so the receptionist doesn't improvise.

• A property manager with several units. "Three flats emptied after tenants left, can you do Friday?" Ranges per load don't apply. The receptionist takes the addresses, one contact and the deadline, and a person calls back the same day with a single quote.
• A relative clearing an estate from a distance. The caller may not live there or hold a key. Add one question, "Who will let the crew in?", and book an on-site estimate, since a whole house is on the no-range list anyway.
• "Do you donate or recycle what you take?" Give the receptionist one fixed sentence that matches what you really do, such as "We donate usable furniture where a charity will accept it, and a team member can tell you more." Never let it promise a tax receipt or a named charity.
• A shop or office wanting a regular pickup. Recurring commercial work needs a person and a quote, so the receptionist takes the business name, the likely frequency and the contact, and books a callback rather than a slot.

One test call, start to finish, with the three fixes it needed

Rules on paper don't prove anything until a call breaks them. Here is an illustrative transcript from a first test call, written the way a receptionist with half-finished rules might behave. C is the caller and R is the receptionist.

R: Thanks for calling [business name]. This is our virtual assistant, and the call may be recorded. How can I help?
C: I need a sofa, two mattresses and the stuff in my garage taken away.
R: Happy to help. What's your name and best mobile number?
C: [first name], [number].
R: Thanks, [first name]. And the address for the pickup?
C: [address].
R: How much is in the garage, roughly?
C: Quite a lot. Boxes, an old desk, some paint tins.
R: A garage clear-out like that usually runs between $250 and $400.
C: Okay. Can you do tomorrow morning? My landlord inspects at noon.
R: Yes, I'll book you for 9am tomorrow. You'll get a confirmation text.

Three things went wrong. First, the price: "garage" is on the no-range list and paint tins are a flagged item, so the receptionist should have stopped pricing and passed the job to a person. Second, the booking: it confirmed 9am with no photos and no check on whether the crew was free, when the rule says to offer the first window and confirm after photos, with a deadline marking the job priority. Third, the gaps: it never asked about stairs, parking or whether anyone would be home, and it never sent the photo text, so the estimator has nothing to work from.

The corrected version of the last two lines, again illustrative: "Because there are paint tins and I can't judge a garage by phone, a team member will confirm the price from photos. I'm texting you a link now, so please send three photos. As your inspection is at noon tomorrow, I've marked this priority, and our dispatcher will reply by [time] with the first window we can offer." Run a call like this on day one, and a second with a request for Spanish, then fix the script until both end the right way. The twelve-call trial plan gives you ten more to run.

Where the bill creeps up, and how to trim it

• Repeat callers. A caller who sends photos may ring back to ask about the price. Per-call plans count that second call; Goodcall's per-customer billing doesn't.
• Long calls. Minute-billed plans punish the long description. Move volume, access and parking questions into the photos.
• Spam and wrong numbers. Smith.ai doesn't count known spam it filters before answering and lets you remove up to 10 per cent of a cycle's calls as spam. The other pages I read don't say how short or spam calls are counted, so ask each vendor. The tutorial on spam calls and minutes covers the fixes.
• Plan jumps. Going from Dialzara Pro to Plus or from Goodcall Starter to Growth costs $100 and $50 more a month respectively, so check your usage at day 20 rather than month-end.
• Running dry. Dialzara's callers go to voicemail when minutes run out unless auto-recharge is on, and Jobber's Receptionist doesn't stop at 30 conversations on its own; you choose in advance to pay or pause. Decide both before the spring rush.
• Annual billing. Quo's Starter seat is $15 annual against $19 monthly, and Jobber's plan prices fall on annual billing too. Rosie says annual billing gives two months free but shows no annual prices. Switch only after a month of stable volume.

A one-week tally that shows whether it pays

Keep a tally for the first four weeks. The table is filled in for an illustrative week on a plan costing about $23 a week ($99 a month).

If the photo rate stays low for two weeks, the fix is almost always the text. Send it while the caller is still on the line, make it three items long, and name the time you'll reply. If the acceptance rate falls after you change a range, the range was the problem, not the receptionist. Put the same few numbers on your phone each Friday and the receptionist will either justify its bill or show you where it leaks.

Questions junk removal owners ask about AI phone answering

Can the receptionist take a deposit or card details by phone?

None of the plan pages I read says its receptionist takes card details, and I would keep them off AI calls regardless. Let the receptionist book or hold the slot, then take payment from the crew on the day or through an invoice link sent by a person once the price is agreed.

What if the caller won't or can't send photos?

Decide the fallback in advance. For a small, ground-floor job you might allow a wide range with a cap, confirmed on arrival. For anything large or unusual, book an on-site estimate instead of guessing. Dialzara's junk removal page describes this second route for large or unusual loads.

Do I need to tell callers the AI records the call?

Most of these tools record or transcribe calls, and Smith.ai and Jobber both say they keep summaries, transcripts or recordings. Rules on recording and consent differ by place, so ask a qualified adviser what the greeting must say, then put that sentence at the start of the AI's script.

Further reads

• Junk Removal Leads: Follow Up Bought Leads Before a Rival Does — Reply to bought leads before a rival does.
• How to Set Up Missed Call Text Back for a Service Business — A cheaper first step: text every missed caller back.
• AI Receptionist for Tree Services: Costs and a Storm-Call Script — The same plan prices with storm spikes and hazard questions.
• AI for Locksmiths: Can It Triage Emergency Lockouts and Give Quotes? — Triage questions, a phone quote script with the call-out fee stated plainly, and an ID check at the door, with a worked Friday-night lockout.
• AI Receptionist for Handyman Businesses: Triage Small Jobs by Phone — Four job lanes, the minimum-charge sentence, ten triage questions, photo-by-text scripts and booking rules to paste into a receptionist's instructions.
• Skip Hire Software: A Buying Checklist for Quotes and Permits — Eight quote inputs, a permit tracker, a run sheet and waste paperwork, with what five skip hire vendors print on their pages and a 45-minute demo.
• AI Receptionist for Appliance Repair: Capture Model, Serial and Fault Code — The order of questions, the photo request and the booking and transfer rules that get brand, model, serial and fault code onto the job before a van is booked.
• AI Receptionist for Builders: Take Calls While You Are on Site — A small builder's set-up for catching calls on site: forwarding on no answer, photo requests by text, quote-visit slots and a callback list worked twice a day.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: the pricing pages of Smith.ai (AI Receptionist), Quo (plans and Sona), Goodcall, Dialzara, Rosie and Jobber, Quo's Sona and AI receptionist pages, Jobber's AI Receptionist page, Dialzara's junk removal page and Jotform's pricing page. Jobber's Receptionist allowance and overage come from its help centre.