12 quotes. One pile. $1,025 apart.

The last price in America that is still invented at the kerb

Almost every price an American household pays has been dragged into daylight over the last twenty years. Airfares were opaque, then they were a grid of numbers you could sort. Hotel rooms were opaque, then they were a map with prices on it. Car insurance, mortgage rates, used cars, electricity in the deregulated states, the sticker on a hospital procedure — one by one, each of them stopped being a number that someone said to you and became a number you could look up, compare, and walk away from.

Junk removal never made that trip. The price of getting a sofa out of a living room is still set the way it was set in 1994. A truck arrives. Somebody looks at the pile. A number is said out loud, in a driveway, to a person who has already spent the morning carrying furniture down a staircase.

That number does not exist anywhere before it is spoken. There is no menu behind it, no rate card, no published schedule you can check afterwards. Search for what junk removal costs and you will find ranges — wide, confident, uncited ranges — and every single one of them was assembled the same way: somebody phoned a few companies, or read a few reviews, or reasoned backwards from the size of a truck. Not one of them is a record of what a specific pile was actually quoted by the people who would actually have hauled it.

We think this is the most interesting unsolved consumer-pricing problem left inside the American home. And we think it stayed unsolved for a boring reason rather than a hard one. It is not that the price is unknowable. It is that nobody ever ran the auction.

Between 2026 and 08-29 – 2026-09-01, we ran it. 12 independent haulers priced one pile of furniture in Rockville, MD. They worked from 9 photographs taken on a phone. Not one of them drove out to look. Not one of them saw what any of the others had written. The quotes came back between $225 and $1,250.

This page is about the space between those two numbers. It is long on purpose, because the gap is $1,025 on a single household job and a number like that deserves more than a headline. What follows is the whole measurement: how it was taken, what it does and does not prove, why we believe the same thing is sitting in every metro in the country, and what we intend to do about it.

Nothing about the pile changed between the cheapest quote and the dearest one. Only the person standing next to it changed.

Why the number gets invented at the kerb

The honest explanation is not villainy. It is structure. Three perfectly ordinary facts about this transaction combine into something that stops being ordinary.

The first fact: the customer cannot leave. By the time a quote is being given, the pile is already on the driveway, or the tenant is already out on Saturday, or the buyer's walkthrough is already on Monday. The job is not a want. It is a deadline with furniture attached to it.

The second fact: the truck is already there. A crew drove across a metro, burned forty minutes and a tank of fuel, and is now standing in front of a customer. Walking away from that visit with nothing costs the company real money. Both sides know it, and both sides know the other one knows it.

The third fact: neither party has a reference price. Not the homeowner, who has never bought this before and will not buy it again for years. And — this is the part people miss — not the hauler either. A small operator prices from their own book: what they charged last month, what the transfer station cost them last week, how full next Tuesday looks. There is no market number for them to check against, so their own book is the only truth available.

Put those three together and you get a pricing moment with a very specific shape. The number is named at the exact instant the customer has the least leverage they will ever have, by a party who is also guessing, with no third number anywhere in the room. That is not a market. That is a conversation.

And a conversation produces conversation outcomes. It rewards whoever has more information, more time, and less to lose by walking away. On a driveway, at nine in the morning, with a sofa already outside, that is never the homeowner.

What you are actually buying

Strip the transaction down and there are four things inside the price. Labour: two people, some number of minutes, stairs or no stairs. Disposal: the fee charged at the transfer station, the landfill or the recycler, usually by weight, sometimes by item. Vehicle: the truck payment, fuel, commercial insurance, maintenance, all of it amortised across the jobs that truck does that week. And margin: what is left over for the company that organised the other three.

Three of those four are knowable before anyone drives anywhere. Labour is a function of item count and access, both of which a photograph shows. Vehicle cost per job is arithmetic a company already does. And disposal — the one people assume is the mystery — is frequently a published number. Municipal transfer stations post their rates. Recycling programmes post their rules. We spend a great deal of our time collecting exactly that: what your city takes, what it charges, what it refuses, with the municipal source and the date we last checked it printed on the line.

Which leaves the fourth item. Margin is the only genuinely discretionary component, and it is the one that has never had to survive contact with a competing number.

This matters more than it sounds. If the spread on a job were being driven by cost — one company facing a higher tip fee, another with a longer drive — you would expect the quotes to cluster loosely around a common floor, separated by the real differences between operators. That is not what a wide spread looks like. A wide spread is what you get when the cost floor is roughly shared and the number on top of it is a judgement call made in private.

We are not accusing anyone of anything. A company that quotes high on a Tuesday when its truck is already full is behaving completely rationally. A company that quotes low to fill a dead afternoon is also behaving completely rationally. The problem is not that haulers price differently. The problem is that the homeowner only ever gets to see one of them.

The "free estimate" is not an estimate, and it is not free

An estimate, in any other trade, means a calculation performed in advance and then applied to observed facts. A builder estimates. An insurer estimates. What happens in a junk removal driveway is something else: a price offer, made under time pressure, with no competing offer present and no obligation to explain how it was reached.

The word "free" in "free estimate" describes the visit. It has never described the number. And the visit is not free either — it simply is not billed to the person receiving it.

Follow that cost. A company sends a crew to quote ten jobs. Some number of those convert; the rest are a total loss — fuel, hours, wear, and the opportunity cost of the work that truck did not do instead. Those losses do not evaporate. They are recovered from the customers who said yes. Every homeowner who accepts a kerbside quote is paying, invisibly, for the driveways where the answer was no.

That is worth sitting with, because it reframes what a photograph actually replaces. Quoting from photos is not a convenience feature. It does not merely save the customer an appointment window. It removes an entire category of unpaid cost from the industry — and unpaid cost, in any market, ends up in the price.

So when a system lets a hauler price a job in four minutes instead of ninety, the saving is not cosmetic and it does not belong to the hauler alone. It is the first structural cost reduction this trade has been offered in thirty years, and it is available to whoever is willing to give up the driveway.

The word "free" describes the visit. It has never described the number.

Three ways the internet tried to fix this, and where each one stalls

This is not an unnoticed problem. It has been attacked three times, by serious people, and it is worth being precise about why each attempt gets close and then stops.

The first attempt is the price guide. Thousands of pages answering "how much does junk removal cost" with a range and a table. The failure is arithmetic, not honesty: an average taken across unlike jobs is not the price of any job. A single-item mattress pickup and a full garage clear-out are different transactions in different cities under different disposal regimes. Averaging them produces a number that describes nothing that exists. Worse, the ranges are usually assembled from anecdote — what people said they paid, gathered by someone who was not there.

The second attempt is the lead form. "Get four quotes!" You fill in a box; your details are sold to four companies; four companies telephone you; and then four companies come and quote you in the driveway, one at a time, exactly as before. This moves the phone call. It does not move the pricing moment. In many cases it makes the experience worse, because you have now traded your phone number for the privilege of having the same conversation four times.

The third attempt is the national flat-rate brand, which publishes a truck-fraction price list and sticks to it. This one is genuinely useful and genuinely more honest than the first two. Its limit is structural: it is one company's price list, and the fraction of the truck your pile represents is still judged on site, by that company. A published price from a single seller is a rate card. It is not a market.

What all three are missing is the same thing, and it is not effort or good faith. It is simultaneity. None of them ever puts multiple real, competing, binding numbers on one identical, documented job at the same moment. Until somebody does that, there is nothing to compare, and the price stays a conversation.

How the trade got here

It is worth a short detour into how an entire industry ended up pricing this way, because the answer is not that anybody chose it.

Hauling rubbish for households began as the most local business imaginable: one person, one truck, a phone number written on the door. Pricing was whatever the driver thought the job was worth and whatever the customer would agree to. With one truck and one neighbourhood, that works. There is no meaningful information asymmetry when the man with the truck lives four streets away and will see you at the shop on Saturday.

Then the trade professionalised — uniforms, dispatch, call centres, franchising, regional and national brands, proper insurance, legal disposal, background-checked crews. All of that was a genuine improvement and most of it made the service better. But the pricing convention came along untouched. The truck still drove out; somebody still looked at the pile; a number was still said out loud. What changed was the size of the company saying it, not the mechanism.

Then the internet arrived, and here is the thing that matters most: it arrived as advertising. It did not arrive as a market. The first serious money into home services online went into buying the customer's attention and reselling it — directories, then lead generation, then paid search. Every one of those businesses is compensated for delivering a customer to a company. None of them is compensated for delivering a price to a customer. So two decades of investment went into making the introduction faster while leaving the driveway conversation exactly as it was in 1994.

That is not a conspiracy, it is an incentive. If your revenue is a fee per introduction, price transparency is at best irrelevant to you and at worst actively harmful, because a customer who can compare numbers needs fewer introductions. The gap in this market was never technological. It was that nobody whose business model pointed at it had any reason to close it.

Which is, in the end, an encouraging thing to discover about a problem. Hard problems stay unsolved because they are hard. This one stayed unsolved because everyone standing near it was being paid to look the other way.

What actually moves the price

If you are going to compare quotes, it helps to know what the people writing them are looking at. These are the things that genuinely change the number, roughly in order of how much they change it.

Volume, first and by a distance. Not weight, in most cases — how much of the truck your pile occupies. This is why the trade quotes in fractions of a load rather than in items, and why two sofas can cost less than one oddly-shaped exercise machine that nothing else will stack against.

Access, second, and it is the one households consistently underestimate. Ground floor with a driveway is one job. Third floor walk-up with a shared staircase and permit parking is a different job with the same furniture in it. Add a lift with a booking sheet, a long carry from a rear entrance, or a street where the truck cannot legally stop, and you have added real hours.

Disposal class, third. A load of mixed household furniture goes one place at one rate. Mattresses, tyres, paint, electronics, appliances with refrigerant, anything with a battery in it — each of those may be regulated separately, charged separately, or refused outright, and the rules change by city and sometimes by county. This is the part that people assume is trivial and that quietly decides whether a job is a hundred dollars or several hundred.

Labour, fourth, which is mostly downstream of the first three, plus whatever has to be dismantled to get through a doorway. A wardrobe that has to come apart on a landing is a different proposition to one that walks straight out.

And timing, fifth. Saturday costs more than Tuesday, because everybody wants Saturday. Same-day costs more than next-week, because it displaces something already booked. A household with a flexible date has a lever most households never think to pull, and it is frequently worth more than any haggling they could have done in the driveway.

Nothing in that list is secret, and none of it is difficult. But notice what it takes to apply it: you have to be able to see several numbers written by people who weighed those factors differently. One quote, taken alone, tells you nothing about which of those five the company was pricing conservatively. Twelve quotes tell you a great deal.

One pile, 9 photographs, 12 sealed numbers

Here is the entire event, with nothing removed and nothing dramatised.

A household in Rockville, MD had a furniture removal job. Furniture, photographed on a phone — 9 frames, taken the way anyone takes them: whatever was in the room, from whatever angle was available. Those photographs, a postcode, an access note and a preferred date were the complete input. That is the whole listing. There was no phone call, no appointment window, no crew standing in the driveway with a clipboard, and no conversation with anybody before the numbers arrived.

Between 2026 and 08-29 – 2026-09-01, 12 independent haulers priced it. Each of them looked at the same 9 frames. None of them visited. None of them saw another hauler's number, before or after submitting their own. Each wrote down what they would do the job for.

The lowest was $225. The highest was $1,250. The middle number was $525. The distance between the cheapest and the dearest was $1,025, which is 5.6 times the cheapest quote — on a pile that did not move, did not change, and did not get any larger between the first quote and the last.

That is the measurement. It was verified against our bid ledger on September 4, 2026 and the raw figures are published, with a CSV, on our bid data page. There is no proprietary interpretation layer here and no adjustment. Twelve people looked at the same thing and wrote down what it was worth to them.

We want to be precise about what the ledger records, because precision is the only thing that makes a page like this worth reading. The ledger records quotes. It records what 12 businesses were willing to commit to in writing, from photographs, without negotiation. It does not record which quote the household chose, or whether they chose one at all. This page is about price discovery, not about a completed haul — and we would rather publish the narrower true thing than the wider impressive one.

The 12 numbers, and what the shape of them means

A range is a weak way to describe a distribution. "$225 to $1,250" tells you the two extremes and hides everything in between, and everything in between is where the finding actually lives.

Look at the strip above and the table below. The quotes are not spread evenly across the range. They bunch. Most of the 12 numbers sit in the lower half, close to one another, within a band that looks very much like a real market price for the job. Above that band the points thin out fast, and the top of the range is held up by a small number of quotes standing well clear of the rest.

That shape is the whole argument of this page, and it argues something more interesting than "haulers overcharge". It says there was a consensus price for this pile — a real one, arrived at independently by most of the people who looked at it — and that a household calling one company would have had no way of knowing whether the number they got was inside that consensus or a long way above it.

Consider it from the homeowner's side. You make one call. You get one number. Whatever that number is, it arrives with the same confidence, in the same tone, from a person who does this every day. Nothing in that moment tells you whether you are being quoted near the middle of the market or near the top of it. The information that would tell you does not exist, because it is only created when several people price the same thing at once.

And consider it from the hauler's side, because they are not villains in this story. A company that quoted near the top was not necessarily gouging. They may have been full that week. They may price stairs conservatively. They may simply have a more expensive cost base. Under the old arrangement none of that had to be explained, because nothing forced the comparison. Under this one it does — and the ones who price keenly, who have a dead Tuesday to fill, who are efficient at the transfer station, finally have a way to be found.

That is the part we would ask an investor to look at twice. This mechanism does not work by taking margin away from suppliers and handing it to consumers. It works by making the existing variance visible, so that demand can find the operators who were always cheaper and never discoverable.

What $1,025 actually is

Money on a page goes numb quickly, so here is the same figure said a different way.

The gap between the cheapest and the dearest quote on this one job was $1,025. The middle quote on that same job was $525. The gap is 1.95 times the price of the job itself. The difference between two opinions about one pile of furniture was worth substantially more than the pile was worth moving.

Think about what else that is. It is a month of groceries in most American households. It is a security deposit. It is the difference between replacing the appliance and repairing it again. For a family clearing a parent's house — which is a large share of why anybody books this work at all — it is a meaningful fraction of everything the week is going to cost them.

This is why we are uninterested in arguing that junk removal is cheap. It is not always cheap, and pretending otherwise would be the same kind of lie the price guides tell. The argument is narrower and much harder to rebut: whatever the job is worth, you should be allowed to see more than one opinion about it before you decide, and until now you could not.

That is also the answer to the question we get most often, which is some version of "how much cheaper are you?" We do not know, and any company that claims to know that number about your particular job is making it up. What we know is that on this job, the numbers ranged $225 to $1,250, and that a household shown all 12 of them is in a completely different position from a household shown one.

The difference between two quotes for the same pile was 1.95 times the job's own middle price.

The job where the market agreed — and why we publish it

If every job we ever ran produced a spread like Rockville, MD's, this page would be propaganda, and a careful reader would be right to distrust it. So here is the other job in the ledger, and it says something close to the opposite.

In Los Angeles, CA, a mattresses & beds removal job drew 4 quotes. The lowest was $800. The highest was $920. The entire spread was $120 — a factor of 1.15. Four independent companies looked at the same job and landed within a few percent of each other. For that household, running an auction confirmed the price rather than collapsing it. Everyone was, in effect, right.

A less careful company would leave this out. We publish it first because it is true and second because it is the most important finding we have. Compare the two jobs and the conclusion is not "haulers are erratic" — it is that spread is a property of the job, not a property of the industry.

Some work is standardised enough that the whole market agrees on it: a known item, a known weight, a known disposal fee, a predictable hour. Other work — mixed furniture, ambiguous access, judgement about what fits and what has to be broken down — is priced from experience, and experience differs. The first kind produces a cluster. The second produces a canyon.

Here is the consequence, and it is the one that makes this a business rather than a curiosity: you cannot tell which kind of job you have until after the quotes are in. The household in Los Angeles, CA did not know their job was a cluster job. The household in Rockville, MD did not know theirs was a canyon job. Neither could have known, because the only instrument that measures it is the auction itself.

Which means the auction is worth running on every job, including the ones where it will turn out to have changed nothing. A household that runs it and learns the price was fair has bought certainty at no cost. A household that runs it and finds a $1,025 spread has bought something rather more valuable. There is no third outcome where running it made things worse.

What this page does not claim

A launch page is expected to overstate. This one is trying not to, so it is worth writing the limits down explicitly rather than burying them — partly because it is right, and partly because in a category this full of unverifiable claims, the arithmetic you are willing to show is the only thing that distinguishes you.

We do not claim this job was completed by us. The ledger records what was quoted, not what was accepted or hauled. We are publishing a price-discovery event, and we would rather say so plainly than let the word "project" imply a finished job.

We do not claim these 12 firms represent the national market. The full ledger behind this page is 16 quotes, 2 jobs, 2 metros. That is enough to show what happens on a single job. It is nowhere near enough to describe a country, and anyone reporting these as average American junk removal prices is misreading them.

We do not publish company names. The haulers quoted commercially, not for publication, and putting their pricing on the internet under their own names is not something they agreed to. We publish the numbers because the numbers are the finding; the identities are not ours to spend.

We do not publish the household's name, address or photographs. They are a private person with a private driveway. The city, the job type, the frame count and the figures are the entire published record, and that is by design rather than by omission.

We do not promise anyone $225. $225 was the lowest of 12 quotes on one specific pile in one specific week. Your pile is not that pile. Any company that quotes you a price before seeing your job is guessing, and that includes us.

And we are not a hauler. We do not own trucks, employ crews, or take your sofa anywhere. Companies bid; you choose; they do the work. We say this on every page we publish because it is the fact that most changes how you should read everything else here: we have no fleet to keep busy, and therefore no reason to steer you anywhere except toward the number you like best.

The other price: what the old way costs in hours

Everything so far has been about money, which is the measurable part. There is a second price in this transaction and households pay it without ever seeing it itemised.

Doing this properly under the old arrangement means three companies, three phone calls, three appointment windows and three strangers in your hallway — and the windows are not an hour, they are a morning. Most people, entirely reasonably, decline to spend two working days collecting quotes for a job that might cost a few hundred dollars. So they call one company, take the number, and never learn what the other two would have said.

That is the decisive fact about the old mechanism, and it is worth stating flatly: the reason households do not comparison-shop junk removal is not laziness. It is that comparison-shopping was priced, in hours, above the value of the information it produced. The market was not failing because people were careless. It was failing because it had made carefulness irrational.

Which is exactly what changed here. The Rockville, MD household spent a few minutes photographing a pile and received 12 numbers between 2026 and 08-29 – 2026-09-01. They did not take a day off. They did not meet anybody. The cost of getting the twelfth opinion was identical to the cost of getting the first, because the listing was already written and the marginal hauler simply opened it.

Pushing the cost of an additional quote to zero is the whole mechanism, described in one sentence. Everything else on this page — the spread, the cluster, the consensus band — is a consequence of that one economic change, and it is the same change that made every other opaque consumer market legible. Nobody compared airfares by ringing eight airlines either.

9 photographs are what turn a sequence into an auction

The single design decision everything else rests on is unglamorous: the job is described by photographs instead of by a visit.

People hear that and assume it is about convenience — no appointment window, no stranger in the hallway on a Saturday morning. Those are real, and they are not the point. The point is arithmetic. You cannot put 12 companies in one driveway at the same moment. It is physically impossible and commercially absurd. You can put 12 companies in front of the same 9 frames at the same moment, and it costs nobody anything.

The photograph is the unit that makes simultaneity possible, and simultaneity is the entire difference between a series of negotiations and an auction. Everything this page describes — the spread, the cluster, the consensus band, the discovery — exists downstream of that one substitution.

The obvious objection is whether a photograph is enough. In practice a working hauler reads a great deal from a frame: how many pieces, what material, whether anything has been taken apart, whether the sofa is a two-person carry or a four, where the truck will park, how many stairs, whether the doorway is going to be the real problem. These are the things that drive the cost, and most of them are visible.

Some things are not visible, and we do not pretend otherwise. Weight is the honest example: a waterlogged mattress and a dry one look identical and do not cost the same at the scale. So haulers can ask. A question goes back to the household, the answer goes to everyone, and the ones who needed it price with it. The ones who did not need it have already bid. Nothing about the mechanism requires the photograph to be perfect. It only requires it to be shared.

And there is a second-order effect worth naming. Because the description is a durable artefact rather than a conversation, everybody is pricing the same job. Under the old arrangement, four companies visiting on four different days were quoting four subtly different jobs — the pile grew, the customer remembered something else, the access looked different in the rain. A quote set gathered that way cannot be compared even in principle. One taken from one set of frames can.

You cannot put 12 companies in one driveway at once. You can put 12 companies in front of the same 9 frames at once.

Sealed bids, not a race to the bottom

Bidders cannot see each other's numbers. That is a deliberate choice and it is worth defending, because the obvious alternative — an open, visibly descending price — looks better for the customer for about one quarter.

In an open undercut, each company sees the leading number and shaves it. The floor falls until it reaches the cost base of whichever operator is most desperate, or most mistaken about their own costs. The customer wins that job spectacularly. Then the careful operators, the insured ones, the ones who pay crews properly and dispose legally, stop bidding — because they cannot win and they are tired of losing — and within a year the only companies left in the mechanism are the ones who should not be trusted with your staircase. The customer who won in January is choosing from a worse market in December.

A sealed single-shot bid does something different. Each company prices from its own economics: what the job costs them, what their week looks like, what margin they need. Nobody is bidding against a number on a screen. They are bidding against their own honest floor, which is exactly the number we want the household to be able to see.

The second half of that design is that the household is not obliged to take the lowest quote, and we never present the lowest as the correct answer. The quotes arrive alongside the profile of the company that made them — what previous customers said and what they will and will not take. Plenty of households look at a set of numbers and choose the third one because the company that wrote it answers questions properly. That is a market working, not a market failing.

This is also why we do not run a reverse auction with a timer, and why there is no countdown clock anywhere in the product. Urgency is the thing that broke the original transaction. Rebuilding it in software would be a strange way to fix it.

Why a hauler agrees to bid at all

A reasonable sceptic stops here and asks the sharpest question available: why would any company participate in a system whose stated purpose is to expose how differently they price?

The answer is that we are not asking them to give up margin. We are offering to buy back their worst cost.

The scarce asset in this trade is not trucks. Trucks can be bought. The scarce asset is booked truck-hours. An empty Tuesday afternoon is a total loss, and it is a loss that is already paid for: the vehicle payment is due, the insurance is running, and in most operations the crew is being paid whether or not there is a job to do. A job that fills that afternoon at a thin margin is not a bad job. It is dramatically better than the alternative, which produces nothing at all.

Now put the driveway estimate next to that. A visit consumes forty minutes of driving, twenty minutes on site, forty minutes back — most of a productive block — and a significant share of those visits end in no work. That is the industry's largest unpaid expense and it never appears on an invoice, which is precisely why it has never been attacked.

Quoting from photographs costs a hauler minutes. A dispatcher opens the listing between jobs, reads the frames, writes a number. If they win it, the truck is full. If they lose it, they have spent four minutes rather than a half-day. Over a month, a company doing this at any volume gets back a substantial amount of capacity that it was previously giving away for free.

That is the trade, and we state it to haulers in exactly these terms: you give up the pricing advantage of being the only number in the room, and you get back the hours you were burning on estimates that went nowhere. For an efficient operator with gaps in the calendar, that trade is not close. For an operator whose model depends on being the only number in the room, it is a bad trade, and they do not join. We are comfortable with both outcomes.

There is a longer-run argument too. The variance this page documents is not stable. Price transparency arrives in every category eventually, and it has never once been repelled by the incumbents who preferred opacity. The companies that do well through that transition are the ones that were already competitive and simply could not prove it. Those are the companies we are built for.

Five things we deliberately do not do

Most of what makes this mechanism work is restraint. Each of the following is something the business could do tomorrow, would make money from immediately, and would break the thing that makes the data worth publishing.

We do not sell the same lead to several companies. That is the dominant business model in home services and it is the reason the category has the reputation it has. A lead sold four times produces four phone calls and zero price discovery, because the four companies are still each quoting you separately and privately. We are not in the business of selling your phone number to people who will then quote you in a driveway.

We do not mark up the hauler's number. The figure you see is the figure the company wrote. If we quietly added to it, every number on this page would be fiction and the ledger would be worthless — which is the practical reason as well as the ethical one.

We do not own trucks. This is the structural guarantee behind everything else. A marketplace that also operates a fleet has a reason to steer jobs toward itself, and everybody in that market knows it. We have no crew to keep busy, so we have no reason to prefer any quote over any other.

We do not hide who is cheap. There is no paid placement, no sponsored bid, no way to buy the top of the list. If a company's number is the lowest, it appears as the lowest. Selling that position would be the most immediately profitable thing on this list and it would destroy the product in a single quarter.

And we do not publish averages we assembled. Every figure we put on the internet is either a real quote from the ledger, or a municipal rule with its source and check date attached. When we do not have the data for a city, the page says so instead of filling the gap with a plausible range. An incomplete page is merely incomplete. A confident wrong page is the thing we are trying to replace.

The four objections we hear most

These come up every time we explain the mechanism, usually in this order, and all four are reasonable.

"The cheap bid is bait — they will raise it on the day." This is the fear that kills comparison shopping, and it is not imaginary; anybody who has hired a trade has a story. The structural answer is that a quote written from photographs, visible next to eleven others, alongside the company's own record, is a far worse instrument for baiting than a number said out loud in a driveway with nothing to compare it to. Baiting works in private. It is a poor strategy in a room where the customer is holding the other numbers. Where a job genuinely turns out to be different from its photographs — it happens — the honest resolution is a re-quote the household can decline, not a surprise at the kerb.

"A low bid means a bad company." Sometimes. Often it means an empty Tuesday, a transfer station eight minutes away rather than forty, a two-person crew that is quick on stairs, or a smaller operation with a smaller overhead to carry. The whole point of putting the quotes next to each other is that you are not forced to guess: the number arrives with a profile attached, and nothing obliges you to take the lowest one. Plenty of households do not, which is a market working rather than failing.

"What about insurance and liability?" A fair question and one this trade has historically answered badly. It is also one of the few things a marketplace can enforce more consistently than a phone call can, because participation is conditional and revocable. A household hiring from a driveway has whatever assurances the person in front of them chose to offer.

"What if nobody bids?" Then we have failed on that job, and the honest thing is to say so rather than to present one quote as a market. This is the real constraint on expansion and we treat it as such: the mechanism needs a floor of participating haulers in a metro before it produces anything worth the household's attention. It is the reason the map fills in city by city instead of all at once, and the reason we would rather be genuinely useful in a small number of places than nominally available everywhere.

Where it actually goes — and why that is half the price

There is a part of this transaction that almost no consumer page discusses, and it happens after the truck leaves.

Every load ends up somewhere, and somewhere charges. Transfer stations weigh the vehicle in and out and bill by the ton. Landfills have their own rates and their own refusals. Recyclers take some materials gladly and others only under a programme with its own rules. Mattresses are the clearest example in the country: in several states there is a funded recycling scheme with specific drop-off points and specific conditions, and in the rest of the country there is not, which means the same object costs a very different amount to get rid of depending on which side of a state line it is sitting on.

That is why we spend so much time on something that looks, from outside, like an odd thing for a marketplace to publish: municipal disposal rules, city by city, with the source and the date we last checked them. It is not content marketing. It is the cost floor of the service, and it is the only component of the price that is genuinely public information — which makes it the one part of the number a household can verify without us.

It matters for a second reason too, and this one is less comfortable. When legal disposal is expensive and nobody is checking, there is a shortcut, and it is the reason for every dumped sofa on every quiet road in America. A price that looks too good sometimes reflects an efficient operator. Sometimes it reflects a company that has decided the transfer station is optional. Publishing what legal disposal actually costs in a given city is the cheapest available defence against that, because it tells a household what a number has to cover before it can possibly be legitimate.

We would rather a household paid the middle quote to a company with a weighbridge receipt than the lowest quote to one without. The mechanism cannot make that choice for anyone. What it can do is put both numbers in front of them, next to the rules, on the same screen.

Reviews stopped working. Numbers still work.

For about fifteen years the answer to "which company should I use" was reviews, and for a while reviews were genuinely good. A five-star average from two hundred people told you something no advertisement could.

Then the score inflated. Every company in every category is now rated somewhere between 4.6 and 4.9, which means the number has stopped separating anybody from anybody. Households know this. They have learned to skip the average entirely and read the one-star complaints looking for a pattern, which is an enormous amount of work to extract a signal that used to be sitting at the top of the page.

There is a deeper problem too, and it is specific to this trade. Reviews measure the experience — were they punctual, were they polite, did they break the bannister. They do not measure the price, because the reviewer does not know whether they were quoted well. Nobody writes "three stars, pleasant crew, but I later discovered eleven other companies would have done it for less." They cannot write it, because they never found out. The one dimension where this market fails hardest is the one dimension reviews are structurally blind to.

A quote set is not blind to it. Put 12 numbers on one job next to each other and you have measured exactly the thing the star rating cannot see — and you have measured it without asking anyone to be honest, without a moderation team, and without a way for anyone to buy a better result. The number is the company's own commitment, written before it knew what the others had said.

This is why we put so much weight on publishing figures rather than testimonials, and why you will not find a wall of quoted praise anywhere on this site. We would rather show you a distribution than a compliment. A compliment can be manufactured for a few dollars. A distribution has to be produced by actually running the thing.

Why this is not a Rockville, MD story

The temptation at this point in a launch page is to multiply. Take the gap, multiply it by every household in America, and put the result on a slide. We are not going to do that, and the reason is not modesty. It is that a number produced that way would be the same species of fiction as the price guides we have spent this page criticising, and we would deserve to be caught.

So let us separate what we measured from what we infer, and label both.

Measured: 12 quotes on one furniture removal job in Rockville, MD, spanning $225 to $1,250, gathered between 2026 and 08-29 – 2026-09-01, verified September 4, 2026. Plus 4 quotes on a mattresses & beds removal job in Los Angeles, CA, spanning $800 to $920. Total ledger: 16 quotes, 2 jobs, 2 metros. That is the entire evidentiary base and it is small.

Inferred, and labelled as inference: the three conditions that produced the Rockville, MD spread are not local features of one metro. The first is that no price is published anywhere before the quote is given. The second is that the quote is given at the moment of maximum customer urgency. The third is that supply is fragmented into a very large number of independent operators, each pricing from a private book. Those three conditions describe essentially every metropolitan area in the United States, and they are the mechanism by which a wide spread is generated.

There is a check available on that inference, and it is the reason the second job in the ledger matters so much. If the Rockville, MD spread had been caused by something particular to that city — an unusual disposal regime, an odd cluster of operators, a bad week — then a job in an entirely different market with entirely different companies should have behaved similarly. It did not. Los Angeles, CA came back as a tight cluster. The variance is tracking the job type, not the postcode, which is precisely what you would expect if the cause is structural rather than regional.

None of that makes a national claim. It makes a testable one, which is a better thing to have at this stage. The test is simple and it is the only thing we are actually asking anyone to fund: run the auction in more metros, publish every result including the boring ones, and find out whether the pattern holds.

Fragmentation is the asset, not the obstacle

Most marketplaces are built against consolidated supply and spend their first five years fighting for the right to exist. This one is the opposite case, and the difference is worth making explicit because it inverts a lot of conventional intuition.

A market with three large sellers cannot be arbitraged. The prices are already known, already near-identical, and already visible on three websites; a comparison layer over that market adds convenience and nothing else. Junk removal is not that market. It is tens of thousands of independent operators, single-truck and five-truck businesses, each with its own cost base, its own week, its own view of what stairs are worth, and no mechanism whatsoever for any of them to learn what the others are charging.

Extreme fragmentation with no price signal is the precise condition that produces enormous internal variance. And variance is the raw material an auction converts into consumer surplus. The more fragmented the market, the larger the prize — which means the thing that has kept national players from consolidating this industry for forty years is the same thing that makes it worth building a marketplace on.

Our own ledger, small as it is, shows the effect in miniature. 16 quotes were gathered across 2 jobs in 2 metros. Two ordinary household jobs, posted in two cities, drew that many separate pricing opinions — and produced a spread on one of them of $1,025. Not because anyone was hunting for outliers, but because that is what the market looks like when you ask more than one person.

It is also why the incumbents cannot simply copy this. A national brand can publish its own rate card tomorrow; several have. What it cannot do is publish what its competitors would have charged for the same pile, because it has no way to find out and no reason to want to know. The data on this page is not a content asset that a larger company could commission. It only exists as a by-product of running the auction.

A directory goes stale. An auction gets truer.

Everything that has previously been built in this category has been, underneath, a directory: a list of companies, a set of pages, a snapshot of a market taken on a particular day. Directories decay. The day you publish one it begins drifting away from reality, and the only remedy is to keep re-typing it.

An auction has the opposite property. Every job that runs through it produces new real quotes. Those quotes improve the published price data. Better price data is what people actually searching for "how much does this cost" are looking for, so it earns the traffic. That traffic brings more jobs. More jobs bring more quotes. The asset is not a list someone maintains; it is a residue the machine leaves behind while doing its job.

This is also the defensibility answer, and it is an unusually clean one. Our bid data cannot be scraped, because it does not exist on anyone else's website. It cannot be licensed, because no one owns it. It cannot be bought, because it is not for sale anywhere — it does not exist at all until the moment several companies price the same pile at once. A competitor who wants it has to run the auction, and running the auction is the business.

The same compounding shows up on the supply side. A hauler who has won work through the mechanism has a reason to check it tomorrow, which shortens the time to first quote, which makes the experience better for the next household, which brings the next job. None of this is exotic network-effect theory. It is just the ordinary consequence of a mechanism whose output happens to be the same thing that attracts its input.

And it is why we publish everything. The free-to-cite dataset, the CSV, the municipal disposal rules, the city pages, the cost breakdowns: each of those is the residue being put to work. A company whose moat is a secret has to spend money keeping it secret. A company whose moat is a measurement can spend the same money making sure everyone sees it.

What scales, and what stubbornly does not

Being useful to anyone evaluating this seriously means being as clear about the hard part as about the easy part.

What scales cleanly is almost everything on the demand side. The listing format is the same in every city. The bidding mechanics are the same. The price data compounds nationally rather than locally, because a household comparing quotes wants to know how spreads behave, not only how they behaved on their own street. And the reference layer scales with effort but without invention: we publish 30 cost guides and 577 city pages today, each carrying municipal sources and check dates, and adding the next one is work rather than research.

What does not scale automatically is supply density, and there is no clever way around it. The auction needs a floor of participating haulers in a metro before it produces a spread worth looking at. One quote is not an auction; it is a quote. Two is barely better. The mechanism only starts doing its job somewhere above that, and until it does, the honest thing is to say so to the household rather than to present a thin result as a market.

That constraint dictates the shape of expansion. It has to be metro by metro, and it has to be demand-led — which, fortunately, is measurable. We can see which cities are already asking the question, in volume, in search. Those are the cities where a household will find us the day the supply is ready, and they are the order in which the map gets filled in.

The second hard part is the one nobody enjoys writing down: we do not yet know the durable conversion rate from a set of quotes to a booked job, because we have not run enough jobs to know it. We know what the quotes do. We do not know what the households do, at scale, over time. Anyone who tells you they know that number about an early marketplace is showing you a model, not a measurement, and we would rather hand you the gap than fill it in with something that sounds good.

We will publish that number when we have it, in the same place and on the same terms as everything else here — including if it disappoints us.

What is actually running today

Launch pages have a habit of describing a company's intentions in the present tense. Here is the separation between what exists and what is planned, so you can tell which is which.

Running now: the listing and bidding mechanism that produced every number on this page. A household posts photographs, a postcode and a preferred date; registered haulers in that area price it without a visit; the quotes arrive together and the household chooses. That is not a prototype or a pilot programme. It is the thing that generated the 12 quotes in Rockville, MD and the 4 in Los Angeles, CA.

Running now: the reference layer. 30 cost guides explaining how a price is built rather than asserting one, and 577 city pages carrying local disposal rules with the municipal source and the date we last checked each line. This is the least glamorous work in the company and the most quoted, because it answers questions nobody else has bothered to answer precisely.

Running now: the published ledger. Every quote behind this page is on the bid data page with a CSV, free to cite, updated from the same source this page reads — which is why the numbers here cannot drift away from the numbers there.

Running now: a daily local desk. We track what changes in the cities we cover — collection schedules, drop-off events, programme changes, the ordinary municipal news that decides whether a household can get rid of something for free this week — and publish it as it happens. It is the same discipline as the rules pages, applied to a calendar instead of a rulebook.

Not running yet, and we are not going to describe it as if it were: national coverage. The auction is only as good as the number of haulers bidding in a given metro, and that number is built city by city. Where we do not have the supply to produce a real quote set, we would rather tell a household that plainly than hand them one number and call it a market.

That is the entire inventory. It is smaller than the ambition on this page and it is deliberately listed next to it.

If you have a pile

Photograph it. Whatever is there, from wherever you are standing. The household in Rockville, MD used 9 frames taken on a phone, and that was enough for 12 companies to price the job.

Post it with a postcode and the date you would like it gone. It takes a few minutes and it costs nothing, and there is no call afterwards unless you ask for one.

Then wait, and read the numbers when they arrive. Some of them will surprise you. Take the one you want — the cheapest, or the third one because you liked how they answered a question, or none of them, which is also a perfectly good outcome and leaves you knowing something you did not know this morning.

The only thing we would ask you not to do is what almost everyone still does, which is take the first number because it was said out loud by somebody standing in your driveway.

If you drive the truck

You already know the two numbers this page is about, because you have lived inside both of them. You know what an empty Tuesday costs you, and you know how many hours a month you give away quoting jobs that go nowhere.

What we are offering is narrow and specific: real jobs, described well enough to price from a screen, in your service area, with no charge for looking. You read the frames, you write a number, you move on. Minutes, not a half-day. If you win it, the afternoon is full. If you do not, you are out four minutes.

What we ask in return is equally specific. Price honestly, because the household is seeing several numbers and yours will be read next to them. Turn up when you said you would. Dispose of the load legally, which in most metros means a transfer station with a scale and a receipt rather than a quiet corner of somewhere else.

We are not trying to squeeze you. We take no cut of your number, we do not sell your competitors the same job as a lead, and we do not put anybody at the top of a list for money. If you are efficient and you have been unable to prove it to anyone who was not already standing in front of you, this mechanism exists to make that provable.

If you are reading this as a business

Here is the case, stated the way we would want it stated to us.

The wedge is a question with large, measurable, national search demand and no honest answer anywhere on the internet. Millions of American households ask what this costs. Every page that currently answers them is estimating. We are, as far as we can establish, the only party publishing what independent companies actually quoted on identified jobs — with the raw figures, the date, the frame count and the CSV attached.

The asset is that data, and it compounds. It is created as a by-product of the transaction rather than bought, so its marginal cost is approximately zero and its volume rises with usage. It cannot be scraped from a competitor, licensed from a data vendor, or commissioned by a larger incumbent, because it does not exist anywhere until somebody runs several companies against the same pile at once. The moat and the product are the same object.

The cost structure is unusually clean for this category. No fleet. No crews. No fuel, no commercial vehicle insurance, no depreciation. And — the one that matters most in home services — no unpaid estimating cost, because nobody drives anywhere to produce a quote. The marginal cost of running one more auction is close to nothing, which is the property that makes the fragmentation argument above worth anything at all.

What we do not know, and will not model for you: the durable take rate, the repeat rate, the acquisition cost by metro, and the conversion from a quote set to a booked job at scale. We have not run enough volume to know any of those, and every one of them is a number that early marketplaces routinely invent. We would rather show you the hole.

What we do know is written on this page and checkable within about ten minutes: 16 quotes across 2 jobs in 2 metros; a $1,025 spread on one of them and a $120 spread on the other; 9 photographs and no site visit; 30 cost guides and 577 city reference pages published with municipal sources and check dates. It is small. We are telling you it is small, on the marketing page, in the same typeface as the good news.

That last decision is not squeamishness, it is the strategy. This is a category where every claim is unverifiable and consumers have learned to discount all of it to zero. The company that publishes its arithmetic — including the n, including the job where the auction changed nothing, including the things it has not measured yet — is the one that gets cited, gets linked, gets ranked, and gets believed. Trust is the scarcest input in home services and it is the only one we can manufacture from a standing start.

The ask is straightforward: this pattern needs to be tested in more metros, and the results need to be published whether or not they flatter us. If you build, invest, write, or run a company on the supply side of this trade, the contact page reaches a person rather than a form.

In a category built on unverifiable claims, showing your arithmetic is not a virtue. It is the strategy.

One more thing

Everything on this page is free to cite.

The bid data page carries the full ledger — every individual quote, the job, the city, the window, the verification date — and a CSV download. No account, no email, no attribution deal, no licence negotiation. Journalists, researchers, competitors and the companies who bid are all welcome to take it and use it.

That looks like a strange thing for a company to do with its only proprietary asset, so here is the reasoning. The dataset is not valuable because we are the only ones holding it. It is valuable because we are the only ones who can keep producing it. Giving away last month's measurement costs us nothing and makes the next one more likely to be read.

And there is a simpler motive underneath. The reason this trade prices the way it does is that nobody has ever been able to look the number up. Charging for the answer would be a peculiar way to fix that.

The measurement is the product. So we are giving the measurement away.

Method, sources, and how to check us

Every figure on this page is read at page load from the bid ledger. Not one of them is typed into the text. If the ledger changes, this page changes with it, which is a deliberate constraint: it makes it impossible for the copy to drift away from the data over time, and it means a stale claim here would be a code defect rather than an editorial one.

The ledger records, per job: the metro, the job type, every individual quote amount, the window during which the quotes arrived, the number of photographs in the listing, and the date the entry was last verified against our own bid records. The entries behind this page were verified on September 4, 2026.

What is deliberately absent: hauler names and identifying details, the household's name and address, and the listing photographs. Those omissions are consent decisions, not gaps in the data, and we would rather publish a narrower record than a fuller one nobody agreed to.

What we publish elsewhere follows the same rule. Municipal disposal rules carry the city source and the date we last checked them. Cost guides show how a number is built rather than asserting one. Where we have no data for a city, the page says we have no data for that city.

If you find something on this page that is wrong, tell us and we will correct it and say that we did. That is a low bar. It is also, in this category, an unusual one.