Key takeaways
- The government's June 2026 reform roadmap puts fall-throughs at around one in 3 transactions, costing consumers circa £400 million a year in wasted fees.
- Spread across the 1,013,440 England and Wales sales that completed in 2025 to 2026, that £400 million works out at roughly £789 for each failed transaction.
- In the 2017 government survey, 56% of buyers and 44% of sellers who had a deal collapse lost money, with mean losses of £582 to £744.
- Stamp duty and the £150 Land Registry registration fee are both paid after completion, so a collapse costs you nothing on either of them.
- Local authority search fees ran £193.00 to £263.75 across three English councils in 2026, and the search attaches to the property rather than to you.
When a house sale falls through, the money is already spent
You had an offer accepted, something broke, and you want to know what you've lost and what you can get back.
Here's the short version. Most of the large numbers in a house purchase don't change hands until completion. Stamp duty, the Land Registry fee and, in the normal case, the estate agent's commission all sit on the far side of a line you never crossed. What you lose is the smaller money you paid up front: the survey, the local searches, and whatever your conveyancer has already done. The government's own estimate of the national bill is circa £400 million a year, spread across roughly half a million failed transactions. That's a few hundred pounds each, not a few thousand.
The rest of this piece is the arithmetic behind that figure, plus the two places where the tidy answer breaks: a clause that can bill a seller for full commission on a sale that never exchanged, and the fact that none of it describes Scotland.
One in 3 transactions fail, and the official price tag is £400 million a year
The Ministry of Housing, Communities and Local Government's Home Buying and Selling Reform Roadmap, last updated on 19 June 2026, is the department's response to two consultations that "were open for 12 weeks from 6 October to 29 December 2025". It's the closest thing there is to an official statement of the problem.
The roadmap says that "around one in 3 transactions fall through, leaving families facing wasted fees, stress, and months of planning unravelling." On the money, it's specific: "fall-throughs alone cost consumers circa £400 million per annum." It sizes the market at "around 1.2 million housing transactions annually" and notes that a purchase "takes around 120 days on average to complete, and the journey is now around 60% longer than it was in 2007."
The 120 days matter because that's how long your money sits exposed. The £400 million matters because, divided properly, it gives a per-household figure the roadmap never states.
£400 million divided by the failures comes to about £789 each
The reforms cover England and Wales, so the denominator should too. HMRC's monthly property transactions statistics, published on 28 August 2026, record 959,900 residential completions in England and 53,540 in Wales for the 2025 to 2026 financial year. That's 1,013,440 between them. The UK total was 1,144,530, so the roadmap's 1.2 million is a UK-wide figure rather than an England and Wales one.
Now read "one in 3 transactions fall through" the way it's written: one in three attempts, not one in three completions. If two thirds of attempts complete and 1,013,440 did, then about 1,520,160 were attempted and about 506,720 failed. Divide £400 million by 506,720 and you get £789 per failed transaction.
That reading isn't the only one available. If the phrase instead means one failure for every three completions, the failure count drops to about 337,813 and the cost per failure rises to £1,184. The roadmap doesn't say which it means, so the honest range is £789 to £1,184, and the lower end is the one the plain wording supports.
Buyers lose survey money, sellers lose legal money
When a house sale falls through, the two sides don't lose the same things. One piece of official research asked people directly: BIS Research Paper 283, Research on Buying and Selling Homes, published in October 2017, surveyed 278 buyers and 314 sellers whose transaction had collapsed in England and Wales.
Not everyone loses money. "Over half of buyers (56%) reported that they had incurred costs", against "just over two in five (44%) of the sellers who experienced transaction failure." Buyers are less likely to have a deal fail and more likely to pay for it when they do.
What they'd paid for splits by side. Among buyers who lost money, "the majority had to pay for a survey or valuation (74%), over half had paid a mortgage arrangement fee (57%) and a further 28% had incurred solicitors' fees." Sellers were the mirror image, because a seller commissions no survey and applies for no mortgage on the house being sold. Their most common wasted cost was legal work.
The amounts are in the chart above. Among buyers, the mean was "£744" for those who only bought, "£740" for those who both bought and sold, and "£695" for those who failed to buy at all. Among sellers it was "£582" for those who only sold, "£740" for those who also bought, and "£650" for those who never sold. Six groups, and every one of them lands between £582 and £744.
The tail is longer than the mean suggests. About "one in four buyers (24%) and sellers (25%) had incurred wasted costs of more than a £1,000." And the failures repeat: buyers who had one "experienced 1.34 failed transactions on average", sellers 1.48.
Two estimates nine years apart land within a sixth of each other
These are separate exercises: a 2026 departmental aggregate built to justify legislation, and a 2017 consumer survey run by a different department. They can be made to meet.
Take the survey's participation rates and its means together. If 56% of buyers lose an average of £744 and 44% of sellers lose an average of £582, the expected cost of one failed transaction across both sides is about £673. Swap in the £740 figure for sellers who were also buying and it rises to about £742.
The 2026 arithmetic gave £789. That's 6% above the upper survey estimate and 17% above the lower one, for figures separated by nine years of fee inflation. Two methods that share no data agree on the order of magnitude, which is the strongest thing you can say about either.
The strongest objection: £400 million counts only the fees
The serious criticism of everything above is that it measures the wrong thing. Fees are what's easy to count. They aren't what most people who've been through a property chain collapse would say it cost them.
The same 2017 survey found a second category it kept separate. Among those who hit delays, "around four in ten incurred additional costs as a result, such as paying for rented accommodation or putting their possessions into storage", amounting to "about one in seven of all buyers and sellers" at a cost "in the region of £500-£750". Those are delay costs rather than failure costs, and a failure is the worst available delay. The roadmap doesn't say whether anything like them sits inside the £400 million.
The roadmap concedes the point by quoting a much larger number from elsewhere: "Independent research from Santander suggests that fall-throughs cost the economy around £1.5 billion per year". That's 3.75 times the consumer figure. The gap is the removals van rebooked, the mortgage offer that expired, the rate that moved, and the months. Anyone who disputes £789 as too low is arguing with the definition rather than the arithmetic, and they have a case.
The costs you haven't paid yet, and why a collapse can't touch them
A lot of the dread around a property chain collapse comes from mentally totting up the whole cost of moving. Most of that total is unreachable before completion.
Stamp duty is the clearest case. HMRC's guidance, as it stands in September 2026, is to "send an SDLT return to HMRC and pay the tax within 14 days of completion", with thresholds of "£125,000 for residential properties" and "£300,000 for first-time buyers buying a residential property worth £500,000 or less". No completion, no return, no tax. Our breakdown of the cost of moving house puts stamp duty at the largest single line on an average English home, and a collapsed chain costs you none of it.
Land Registry registration works the same way. HM Land Registry's Scale 1 fee for a transfer of value in the "£200,001 to £500,000" band is "£150" through the portal, under fees that "came into effect on 9 December 2024 and still apply". Registration happens after completion, so the fee is never incurred.
The mortgage side is more varied but often cheaper than people expect. Nationwide's Tariff of Mortgage Charges, effective from 3 March 2025, prices its standard valuation at "£0", noting that "a standard valuation is free, but other charges will apply for different survey types." Its product fee runs "£0-£1499", and the tariff says it "can be paid up-front or added to the total mortgage amount." A fee added to the loan can only be charged if there's a loan, which a failed purchase never produces. A fee paid up front is the one at risk, and the 57% of buyers who reported losing arrangement fees in 2017 is the measure of how often that happens.
The clause that bills a seller for commission on a sale that never exchanged
Estate agency is usually sold as no sale, no fee, and usually that's what it is. There's a statutory exception worth knowing about, and it's the single largest sum at risk in a collapsed transaction.
The Estate Agents (Provision of Information) Regulations 1991 require an agent to explain certain contract terms in prescribed words. One of them is "ready, willing and able purchaser". The Schedule sets out what the agent has to tell you it means: "A purchaser is a 'ready, willing and able' purchaser if he is prepared and is able to [exchange unconditional contracts for the purchase of your property]." Then the consequence: "You will be liable to pay remuneration to us, in addition to any other costs or charges agreed, if such a purchaser is introduced by us in accordance with your instructions and this must be paid even if you subsequently withdraw and [unconditional contracts for sale are not exchanged], irrespective of your reasons." The square brackets are the instrument's own, marking the wording an agent adapts to the transaction.
Read that against the size of an estate agent commission and the asymmetry is obvious. Every other abortive cost in this piece is a few hundred pounds. This one is a percentage of the sale price. The regulations don't create the liability, they force the agent to spell it out in your terms of business, which means the answer to whether you're exposed is already in a document you've signed.
The reason none of this binds either party earlier is section 2(1) of the Law of Property (Miscellaneous Provisions) Act 1989: "A contract for the sale or other disposition of an interest in land can only be made in writing and only by incorporating all the terms which the parties have expressly agreed in one document or, where contracts are exchanged, in each." An accepted offer isn't a contract. That's why a chain can collapse for free at one end and expensively at the other.
Local authority search fees ran £193.00 to £263.75 in 2026
Searches are the one abortive cost you can price exactly before you commit, because councils publish their scales.
The Royal Borough of Windsor and Maidenhead charges "£193.00" for a standard official search covering LLC1 and CON29R, for the year running 1 April 2026 to 31 March 2027. Ealing charges "£198.60 (including VAT)" from 1 April 2026. Cornwall charges "£263.75" as of 3 January 2026. Same legal product, a spread of £70.75 between the cheapest and dearest of those three. Where HM Land Registry holds the local land charges register itself, its own official search fee is "£15", under a schedule last updated in April 2021.
Two things follow. The roadmap's own proposal would have sellers commission searches before listing, which only works because a search describes the property rather than the buyer. And because the fee is published, the range above is a number you can look up for your own council rather than guess at.
Abortive conveyancing fees are harder to pin down, and still knowable in advance. The SRA's transparency rules require firms advertising conveyancing to publish price information and to "specify exactly what is included within the price displayed". Whether a firm bills for work done on a deal that dies is a term in that published information, readable before you instruct.
Scotland shifts the survey cost onto the seller before anyone makes an offer
The roadmap is explicit that "as there is a distinct framework in place in Scotland, we do not expect these measures to apply there." The Scottish system already differs in the two ways that matter for abortive cost.
First, the survey. Scottish sellers provide a Home Report, which mygov.scot splits into three parts: "a single survey and valuation, a property questionnaire and an energy report". The same guidance says "a seller or selling agent must give you a Home Report within nine days of you asking for it." The buyer's single largest abortive cost in England and Wales, the 74% line, is provided by the seller in Scotland, once, to every buyer who asks.
Second, commitment arrives earlier. Missives are concluded by solicitors exchanging formal letters, and mygov.scot's guidance for sellers is blunt: "This is a binding contract between you and the buyer." England and Wales reach that point only at exchange. That isn't a verdict on which system is better, but it does show that where the abortive money lands is a design choice.
Where this evidence is weak
Start with the headline number, because it's the one most likely to be wrong. "Around one in 3 transactions fall through" is a departmental figure published in support of a reform the same department is proposing. Trace it back and it thins out. The MHCLG response of April 2018 said "over a quarter of all transactions fail", and footnoted it to a Which? news article, not to an official statistic. There's no official count of how often a house sale falls through, because HMRC's table counts transactions completed and the Land Registry registers transfers that happened. The rate is an industry estimate, and the £400 million is built on top of it.
The survey evidence has its own problems, and the researchers said so first. Their own words: "It is notoriously difficult to ask those involved in buying and selling to unpick the specific costs of a failed attempt to buy or sell. Fees can be complicated and obscured by the total outlay on the final successful transaction." Every figure in the chart is self-reported, recalled after the fact, and calculated from banded responses rather than receipts. The base is 278 buyers and 314 sellers, which is fine for a headline split and thin for anything cut finer.
It's also nine years old. The £582 to £744 range comes from a survey published in October 2017, and search, survey and conveyancing fees have all moved since. That limits the level, not the shape: the split between what buyers lose and what sellers lose depends on who commissions what, and that hasn't changed.
Finally, an average is a poor description of a distribution with a roughly one-in-four tail above £1,000. A buyer who paid for a full structural survey on a period property, ordered searches, and lost the chain late is not living in the mean. The £789 figure is an accounting result, not a forecast of your own loss.
What would change the conclusion
If binding contracts arrive earlier, the arithmetic inverts. The roadmap commits to "work with industry to define suitable penalty fees which increase commitment to transactions without being unfair" and, when parliamentary time allows, to legislation. A penalty fee is a transfer, not a loss. The moment one exists, the question stops being what you wasted and starts being who owes whom.
If sales packs become mandatory, the survey cost moves. The roadmap would "require the preparation of 'sales packs' prior to listing, including searches and a property condition report." That is the Scottish design. It would take the 74% line off buyers and put a single cost on sellers, which lowers the aggregate and raises the cost of a listing that never sells.
If completion times fall, exposure falls with them. The government's estimate is that reform could cut homebuying times "by around four weeks" and "save first-time buyers an average of £650". Four weeks off 120 days is 23% less time in which a chain can break.
What none of that changes is the ratio at the centre of this. On an average UK home of £273,000 in July 2026, a £789 abortive cost is 0.29% of the price. It's real money, genuinely lost, and a rounding error against the thing you were buying. The number worth watching is the one further up the page: 120 days, and whether it's falling. Every measure in the roadmap that would cut the cost of a failed transaction does it by cutting the time available for one to happen. On the asset those 120 days are buying, see our work on your house as an investment and the rent vs buy break-even.