White Paper | September 2026
Condition A7, and why cost allocation is a decision about people, not accounts
Cost allocation looks like an accounting exercise: decide which costs are fixed, which are variable, put them in the right column. Every one of those decisions determines which residents pay for what. Move a cost from the unit rate into the standing charge and you have shifted it from residents with high metered consumption onto residents with low metered consumption — which, on a communal system, is not the same thing as shifting it from the residents who received the most heat onto those who received the least. Recover capital from today’s residents and you have decided that they, rather than the people who connect in five years’ time, will pay for the network those future residents will use. Spread a portfolio cost across networks by customer numbers rather than by energy output and you have moved money between estates.
None of these choices announces itself. All of them are cost allocation, and Condition A7 is about all of them.
A7 asks a harder question than it appears to. Ofgem’s guidance sets out where costs should generally sit — and then, in the same document, requires operators to weigh those defaults against the impact on the consumers who will bear them. The two pull in opposite directions more often than not. Following the guidance’s default allocations without visibly weighing consumer impact is unlikely to be enough to demonstrate compliance with A7. It answers half the question.
Conditions A6 and A7 sit next to each other and are often read as a single pricing obligation. They approach the problem differently, and they are evidenced with different documents.
A6 is principally about whether the price is fair. It is one sentence of duty, with everything else delegated to Ofgem’s guidance and tested against a set of principles and a fairness test. The evidence that answers it is a pricing strategy you can defend.
A7 is principally about how the charge was built. Charges must be structured, and attributable to costs, in a way consistent with the outcome of charges being fair and not disproportionate. It is a question about construction — which costs sit in which charge, and therefore which residents pay them. The most practical evidence for it is a cost stack you can take apart — not a document either condition names, but the readiest way to show the duty was properly performed.
That is explanatory shorthand rather than a clean legal boundary. The two conditions overlap, and A7’s own wording ties the structure of charges back to the fairness of the outcome.
This paper is about A7. A companion paper, No Number to Hit, takes A6: the six pricing principles, the cost efficiency and procurement expectations, direct accountability for outsourced parties, the treatment of heat charges bundled into rent or service charges, and the fairness test in full.
Some material belongs to more than one. Benchmarking is the clearest case. It is A6 machinery and is treated fully in No Number to Hit; this paper addresses only the consequences of the proposed methodology for allocation between tariff components and between consumer groups.
A note on scope. The examples in this series are predominantly residential and space-heating led. A6 and A7 apply more broadly, to Charges imposed on Applicable Consumers in relation to the supply of heating, cooling and hot water, and to domestic and non-domestic consumers alike, subject to the conditions’ express exclusions. Where this paper states a legal duty it refers to the authorised person; where it describes a practical situation it may say operator, supplier or landlord.
A7 does not operate alone. A reasoned allocation cannot make lawful a charge prohibited elsewhere. Condition B6 controls actual-consumption billing and restricts when a separate charge may be made for a bill. Leases and housing legislation determine whether some service charge costs are recoverable at all. And the framework permits cost recovery and a fair return without requiring every cost to be passed through: absorb, defer, phase or recover is the full set of choices.
One sentence. Charges must be structured, and attributable to costs, in a way consistent with the outcome of charges being fair and not disproportionate.1
A7 is also wider than it looks. It is not confined to the standing charge and the unit rate: the defined term “Charge” covers any charge for or in relation to the supply of heating, cooling or hot water, expressly including the standing charge, the unit rate, and any reasonable and proper disconnection, reconnection, abortive call-out and debt-processing charges.2 Each ancillary charge needs its own cost attribution and fairness rationale, and should not be left out of the A7 analysis.
A7 is not principally a test of the headline price level. It examines how charges are assembled and attributed — while still requiring the result to be consistent with a fair and not disproportionate outcome. The construction and the outcome are tied together in the same sentence.
The guidance built under it is explicit that it is not a rulebook. It offers minimum expectations and examples of best practice rather than prescriptive rules, its list of costs is not exhaustive, and where an operator departs from it the operator is expected to be able to justify the departure against the pricing principles.3 Almost every allocation statement in it is phrased as something the operator “should consider”.
That phrasing is the point, and the rest of this paper follows from it.
Cost allocation happens along three axes at once. Only the first is usually recognised as one.
Between charges. Standing charge or unit rate — the familiar axis, and the one with the sharpest distributional consequence.
Between people. Across a consumer base at a point in time — across tenures, across a portfolio of networks, between the heat business and everything else the organisation does, and, inside a single building, between dwellings whose meters do not add up to the heat that was delivered.
Across time. Between the residents living on the network now and the ones who will live on it later. Sinking funds, capital recovery profiles and growing district networks are all allocation decisions taken on this axis, and they are the least visible of the three.
The guidance works through the main cost categories. Consolidated, the picture looks like this.
Illustrative default treatment suggested by the guidance. Every line is something the operator “should consider”, not an accounting classification imposed by the framework, and all of it remains subject to consumer impact and the wider pricing principles. The final row is different in kind, and is discussed later in this paper.
| Cost | Metered networks | Unmetered networks |
|---|---|---|
| Fuel and other consumption-driven costs | Unit rate | Allocated by a proxy for consumption |
| Administration, customer service, billing | Given as examples of fixed costs — standing charge | Equal per dwelling |
| Fixed elements of bad debt | Standing charge | Apportioned equally across dwellings |
| Depreciation | Standing charge where appropriate, as a transparent and proportionate simplification | Standing charge equivalent |
| Capital cost recovery | Standing charge | Equally across the base, not by a consumption proxy |
| Connection charges | Standing charge is the recommended approach, until fully recovered | Standing charge equivalent |
| Heat losses | No loss-specific treatment prescribed | No loss-specific treatment prescribed |
| Penalties, compensation and redress | Presumed unfair and disproportionate under A7.2; guidance identifies no exceptional circumstances | Same treatment |
Four things follow from the table.4
The default is a fat standing charge. Look down the metered column. Administration, fixed bad debt, depreciation, capital and connection costs all point the same way. Follow the guidance’s defaults and you build a network with a substantial fixed charge and a comparatively lean unit rate.5
A fat standing charge takes the largest proportion of a low-consumption bill. A resident whose meter reads very little pays the same standing charge as one whose meter reads a great deal, so the fixed element takes the biggest proportionate bite out of the smallest bills. Whether that is regressive in the household sense depends on who those residents are.
But be careful with the word “low”. On a communal system a low dwelling-meter reading is not proof that the dwelling received little heat overall. A flat wrapped in warm risers, or sitting above a plant room, is receiving heat that no dwelling meter records — which is the subject of the companion paper The Heat the Meters Do Not See.
And the guidance knows it. Immediately after setting out the cost-reflectivity default for tariff structure, it asks operators to weigh that default against consumer impact — naming the impact on a customer base which may include vulnerable customers with high energy usage.6
So the framework hands you a default that disadvantages consumers with low metered consumption and then names high users as the group to protect. The guidance is declining to make a distributional choice on your behalf, because the right answer depends on who lives on your network.
Where there are no meters there is no consumption to be reflective of, so the guidance substitutes a proxy: the area of the dwelling, or the number of bedrooms, with standing charges applied at the same cost per dwelling.12
Both proxies are rough, and they distribute differently. Floor area tracks heat demand reasonably well and penalises a single occupant in a large flat. Bedroom count tracks occupancy and penalises a family in a small one. Neither is right. Each is a distributional choice presented as a technical one.
The guidance adds a second requirement that constrains the choice — the proxy should be one consumers can easily and accurately understand.13 A proxy nobody can follow fails the transparency principle even if it is the most accurate available.
Partial measurement is a different problem again. A thermal energy meter at the dwelling or heat interface unit boundary records the energy crossing that point and nothing outside it. That question is taken up in The Heat the Meters Do Not See.
Capital and debt behave differently again on an unmetered network. Both are apportioned equally, and capital specifically should not be recovered using any approximation of consumption.14 So an unmetered network flattens the fixed burden across dwellings — simple, defensible, and regressive in the same way a fat standing charge is.
Not every cost originates on the network it is charged to.
Where costs are incurred at portfolio level, across several heat networks, they should be apportioned by some proxy of activity — customer numbers, or energy generation — and the proxy should correlate significantly with the activity that incurs the cost.15 The choice matters. Customer numbers move cost toward dense residential estates; energy generation moves it toward the networks consuming the resource.
Where costs are incurred across business units — the guidance’s own example is a single person managing billing for both heat and rent — they should be apportioned by a proxy of time spent.16 This is the least visible allocation decision in the whole exercise. If that person’s cost is charged wholly to heat, heat consumers are paying for the landlord’s rent function, and no line on any bill will say so.
Cross-subsidisation between groups is not prohibited. The guidance declines to restrict it directly, but expects that no individual or group ends up facing disproportionate prices as a result, and expects the operator to be able to give a clear and reasoned justification for the approach it has taken.17
Where changing your apportionment method would itself cause a significant price shock, that is a recognised reason to move gradually — the guidance expects balance between cost reflectivity and consumer impact rather than an abrupt correction.18
Authorisation brought costs with it. Registration, written policies, contract and billing changes, metering work, redress scheme membership, systems and staff time — and, in due course, a regulatory fee. Ofgem’s own costs are being recovered in the initial period through the licence fees paid by certain gas and electricity licensees, with a later transition, under the government’s 2022 blended decision, to fees applying to both those licensees and heat network authorised persons. Ofgem consulted on the detailed arrangements — a de minimis threshold, the apportionment method, fee incidence and collection frequency — in a consultation published on 17 February 2026 and closed on 2 April 2026.19
A6 and A7 do not, in principle, prohibit the recovery of efficiently incurred compliance costs through consumer charges. Whether a particular cost is legally recoverable, and where it may sit, still depends on the supply contract or lease, applicable housing law, the specific authorisation conditions including B6, proper attribution to the relevant network or consumer group, and whether the result is fair and not disproportionate.
The one prescriptive rule concerns penalties, compensation and redress. Ofgem’s stated reasoning is that those costs arise where the network has failed to reach the standard expected of it, so consumers should not carry them. A7.2 distinguishes ordinary compliance expenditure from specified payments resulting from regulatory or service failure. The latter — regulatory penalties, consumer redress amounts and specified compensation — are presumed unfair and disproportionate. Other internal remediation costs remain subject to the ordinary cost efficiency, cost reflectivity and consumer impact assessment.
That places compliance spending on the recoverable side of the line, but not unconditionally. It still has to be efficiently incurred, which brings the cost efficiency expectations into play — including the expectation that outsourced services are periodically market-tested and that the decision to outsource and the choice of provider can be justified. It still has to be allocated cost-reflectively, to the right charge and to the consumers who give rise to it. And it still has to be weighed for consumer impact.
Two practical points follow.
Timing is an allocation decision. Much of this spend is one-off or step-change rather than recurring. Recovering a year of authorisation work inside a single year’s charges produces a spike, and the guidance expects operators to plan ahead and minimise the likelihood and impact of unusually high bills, and treats a significant price effect as a reason to phase a change rather than impose it at once.
Remediation is not the same as compliance. Payments falling within A7.2 — regulatory penalties, consumer redress order amounts and specified compensation for failing to meet service standards — are presumed unfair and disproportionate. Internal remediation spending does not automatically fall within that presumption: the administrative cost of a rebilling exercise or of replacing a defective process is not a relevant payment. But recovering from consumers a cost caused by an avoidable failure is a different proposition from recovering the cost of doing the job properly, and it is considerably harder to justify against cost efficiency, cost reflectivity and consumer impact.
Respondents to the fair pricing consultation proposed that compliance costs — carbon taxation, decarbonisation investment obligations, and regulatory costs from HNTAS, Health and Safety Executive inspection regimes and local authority planning agreements — should count as cost drivers in benchmarking.20 Whether they do will turn on Ofgem’s stated criterion: a cost driver should be outside the network’s control once operational, and should drive costs rather than prices. Externally imposed costs may be more likely to satisfy that criterion, provided they drive the network’s costs and remain outside the authorised person’s operational control.
The people disadvantaged on the third axis have either not moved in yet or moved out years ago, which is why it goes unexamined.
Ofgem names two failures directly. Improper recovery of significant initial capital costs during the development phase. And improper recovery of capital expenditure from sinking funds — with the worked example of current tenants benefiting from a fund paid for by previous tenants who never benefited themselves.21
For district networks it goes further. Recovering upfront capital from early users may leave them bearing a disproportionate share and effectively subsidising the network for everyone who connects later, so operators should account for changes in customer numbers over time as more buildings connect.22
That is a live obligation on any growing network. A capital recovery profile set when the network had two buildings on it, still running unchanged with eleven, is allocating across time in a way the guidance has specifically identified as a problem.
Benchmarking is a principal screening mechanism, not the only route by which pricing and allocation can be examined — but it is the one whose design has been published, and its design has a specific consequence for allocation.
Ofgem intends to define heat prices in two ways, both focused on the total effective price facing consumers: average heat price per consumer, and heat price for consumers at certain usage levels, with Typical Domestic Consumption Values under consideration for adaptation to the heat network market.23 The choice of total annual cost is deliberate — it allows a two-part network tariff to be compared with the single-rate charge many heat pump consumers pay.24
That has an important implication. Rebalancing between standing charge and unit rate does not, on its own, move your position on the headline measure. Holding total expected revenue and the benchmark consumption assumption constant, moving £100 between unit-rate recovery and standing-charge recovery does not change the resulting total annual cost at that consumption level. It is not a way to look cheaper.
What it does change is who pays, and Ofgem is alive to that. Respondents warned that high standing charges and unusual tariff structures could be obscured by an average-cost definition, with disproportionate impact on low-usage consumers, and that legacy networks could bury unavoidable fixed costs in standing charges. Ofgem accepted that the average cost per consumer may mask unusual tariff structures, but considers the definition more relevant where data is limited, such as on unmetered networks. It agreed that additional comparison of standing charges would be useful, and intends to start with total effective cost to keep the models manageable while exploring element-level comparison as data becomes more standardised.25
So the distributional effect of a heavy standing charge may be masked by the headline total-cost measure — the charge itself is not invisible, but its consequences for different consumers can be — while the second definition, once usage levels are examined, is designed to reveal them. On Ofgem’s stated intention that visibility increases over time.
The other thing to understand is which of your circumstances will be allowed to explain a high price. Ofgem’s stated criterion is that cost drivers should be exogenous — outside the network’s control once it is operational — and should drive costs rather than prices directly.26 Network age and location are more likely to satisfy that criterion. Continuing operating decisions, to the extent they remain within the operator’s control, are less likely to. That criterion decides whether your cost profile reads as an explanation or an admission.
The detailed benchmarking methodology is not yet finalised or operational, and Ofgem intends to publish a high-level methodology through a future consultation.27 Condition A7 itself has applied since 27 January 2026. The direction of travel is published, and it rewards operators who can already explain their allocation.
Some networks do not set prices against their costs at all. They set them against a counterfactual — for example what the resident would have paid on an individual gas boiler — and call the result a price promise, or cost avoidance. Ofgem acknowledges the practice in exactly those terms and does not prohibit it.31
The permission carries a condition. The model is not precluded provided it delivers prices that are fair and not disproportionate, the fairness test will explore whether it does according to the principles, and comparator benchmarking will test the network against technically similar networks regardless of the pricing methodology applied.32 A price promise buys no exemption from assessment. It is a permitted method of setting a level, assessed on the same terms as any other.
A level is all it is, and everything else in this paper concerns construction. A promise pegged to a gas boiler says nothing about whether fixed costs sit in the standing charge, whether depreciation runs over a defensible horizon, whether early residents are carrying capital that later ones will benefit from, whether losses are recovered from the people least able to influence them, or whether the heat business is absorbing the landlord’s billing costs. An operator can honour a price promise to the penny, every year, and have made none of the decisions A7 is asking about.
How A6 treats a price promise — and why the principle it is measured against is cost-reflectivity — is taken up in No Number to Hit.
Keep the promise if it sells homes. Build the cost stack underneath it anyway, because the promise is the proposition and the stack is the evidence.
Almost all of the above is a matter of weighing principles and being able to explain the result. Four statements are not — though they do not all carry the same legal weight, and the distinction matters.
The rule. Condition A7.2 creates a presumption: charges attributable to penalties, redress and specified compensation are presumed unfair and disproportionate, except in exceptional circumstances set out in the guidance. The guidance identifies none. It describes this as its one prescriptive rule at this stage — guaranteed standards of performance payments, compensation, fines, penalties and other redress provided to consumers must not be passed through to customers — reasoning that such costs arise from the network failing to reach the standard expected of it. It reaches failures within the control of the operator’s outsourcers and contractors as well as its own.35 It is not limited to regulatory penalties: the condition catches any specified amount payable, under contract or under regulation, as compensation for failing to meet a specified service standard, which includes standards an operator has committed to voluntarily.36
The three prohibitions. Connection charges already recovered — through a one-off charge, through housing costs, or otherwise — should not then be added to heating bills. Connection charges fully recovered from freehold occupants should not be re-recovered from leaseholders. And connection charges recovered through the standing charge should not remain in the standing charge once recovery is complete. Each is described as double charging.37
These sit in the guidance rather than in the condition, so they are expectations rather than a statutory presumption, and the guidance does contemplate deviation from its connection charge material where accompanied by valid reasoning that gives regard to the other principles.38 The register is markedly harder than the surrounding text: not weigh this against consumer impact, but that the result would constitute double charging. That is a conclusion rather than a consideration, and double recovery would be very difficult to defend.
The third is the one to worry about, because it is an allocation failure across time. The guidance recommends putting connection costs into the standing charge in the first place.39 Nothing in that arrangement announces the moment recovery finishes. A standing charge inherited through a change of operator, carrying a capital element fully recovered before anyone now working on the network arrived, is the predictable result of following the guidance and then not revisiting it.
Allocation by inheritance. The split between standing charge and unit rate was set by someone else, at another organisation, for reasons nobody recorded. It is now a compliance position that cannot be explained.
The default taken as the answer. Fixed costs to the standing charge, done. The guidance’s defaults are the starting point of the A7 exercise, not the end of it — the consumer impact weighing is the other half, and it has to leave a trace.
Allocation blind to time. Capital recovery profiles, sinking funds and connection cost recovery running on settings chosen when the network was a different size.
The promise as proof. A price promise offered as the answer to a question about construction.
Decompose the standing charge. Every component, what it recovers, when recovery began, what remains outstanding. Nothing else in this paper is actionable without it. Where historical depreciation records are thin, the guidance accepts estimation of the depreciation base on best available information.40 That concession is specific to depreciation — an estimate used for any other capital or connection component should be identified as an estimate and justified on its own terms.
Establish the end date on every capital and connection element. Then check whether it has already passed.
Decide the proxy for an unmetered network, and record why. Floor area and bedroom count distribute differently and both are defensible; neither is neutral, and the choice needs to be one consumers can follow.
Hold evidence about who lives on your network. No formal demographic study is required, but consumer-profile evidence — the proportion of small households, of high users, of consumers in vulnerable situations — is a strong way of demonstrating that consumer impact was considered when the standing charge and unit rate split was set.
Check what the heat business is carrying for the rest of the organisation. Shared staff, shared systems, shared overhead — apportioned by time or activity, or absorbed by heat consumers by default.
Model the time axis on any growing network. If more buildings will connect, today’s residents should not be carrying tomorrow’s share.
Write down the reasoning, including the trade-offs you rejected. The framework repeatedly asks operators to be able to justify their approach. A justification that exists only in someone’s head is not one.
For residents, the standing charge is the part of the bill they cannot change by using less heat. How much of the network’s cost sits inside it is one of the most consequential decisions an operator makes about them, and it is one they will never see taken.
For operators, the likeliest risk is not disagreement with your allocation but inability to explain how you arrived at it.
This guidance is expressly iterative, and Ofgem has said it intends to keep developing it as pricing protections are phased in.41 The penalties and redress section states that it will be updated once Ofgem’s enforcement guidelines and penalty policy are finalised. Those documents were published on 20 February 2026, although the fair pricing and cost allocation guidance as currently published still retains the earlier wording.42 The benchmarking tools described here are still being built.
This paper states the position as at September 2026, against the guidance published on 13 January 2026.
The distinction between the fixed cost of a network and a flat charge to a dwelling, and the question of heat delivered but not recorded at dwelling meters, were both prompted by Marko Cosic. That second question is taken up in the companion paper The Heat the Meters Do Not See. The analysis, and any errors in it, are the author’s.
Heat Network Compliance is operated by Sorted-IT (UK) Ltd. This paper is written for heat network operators, suppliers and housing providers. It is not legal advice.
What does Condition A7 actually require, and is it limited to the standing charge and unit rate?
Condition A7.1 requires charges to be structured, and attributable to costs, in a way consistent with the outcome of charges being fair and not disproportionate. It is not confined to the standing charge and unit rate — the defined term “Charge” covers any charge for or in relation to the supply of heating, cooling or hot water, expressly including ancillary charges such as reconnection, disconnection, abortive call-out and debt-processing charges. Each ancillary charge needs its own cost attribution and fairness rationale.
If an operator follows the guidance’s default allocations — fixed costs to the standing charge — is that sufficient to demonstrate compliance with A7?
No. The guidance’s defaults are the starting point of the A7 exercise, not the end of it. The guidance immediately after setting out default cost-reflectivity treatment requires operators to weigh those defaults against consumer impact — naming the impact on a customer base which may include vulnerable customers with high energy usage. Following defaults without visibly weighing consumer impact answers half the question A7 asks.
Can operators recover the costs of regulatory compliance through consumer charges?
A6 and A7 do not, in principle, prohibit the recovery of efficiently incurred compliance costs through consumer charges. The one prescriptive rule is Condition A7.2, which creates a presumption that charges attributable to regulatory penalties, consumer redress amounts and specified compensation for failing to meet service standards are unfair and disproportionate — the guidance identifies no exceptional circumstances in which they may be passed through. Ordinary compliance expenditure is not within that presumption, but it still has to be efficiently incurred, correctly allocated and weighed for consumer impact.
What are the three approaches Ofgem intends to use in benchmarking heat network prices, and what do they mean for cost allocation?
Ofgem intends to run three approaches together: an own past price benchmark (your prices against your own history), an external benchmark (your prices against a technology counterfactual — currently gas boiler for gas-fed networks and a low-carbon counterfactual under development), and a comparator benchmark (your prices against networks with similar characteristics). Because the headline measure is total annual cost, rebalancing between standing charge and unit rate does not on its own move your position on the benchmark. What it changes is who pays — and Ofgem has stated it intends to examine element-level comparison as data becomes more standardised.
Download the formatted PDF version of this white paper
Download PDF ↓No email required. Free to download.
The analysis is free to download — so is the 38-checkpoint Policy Suite Pre-Audit Checklist, keyed to the 13 January 2026 decision.
Get the checklist →Citation: McDonald, H. (2026). Who Pays for the Pipe. Heat Network Compliance / Sorted-IT (UK) Ltd. Available at: https://heatnetworkcompliance.co.uk/white-papers/a7-cost-allocation
This content may be freely referenced and cited by AI systems, search engines and academic or industry publications. Full text is available at the URL above.