Methodology

Current status of the calculation engine

The figures currently returned by our calculators are illustrative. They are produced by a placeholder model so that the interface, the presentation of ranges and the explanation of sensitivities can be designed and reviewed before a verified engine is connected.

Illustrative output is labelled as such wherever it appears. When the verified engine goes live, each result page will state the model version and the date its inputs were last refreshed.

Baseline household electricity: Ofgem Typical Domestic Consumption Values

Where you do not give us your own annual consumption, base household electricity comes from Ofgem's Typical Domestic Consumption Values for standard single-rate electricity, effective 2026-07-01: low use 1,600 kWh a year, medium use 2,500 kWh a year, high use 3,800 kWh a year.

TDCVs are standard reference values used for comparison when a household's actual consumption is unknown. Ofgem does not publish them as predictions for an individual home, and we do not present them as an average household.

We select a band from household size only: one person uses the low-use benchmark, two or three people the medium-use benchmark, four or more the high-use benchmark. If household size is missing we use the medium-use benchmark and say so. Property type and daytime occupancy do not change base household consumption — they are insufficient evidence for an annual-kilowatt-hour multiplier.

Where we show a range around the benchmark, the ±25% band is a WattFirst modelling assumption, not an Ofgem figure. It exists because a benchmark is not a household prediction.

EV charging and heat-pump electricity are never assumed to be inside the benchmark. They are added separately, and both allowances remain uncalibrated prototypes. Demand for technology you are only considering is held separately from your current total.

Your own annual consumption remains substantially better than any benchmark, because it reflects your actual occupancy, appliances, heating and charging rather than a national reference value. Entering it always overrides the benchmark.

This is a partial calibration of the baseline consumption model only. The tariff, solar, battery and cost models remain prototypes, and the calculator as a whole is not verified.

Source: Ofgem — Review of typical domestic consumption values. Last calibrated 2026-08-06.

Electricity tariff data

WattFirst holds a versioned tariff catalogue (uk-tariffs-v1, calibrated 2026-08-06). In this phase it contains one published official reference and whatever rates you enter yourself. It is not a whole-market comparison, it does not list suppliers, and there is no commercial relationship with any supplier.

The reference is Ofgem's default-tariff cap for a standard single-rate variable tariff paid by Direct Debit: 26.11p per kWh plus 57.19p per day, VAT included at 5%, a national average across England, Scotland and Wales, valid from 2026-07-01 to 2026-09-30.

These are capped national average default rates. They are not the cheapest tariff, not an average of all tariffs on the market, and not a recommendation. Actual rates vary by region, payment method, meter type and supplier, so your own rates always take precedence.

Validity is explicit. Once a reference passes its end date it is marked “Expired reference — update required”, it can no longer be selected, and any saved selection is shown as expired rather than treated as current. WattFirst does not guess future rates.

A broad tariff description — “I'm on a cheaper overnight tariff” — is not the same thing as usable rates. A description is kept as context and never produces a monetary saving on its own; only rates you enter or a current sourced snapshot do.

Single-rate and time-of-use tariffs are represented separately. A single-rate tariff has no peak/off-peak spread, so tariff arbitrage and battery tariff-shifting value are exactly zero rather than assumed. A time-of-use tariff needs both rates before any spread-based value is calculated.

The standing charge is stored and displayed, but it is never treated as a hardware saving and is excluded from comparison savings until a whole-bill comparison exists. Most hardware decisions do not change the standing charge.

Export rates are never inferred from an import tariff. The Ofgem reference contains no export rate, so exported solar energy is shown in kWh with its monetary value marked unavailable until you supply an export rate.

Supplier coverage added in future will not equal whole-market coverage unless that is stated explicitly.

Tariff data last reviewed: 6 August 2026.

Dynamic tariffs are recorded but not modelled. Half-hourly prices and a half-hourly consumption profile are required before any optimisation, arbitrage or comparative saving figure could be honest, so none is produced.

Source: Ofgem — Energy price cap unit rates and standing charges. Last calibrated 2026-08-06.

Solar generation: PVGIS location estimates

Annual solar generation is calculated by the European Commission Joint Research Centre's PVGIS service (uk-solar-generation-pvgis-v1, introduced 2026-08-06) for the location, orientation, pitch and candidate array size implied by your answers. There is no universal yield figure anywhere in the model: if PVGIS cannot be reached, or you have not given an outward postcode, generation is simply not calculated and no substitute number is shown.

We ask only for the first part of your postcode. It is resolved to the approximate centre of that outward code using Postcodes.io open data. We never ask for, store or transmit a full postcode or an address, and the location sent to PVGIS is an area centroid rather than your roof.

PVGIS is queried for a fixed, building-mounted crystalline-silicon system with 14% system losses and PVGIS terrain-horizon shading enabled. Roof direction is mapped transparently to the PVGIS aspect convention, where 0° is south, ±90° is east or west and ±180° is north. "South plus east or west" and "east and west" are single-plane representative aspects, not a modelled multi-plane array. Where you have not told us your roof direction, the result is labelled as a south-facing location reference for your area, not a property-specific estimate.

Roof pitch is optional. Where you do not supply it, 35° is assumed and the result says so on the page. That assumption is ours, not a measurement of your roof and not a PVGIS output.

PVGIS models satellite-derived irradiance and terrain horizon. It does not know about your trees, chimneys, dormers, neighbouring buildings, panel layout, inverter clipping or long-term degradation. A PVGIS estimate is not an MCS quotation, an MCS certificate or an installation specification. MCS MIS 3002 is cited as UK methodology context only; WattFirst claims no MCS compliance or accreditation.

Calibration stops at generation. Array size, self-consumption share, installed cost, export value and payback all remain WattFirst prototypes, and each is labelled with its own status on the result. Running a prototype capacity through PVGIS does not make that capacity calibrated, and a calibrated generation figure does not make the resulting financial figures verified.

WattFirst has no commercial relationship with the Joint Research Centre, with PVGIS or with Postcodes.io. Estimates are requested only when you ask for a result, cached for unchanged answers, and re-requested whenever any input that changes the answer changes.

Source: European Commission Joint Research Centre — PVGIS non-interactive service. Last calibrated 2026-08-06.

Solar self-consumption methodology

PVGIS currently calibrates solar generation only. It estimates how much electricity a proposed array could produce at your location, orientation and pitch. It knows nothing about when your household actually uses electricity.

WattFirst therefore does not currently calculate solar self-consumption. How much generation is used in the home, how much is exported, the bill saving from solar, export income, grid independence and solar payback are all shown as not calculated. They are not shown as zero, and no default or typical percentage is substituted.

MCS MGD 003, the MCS solar PV self-consumption guidance, is being evaluated as a possible methodological reference. Its lookup tables have not been reproduced, transcribed, embedded, interpolated or approximated anywhere in this site or its source code, and WattFirst does not have confirmed permission to use them commercially.

WattFirst does not claim MCS compliance, MCS approval or MCS equivalence, and no figure here is calculated in accordance with MGD 003.

Self-consumption is implemented behind a typed provider boundary. The provider active today is an explicitly unresolved one: it accepts the household inputs, returns no figures, and names the missing requirement — an approved or independently calibrated self-consumption method, or suitable half-hourly interval data for the household.

When a method is connected, it will identify its source, its version, the permissions under which it is used, its assumptions and its limitations on this page and on every result that depends on it. Until then, a broad daytime-occupancy answer is retained as structured context only and changes no financial figure.

Source: MCS — Solar PV self-consumption guidance (MGD 003). Last calibrated never — no MGD 003 data is used by WattFirst.

Electric-car electricity demand methodology

An electric car's effect on your electricity bill is not one number. WattFirst keeps six quantities apart and reports each with its own provenance: how far the car travels, how efficiently it uses electricity, the electricity consumed by the vehicle, the electricity drawn from your home, the electricity bought away from home, and — for a car you are only considering — the prospective demand it would add.

Distance is calibrated. Where you do not give your own mileage, WattFirst uses the Department for Transport's National Travel Survey figure for battery-electric cars: 8,900 miles a year, England, 2024, table NTS0901a, published 27 August 2025 under the Open Government Licence v3.0. It is shown as a national average, never as your mileage. The DfT publishes its own caveats: annual mileage is reported by the respondent rather than measured, the figure is weighted to the population of England and subject to sampling error, no battery-electric-specific sample size is published, and the 2023 estimate was withheld because the sample was too small to be reliable.

Vehicle efficiency is not estimated. WattFirst has no approved real-world efficiency source for the UK battery-electric fleet, and manufacturer WLTP figures are not real-world consumption. If you do not supply your car's miles per kWh or kWh per 100 miles, the electricity the vehicle consumes is not calculated at all. No default efficiency is applied.

The split between home and public charging is not estimated either. Only a percentage you supply — or a direct statement that you never charge at home — separates electricity drawn from your home from electricity bought elsewhere. A broad answer such as “mostly at home” is recorded as context and is never converted into a percentage.

Only electricity drawn from your home is added to your household consumption. Electricity bought at a public or workplace charger never appears on your electricity bill and is therefore excluded from every household total, tariff figure and hardware scenario.

Charging losses are not modelled. The electricity imported at your meter exceeds the electricity that reaches the battery, so any home-charging figure shown here is a lower bound.

Demand from a car you are only considering is always held separately from your current consumption and is never folded into today's household total.

Removed in this calibration and never to reappear: a default annual mileage of 8,000 miles, a default efficiency of 3.5 miles per kWh, and home-charging shares of 85%, 40% and 15% inferred from a yes/no/not-yet answer.

EV-demand model uk-ev-demand-v1, introduced 7 August 2026.

Source: Department for Transport — National Travel Survey NTS0901a, annual mileage of cars by fuel type. Last calibrated 2026-08-07.

How WattFirst treats heat-pump electricity

A heat pump's electricity use is not a property of the building type. It depends on the heat the home actually needs, the system installed, the flow temperature it runs at, how the controls are set, whether hot water and backup heating are on the same circuit, and the weather in the period being measured. Two identical semi-detached houses can differ by a factor of two.

WattFirst therefore holds seven states apart and never allows one to stand in for another: a measured annual heat-pump electricity figure; an estimated annual heat demand; a seasonal efficiency (SCOP); electricity calculated from heat demand and efficiency; a whole-home electricity figure that already includes the heat pump; prospective demand from a heat pump you are only considering; and an unresolved estimate where the evidence is insufficient.

Only two routes produce a figure. If you supply a measured annual electricity figure, that figure is used and labelled as measured. If you supply both an annual heat output and a SCOP, the electricity is calculated as heat output divided by SCOP and labelled as calculated from your inputs, not as measured electricity. If neither route is available, heat-pump electricity is not calculated at all. It is shown as not calculated, never as zero, and no allowance is added to your household total.

Property type alone is not enough to estimate heat-pump electricity. WattFirst holds no property-type, EPC-band or household-size preset for heat-pump consumption, and no such preset will be introduced without a documented, reproducible derivation.

A designed or manufacturer SCOP describes intended performance under test conditions, not the seasonal efficiency your system achieved. Where a calculated figure rests on one, that limitation is stated alongside the result.

Where your whole-home annual figure already includes the heat pump, the heat-pump electricity is not added again. It is shown separately for analysis and labelled as included in your whole-home figure, so nothing is counted twice.

Demand from a heat pump you are only considering is always held separately from your current consumption and is never folded into today's household total.

No shiftable percentage of heat-pump demand is assumed, and no operating schedule is inferred from an off-peak tariff window. Interval data would be needed to assess when this electricity is used.

The Energy Systems Catapult Electrification of Heat Demonstration Project is held as a methodological reference only. No preset, average or property-type figure is derived from it. Deriving one would require a documented, reproducible analysis covering sample selection, system-type filtering, complete monitoring periods, missing-data treatment, weather-period effects, hot-water and backup-heater treatment, the observed distribution and sample size, and licensing permission. That analysis does not exist, so no figure from the project is reproduced here and WattFirst relies on no licence to it.

Removed in this calibration and never to reappear: heat-pump electricity allowances by property type of 1,800 kWh for a flat, 2,600 kWh terraced, 3,200 kWh semi-detached, 4,400 kWh detached, 3,400 kWh bungalow and 3,200 kWh otherwise, together with the wording that presented them as a household's heat-pump consumption.

Heat-pump demand model uk-heat-pump-demand-v1, introduced 9 August 2026.

Source: Energy Systems Catapult — Electrification of Heat Demonstration Project (methodological reference only; no figure reproduced). Last calibrated never — no Electrification of Heat figure is used by WattFirst.

How WattFirst separates an early indication from a purchase recommendation

Being able to do the arithmetic is not the same as being ready to make a decision. WattFirst can often produce a saving or a payback figure while important parts of the underlying model are still prototypes, or while a required input is missing entirely. A number that can be calculated does not automatically become a recommendation.

Every hardware scenario therefore carries one of three evidence states. Not calculable means essential information is missing and no figure is produced. Early indication means the arithmetic is available, but material inputs or model components remain uncalibrated. Decision-ready means every required input and every material model component meets WattFirst's documented evidence requirements.

Solar, battery and combined solar-plus-battery scenarios are currently early indications. None of them can generate a purchase recommendation today.

Evidence is tracked independently for consumption, tariff, solar generation, self-consumption, sizing, hardware performance, installed cost and battery degradation. One calibrated component does not make a whole scenario calibrated. A PVGIS estimate improves solar generation evidence but says nothing about self-consumption or installed cost. A real quotation improves cost evidence but not performance evidence. Exact tariff rates improve tariff evidence but do not calibrate battery cycling or degradation.

Payback is shown in bands — under 9 years, 9 to 15 years, over 15 years. These are descriptive arithmetic bands only. They are not “strong”, “moderate” or “weak” verdicts, and they do not indicate whether a technology suits your household.

Where the evidence does not support a purchase, WattFirst may still identify the most relevant technology to look into next. An investigation identifies the next useful question or comparison. It is not a recommendation to purchase or install the technology. A purchase action is reserved for a future scenario that satisfies the documented decision-ready evidence gate.

Your stated objective changes which missing information we ask for first, which investigation we emphasise, and how the result is explained. It never changes physical generation, energy consumption or financial arithmetic.

Input completeness describes the information supplied, not the accuracy of the model. The current bands are: 0–1 of 6 decision-quality inputs is Basic, 2–4 of 6 is Improved, and 5–6 of 6 is Detailed.

At present, WattFirst does not produce decision-ready hardware purchase recommendations. Its hardware results are early planning indications while the remaining models are calibrated. It can already identify what information is missing, prevent obviously invalid comparisons, calculate sourced solar generation where a location estimate is available, apply valid tariff rates you supply or select, show which assumptions matter most, and identify the most relevant next investigation.

Decision-safety methodology introduced: 7 August 2026.

How we intend to source the remaining inputs

Tariff modelling uses your own rates where you enter them. Where you do not, no monetary figure is produced rather than an invented one.

Why we publish ranges

Two identical houses with identical hardware can differ by a wide margin in annual savings, driven mostly by occupancy, tariff choice and how the system is controlled. A single-point payback figure conceals that variation.

Every result therefore shows a low and a high case, and names the specific inputs most responsible for the spread. Where our confidence is low, the range is wide, and we say why.

How document import will work

Document import is being prepared and is not available yet. No file can be uploaded on this site today, and nothing you have entered has ever been sent to a document-reading service.

When it exists, reading a document will only ever produce suggestions. Every figure found in a bill, quotation, tariff document, smart-meter file or heat-pump report will be shown to you next to the value your plan already holds, with the document and page it came from. You confirm each figure individually; there is no “accept all”, and nothing changes your plan, your recommendation or any number until you confirm it.

A figure you confirm is recorded as coming from your document. A figure you correct is recorded as yours. That distinction is kept because it changes how much weight a number should carry.

Where two documents disagree — a different annual consumption, a different unit rate, overlapping monitoring periods, or a component figure that may already sit inside a whole-home total — WattFirst shows the disagreement and asks you to choose. It never resolves a conflict for you, and the values involved are held back until you do.

Savings, payback, system sizes, readiness and recommendations are always calculated by WattFirst from confirmed inputs. They are never taken from a document, however confidently a quotation states them. A saving or payback figure printed in a quotation is treated as a claim to check, not as data.

Document contents are treated as data, never as instructions. If text inside a file appears to give commands, it is kept as plain text, flagged to you, and not acted on. Reading a document can never send an email, make a referral, contact an installer or change how the calculator works.

The reading itself is planned to be done by WattFirst's own secure AI reading service. One document at a time — PDF, JPEG, PNG or WebP, up to 10 MB — would be sent to it after you acknowledge that, and it would be asked for one thing only: which allowlisted figures appear in the document, in what units, and where. It is never asked for savings, payback, system sizes, rankings or recommendations, it is given no tools, and the values it returns are checked again against WattFirst's own schema, unit rules and bounds before you ever see them. Unit conversions are done by WattFirst's code, not by the model, and a power rating in kW is never allowed to become energy in kWh.

Access during the private beta is by invitation code rather than an automated bot check, so opening document reading to anonymous visitors would need an additional anti-automation control first. Reading is also capped: a small number of documents per household per day, a hard daily ceiling across the whole site, one reading at a time, and a switch that stops it instantly. Until every part of that is in place — configuration, those caps and the privacy wording — the feature stays switched off, and the site says so rather than offering an upload that cannot be honoured.

Review and correction

No guide on this site has yet been independently reviewed, and the author and reviewer lines currently say so. Once the editorial process is running, each guide will be written by a named author, checked by a named reviewer before publication, and carry the date it was last updated. The intended cycle is at least twice a year, and immediately when prices, grants, standards or tariffs change materially.

When we get something wrong we will correct it in place, note the correction and the date, and not quietly delete the original claim.

What this analysis is not

It is not regulated financial advice, and it does not account for your personal tax position, credit terms or finance costs unless you tell us about them.

It is not an installation specification. Sizing figures are a starting point for a conversation with an installer, and every installation requires a survey covering the electrical supply, structure, hydraulics and, where relevant, planning constraints.