How to Use Datadis Electricity Consumption Data for Commercial Solar

A commercial solar system is worth what its output displaces at the moment it is produced. That makes the timing of a business's electricity use the central input to any assessment, and timing is exactly what an electricity bill does not show. A bill records how much was consumed over a billing period and what it cost. It does not record whether that consumption occurred at eleven in the morning, when a solar array is at full output, or at ten at night, when it produces nothing.

Interval consumption data closes that gap. It records electricity use in hourly or quarter-hourly steps across the year, so modelled solar generation can be laid over actual demand and the result separated into electricity the business would consume on site, surplus it would export and generation it would waste under some configurations.

In Spain, the usual route to that data is Datadis.

What Datadis is

Datadis is a platform created by the Spanish electricity distribution companies to give consumers a single point of access to the consumption data held about their own supply points. It was launched in 2020, and the practical benefit is that a business with several sites served by different distributors can reach all of them through one interface rather than registering separately with each.

An important distinction: Datadis does not generate the readings and does not own them. The measurements are taken by the meters and held by the distribution companies, which supply them to the platform. Datadis provides access. Where a reading is missing, delayed or estimated, that originates upstream in the metering and distribution chain rather than in the platform itself.

The platform provides common access across participating Spanish distributors, including e-distribución, i-DE, UFD, E-Redes and Viesgo Distribución.

Distributor and supplier are not the same

This causes more confusion than any other point, and it matters because the data comes from one of them and the bill from the other.

The distributor, or distribuidora, owns and operates the physical network, installs and reads the meter, and is assigned by geography. A business cannot choose it. The distributor holds the consumption measurements.

The supplier, or comercializadora, is the company the business contracts with, buys electricity from and pays. A business can change supplier freely, and doing so does not change the distributor, the meter or the consumption history.

A business that switches supplier does not lose its consumption record. The distributor still holds it, and it remains reachable through Datadis.

Some suppliers also present consumption curves in their own customer portals, which can be a convenient alternative. The data still originates with the distributor.

Finding and checking the CUPS

The CUPS, Código Universal del Punto de Suministro, is the unique reference for a supply point. It is the key to everything that follows.

It appears on every electricity bill, usually near the supply details, and takes the form of a twenty or twenty-two character code beginning ES followed by digits and letters, for example ES0021000000000000AB0F.

Three practical points. The CUPS identifies the supply point, not the business, so a company with several sites has several CUPS references and each must be handled separately. The CUPS does not change when the supplier changes, which is what makes it a stable identifier across years of contract history. And the digits following ES indicate the distributor, which is a quick way to confirm which network serves a site before starting.

Before requesting data, check the CUPS against a recent bill character by character. A transcription error produces a failed lookup rather than an obviously wrong answer, and it is a common cause of delay.

Who can access the data

Three routes exist, and they involve different amounts of effort and different timescales.

The contract holder. For a company, this means a representative acting on behalf of the legal entity holding the supply contract. Registration requires the company's NIF and the CUPS, and supporting documentation may be requested to verify the relationship between the person registering and the entity named on the contract.

An authorised third party. The holder registers and then grants access to a named adviser, consultant or installer through the platform. This is generally the cleanest arrangement for a solar assessment, because both parties see the same dataset and there is no exchange of files by email.

A partner organisation. Companies that have completed Datadis's own partner registration can request access on the basis of documented consent from the customer, which reduces the steps the customer has to take. This is a formal arrangement with its own requirements rather than something an installer can simply assert.

Access is not automatic in any of these cases. Registration, identity verification and any required authorisation all take time. The interval between requesting data and receiving it varies. Where a project has a deadline, start this early rather than assuming same-day access.

An installer or consultant does not have a right of access. Authorisation must be granted, and it can be withdrawn.

Obtaining the data

The platform interface, its menu structure and the export formats it offers are changeable, so the sequence below describes what has to be achieved rather than which buttons to press. Check the current interface and the platform's own help material when you come to do it.

1. Register on the platform as the supply point holder, providing the NIF of the entity named on the contract and the CUPS reference. Supporting documentation may be requested.
2. Verify the registration through whatever identity check the platform applies at the time.
3. Confirm the supply point appears and that its details match the bill, particularly the access tariff and contracted capacity.
4. Grant authorisation to any third party who needs access, if that route is being used.
5. Select the date range. Request as much as is available, and at minimum a complete twelve months covering a full seasonal cycle.
6. Export the consumption data and retain the file unaltered as the source record before doing anything to it.

Keep a note of the date the export was taken and retain the original file untouched. Working from a dated, unaltered source export makes every subsequent step traceable and allows any analysis to be reproduced or checked later.

What the data contains

According to the platform's own documentation, the information supplied by the distributors relating to a supply point covers a defined set of fields.

For a solar assessment, the load curve is the centrepiece, but the surrounding fields matter too. The access tariff and contracted capacity by period are needed to value the electricity that solar would displace, and the connection voltage is a first indication of what an installation would involve electrically.

Hourly or quarter-hourly

The platform's documentation describes hourly load curves, with quarter-hourly curves in the case of large consumers.

Which applies to a specific site depends on the meter installed, the nature and size of the supply, and the records the distributor holds and has passed to the platform. It should not be assumed in advance. Check what actually arrives before designing an analysis around a resolution that may not be there.

Hourly data is often suitable for preliminary commercial solar sizing, and for a site with a stable, predictable load it may be all that is required. It has a known limitation, though. Averaging consumption across a full hour smooths short load changes within that hour, and because the overlay compares averages rather than instantaneous values, hourly matching tends to overstate self-consumption relative to quarter-hourly matching. Where a load switches on and off within the hour, the hourly average implies a steadier demand than actually existed, and generation that would in reality have been exported appears to have been consumed.

Finer resolution can therefore materially improve the sizing result. It matters most where loads are intermittent, where the site runs batch processes, where large items of plant cycle, or where generation and demand both fluctuate sharply through the day. It is also the resolution needed to see the short demand peaks that determine whether contracted capacity is exceeded, which is a separate exercise often worth running alongside the solar assessment.

Where only hourly data is available and the load is known to be variable, the honest approach is to use it, state the limitation, and treat the resulting self-consumption figure as an upper estimate rather than a precise one.

On history: request as much as the platform offers for the supply point and work with what is returned. Available depth varies, and there is little value in planning around a specific number of years before seeing what is there. Where less than twelve months is available, say so in any resulting analysis rather than extrapolating quietly.

Checking the dataset before using it

Raw consumption data contains features that will distort a model if carried through unexamined. Work through the file before analysing it.

Gaps. Periods where no reading appears. Establish how many hours are affected and whether they cluster, since a scattered handful matters far less than a missing fortnight in July. Record how any gap was handled, whether by exclusion, by interpolation or by substitution from a comparable period.

Duplicate timestamps. Investigate duplicated timestamps individually rather than removing them as a class. A repeated hour at the October clock change may be entirely valid, while other duplicates may indicate an issue requiring correction. Establish which is which before deciding.

Daylight saving. Spain moves the clocks in late March and late October. The March change produces a day with twenty-three hourly records and the October change a day with twenty-five, including a repeated hour. These are correct records of what happened, not corruption, and deleting them or forcing every day to twenty-four hours introduces an error. Handle the clock change explicitly, and be careful about which time reference the export uses, since aligning consumption in one time basis with generation modelled in another shifts the entire overlay.

Zero readings. A zero is not automatically evidence of zero consumption. It may indicate a genuinely idle supply, or a communications failure, or a period when the meter did not report. A whole weekend of zeros at a site with continuous refrigeration is a data problem. A zero at three in the morning at a site that shuts down completely may be entirely real. The way to tell is to compare against the site's known operating pattern and against neighbouring periods.

Estimated readings. Where the export or the supporting distributor record identifies readings as estimated, those reflect a calculation rather than the site's actual behaviour, and their weight in the analysis should be considered. Not every dataset flags this, so where it matters the point is worth raising with the distributor.

Shutdowns and abnormal periods. An August closure, a refurbishment, a strike, a plant failure or a temporary change in operations all show in the data. The judgement to make is whether each will recur. A regular annual shutdown is a permanent feature that will produce surplus every year and belongs in the model. A one-off closure is different: it should be identified, and where the model is intended to represent a typical future year, the affected period should be replaced or normalised using a documented method, such as substitution from the equivalent period in another year or from a comparable operating period. What matters is that the treatment is recorded, not that the period is silently dropped.

Document every adjustment made. An analysis whose data cleaning cannot be traced cannot be checked by anyone else.

Building consumption profiles

Once the data is sound, structure it before overlaying anything.

Daily profile. Average consumption by hour of day, which reveals when the site starts, when it stops, whether there is a midday dip and how much base load runs overnight. Continuous daytime base load provides the most dependable opportunity for on-site solar consumption, since it is present throughout the generating window. How much that consumption is worth is a separate question, determined by the price applying during each interval under the business's supply contract.

Weekly profile. Consumption by day of week, which exposes the difference between operating days and non-operating days. This is often the single most informative view for a solar assessment.

Monthly and seasonal profile. Consumption by month, showing cooling loads in summer, heating in winter, harvest or production peaks and any shutdown.

Combined view. A heat map of hour against day across the year shows the operating pattern, the shutdowns and the anomalies in one image, and tends to surface things that summary statistics hide.

Separating day types

Solar produces the same output regardless of whether the business is working, so day types must be separated before any conclusion is drawn.

Split the dataset into working days, weekends, national and regional public holidays, and seasonal closures. Spanish public holidays vary by autonomous community and municipality, and a site with a local holiday that does not appear on a national calendar will show unexplained low-consumption days until that is identified.

Then compare the average working day, the average Saturday and the average Sunday. For a five-day business the difference is usually stark, and it determines how much of the weekend generation has anywhere to go.

Allocating consumption to tariff periods

Spanish commercial tariffs divide the year into six periods, P1 to P6, applied to both energy and contracted capacity. These periods rank the regulated access tolls and charges, with P1 carrying the highest regulated rates and P6 the lowest. They do not by themselves determine what a business pays for a kilowatt-hour, because the total also includes the energy price agreed with the supplier, which may be fixed, indexed or structured differently again, along with adjustment services and taxes.

What this means in practice is that the periods are the right framework for allocating consumption, but the value attached to each period must come from the business's own supply contract and invoices rather than from the period number.

Each interval in the dataset should be tagged with its tariff period using the calendar applicable to that supply point. The Balearics, the Canaries, Ceuta and Melilla follow their own calendars, so a multi-site business may need more than one mapping.

The output is a table showing consumption in each period, which then allows a weighted value to be calculated for the electricity solar would displace. This is what makes the difference between a defensible model and one built on a single blended price per kilowatt-hour.

One structural feature deserves attention at this stage. Saturdays, Sundays and national public holidays fall within P6 across the full day in every season. Under most commercial contracts the electricity displaced at those times is among the least expensive the business buys, though the actual difference depends on the contract, so weekend generation typically yields a lower saving per kilowatt-hour than the same output on a working day. For a business operating Monday to Friday, weekend generation also arrives when on-site demand is lowest, which compounds the effect.

Overlaying generation on consumption

With the consumption series prepared, the generation side is modelled for the specific site, using PVGIS or an equivalent tool, producing an hourly series for a stated capacity, orientation, tilt and loss assumption.

The two series are then compared interval by interval. In each interval:

• Where generation is below demand, all of it is self-consumed and the shortfall is imported.
• Where generation exceeds demand, the demand portion is self-consumed and the remainder becomes surplus.
• Where the installation operates without surplus, that remainder is curtailed rather than exported.

Summing across the year gives the figures the financial case rests on. Repeat the exercise at several capacities, because the result changes non-linearly as the system grows.

Eight quantities that get confused

The last two are the pair most often confused, and they move in opposite directions. A quoted figure of "seventy per cent" means nothing without knowing which is meant. Both should always be stated.

Why annual kWh alone misleads

The shortcut is to divide annual consumption by an assumed annual yield per kWp and treat the result as the system size. For a site consuming 500,000 kWh at an assumed 1,500 kWh per kWp, that gives 333 kWp.

The calculation is not meaningless. It estimates the capacity that would generate roughly the same quantity of energy over a year as the business consumes. What it does not do is establish the economically correct size, because it says nothing about whether generation and demand coincide. A system that produces the right annual quantity at the wrong times delivers a fraction of the value.

Worked example: identical annual consumption, different answers

The following uses clearly illustrative figures to show why the interval data changes the conclusion. Both businesses consume 500,000 kWh a year and both are assessed at 250 kWp with a modelled yield of 1,500 kWh per kWp, giving 375,000 kWh of annual generation. Self-consumed electricity is valued at €0.155 per kWh and exported surplus at €0.040 per kWh, both illustrative.

Business A is a cold store operating seven days a week with continuous refrigeration and a daytime handling operation.

Business B is a light engineering workshop operating Monday to Friday, 08:00 to 17:00, with a three-week August shutdown and minimal weekend load.

The same capacity on the same annual consumption produces a gross benefit differing by around €13,800 a year. Nothing about the equipment differs. The entire gap comes from the operating pattern, which is invisible in the annual figure and obvious in the interval data.

The practical consequence is that Business B should be assessed at smaller capacities as well, because a proportion of its 250 kWp is producing electricity the site cannot use. Business A, by contrast, may well support a larger system than 250 kWp. Neither conclusion is reachable from annual consumption.

What to send a solar assessor

Providing this at the outset avoids the round trips that delay most assessments.

Where the assessor is to obtain the data directly, granting authorisation through the platform is generally preferable to emailing files, since both parties then work from the same record.

Data protection and authorisation

The consumption data relates to an identified supply point. In the case of an individual or a sole trader it also relates to an identifiable person, so it should be handled carefully rather than circulated freely.

The platform's own documentation states that the data will be accessible only to the holder and to those the holder expressly authorises. Practical implications follow from that.

Authorisation should be granted deliberately, to a named party, for a defined purpose. It should be reviewed when a project ends or when a supplier relationship changes, and access that is no longer needed should be withdrawn. Where several installers are bidding, consider whether each needs platform access or whether providing the same exported file to all of them is sufficient, since the latter also ensures every bidder is working from identical data.

Where data is exported and shared, the business should know where the files have gone and be able to ask for their deletion. If the assessment is being carried out under a contract, the treatment of the data is a reasonable thing to address in it.

Specific obligations depend on the parties, the purpose and the arrangement between them, and where a business has particular concerns those should be put to its own advisers rather than settled from a general article.

When Datadis data is unavailable or incomplete

Several situations can leave a business without usable data, and each has a route around it.

Registration or authorisation is delayed. Start the process before it is needed. In the meantime, the distributor's own portal is an alternative. i-DE, e-distribución, UFD, E-Redes, Viesgo and the others each provide access for the CUPS holder, usually requiring a digital certificate or electronic identification.

Insufficient history. A recently connected supply, a new building or a recent change of occupier may leave less than twelve months available. Options are to use what exists and state the limitation, to supplement with data from a comparable site under the same operation, or to install temporary monitoring and record a representative period.

Resolution is coarser than expected. Hourly data supports a preliminary sizing exercise, with the caveat that it can overstate self-consumption where loads are intermittent. Where finer resolution would materially change the answer, whether for the solar overlay itself or for a contracted capacity review, temporary metering can capture it.

The meter itself. Some meters can be interrogated directly, and the supplier or an energy consultant may be able to obtain records this way where the platform route is obstructed.

Temporary monitoring. Clamp meters or logging equipment installed at the main switchboard record actual consumption at whatever resolution is set. This gives good data quickly but covers only the period monitored, so a month of readings taken in April tells you little about August. Where it is the only option, monitor for as long as the programme allows and be explicit about what the period does not cover.

Significant operational change. Where the business has recently changed materially, historical data describes an operation that no longer exists. The honest approach is to model the new pattern explicitly and flag it as an assumption rather than presenting historical data as predictive.

In every case, the limitation belongs in the resulting report. An assessment built on three months of data can still be useful. An assessment built on three months of data and presented as though it used twelve is not.

Frequently asked questions

What is Datadis? A platform created by the Spanish electricity distribution companies that gives consumers access to the consumption data held about their supply points. It provides access to data supplied by participating distributors rather than generating the readings itself.

How does a business access its electricity consumption data? By registering on the platform as the supply point holder using the company NIF and the CUPS reference, then exporting the data. Access can also be granted to an authorised third party, or obtained through a partner organisation with documented consent.

Where do I find my CUPS number? On any electricity bill, usually near the supply details. It begins with ES and runs to twenty or twenty-two characters. Each supply point has its own CUPS, and it does not change when the supplier changes.

Can an installer access the data on my behalf? Only with authorisation. The holder can grant access to a named third party through the platform, or a partner organisation can request access on the basis of documented consent. Access is never automatic and can be withdrawn.

Is the data hourly or quarter-hourly? The platform documents hourly load curves, with quarter-hourly curves in the case of large consumers. Which applies depends on the meter, the supply and the distributor's records, so it should be checked rather than assumed.

Is hourly data good enough to size a commercial solar system? It is often suitable for preliminary sizing. Because it averages consumption across the hour, it smooths short load changes and tends to overstate self-consumption compared with quarter-hourly matching. Where loads are intermittent or vary sharply, finer data can materially improve the result.

How much historical data is available? This varies by supply point. Request as much as the platform offers and work with what is returned. Where less than twelve months is available, the analysis should say so.

Why are electricity bills not enough to size a solar system? Bills show how much was consumed and what it cost over a period. They do not show when consumption occurred, and the value of solar depends entirely on whether generation and demand coincide.

What is the difference between the distributor and the supplier? The distributor owns the network, reads the meter and holds the consumption data, and is assigned by geography. The supplier is the company the business contracts with and pays, and can be changed freely without affecting the meter or the data.

Does a 25-hour day mean the data is corrupt? No. Spain's October clock change produces a day with twenty-five hourly records, including a repeated hour, and the March change produces one with twenty-three. Both are correct and should be handled explicitly rather than deleted.

What does a zero reading mean? Not necessarily zero consumption. It may reflect a genuinely idle supply, a communications failure or a period the meter did not report. Compare against the site's known operating pattern before treating it as real.

What is the difference between self-consumption rate and solar coverage rate? Self-consumption rate is solar used on site divided by total generation, and it falls as the system grows. Solar coverage rate is solar used on site divided by total site consumption, and it rises. They answer different questions and both should be quoted.

Reviewed 19 August 2026. Platform interfaces, registration procedures and export formats change. Check the current process and the platform's own guidance before relying on the sequence described here.

Sources