EU EV Charging Graph – one EVSE-level inventory of EV charge points in Germany, France and Spain

EU EV Charging Graph merges the three official national registers of public EV charge points – Germany (Bundesnetzagentur), France (IRVE, data.gouv.fr) and Spain (DGT National Access Point) – into one flat table keyed by EVSE ID, and keeps the history the sources do not keep: which charge points appeared, changed or disappeared since your last run.

This is EV charging station data for Germany, France and Spain in one schema. Instead of writing three parsers (a wide German CSV with up to six points per row, a 159 MB French CSV and a DATEX II XML feed), you get the same columns for every country: normalised EVSE ID, operator, location, power, connector types, opening hours, payment methods, quality flags and change tracking.

Rastriq – data infrastructure for official registers. Not affiliated with the Bundesnetzagentur, data.gouv.fr or the DGT.

Use cases: what can you do with EV charging station data for Germany, France and Spain?

1. Benchmark charge point operators (CPOs) - analysts, consultancies and CPO strategy teams. Compare operators by network size and power mix, per country or across all three. Group by operator_name_norm (it merges legal forms and country variants, so Tesla Germany GmbH = TESLA France SARL = tesla) and operator_id (the 5-character eMI3 prefix from the EVSE ID). Count evse_id rows and split by power_kw and connector_type. You get a league table of who operates how many fast chargers where.

2. Site selection for retail, real estate and fleets - retail chains, property developers and fleet managers. Find where chargers already exist before you add or skip them. Filter by postalCodes and minPowerKw, then use lat, lon, power_kw, opening_hours and payment_methods to map the competition around a plot or depot. You get a ranked list of areas with good or poor coverage.

3. Monitor high-power (HPC) roll-out - infrastructure funds, OEM charging teams and trade press. Run delta weekly with minPowerKw: 150 and read the new rows: operator_name_norm, city, power_kw, commissioning_date. removed rows show decommissioned points. You get a weekly count of new fast-charging points by operator and region (recipe below).

4. Investors and infrastructure due diligence - investors, lenders and M&A teams. Check what a target company really owns on the ground: filter operators by name or EVSE prefix (e.g. DEENB), then count points, power classes and commissioning_date, and use quality_flags to see how clean its published data is. You get an independent, register-based view of a network to compare with the figures in the data room.

5. Roaming and EVSE ID matching between platforms - e-mobility service providers, roaming hubs and platform teams. Reconcile your own charge point list with the official registers on the normalised evse_id (DE*ABC*E123 becomes DEABCE123), with evse_id_raw for the original. Rows with missing_evse_id, invalid_evse_format or duplicate_evse are flagged, not hidden, so you can see exactly why an ID does not match. You get a match rate and a fix list per operator.

6. Data quality for maps and apps - navigation, map and charging-app providers. Compare your charger data with the registers: look for points you miss, points that no longer exist (removed), wrong power (changed_fields: ["power_kw"] with previous_values) and coordinates outside the country (coords_out_of_country). You get a prioritised list of corrections.

7. Coverage analysis by postal code - municipalities, mobility consultants and energy utilities. Use postalCodes (full codes or prefixes such as 28 for Madrid province) with includeConnectors: false for one row per station and evse_count. You get chargers per postal area, to compare with population or vehicle registrations.

Recipe 1: weekly roll-out monitor for 150 kW+ chargers

The first run is a full baseline of fast chargers in the three countries; each later run returns only new, changed and removed points. Excluding rows without an EVSE ID and extra copies of repeated IDs keeps the data clean.

{
  "mode": "delta",
  "countries": ["DE", "FR", "ES"],
  "minPowerKw": 150,
  "excludeFlags": ["missing_evse_id", "duplicate_evse_secondary"],
  "includeConnectors": true,
  "maxItems": 0
}

Recipe 2: station-level coverage for Madrid province

One row per station for postal codes starting with 28, DC fast charging only.

{
  "mode": "snapshot",
  "countries": ["ES"],
  "postalCodes": ["28"],
  "minPowerKw": 50,
  "includeConnectors": false,
  "maxItems": 0
}

Why the history matters

Source	EU EV Charging Graph
Germany publishes a monthly file that replaces the previous one	Compares every run with your previous run, per EVSE
France republishes a consolidated base daily, with repeated EVSE IDs from different producers	Keeps every row, gives repeated IDs distinct keys and flags them duplicate_evse
Spain publishes a DATEX II feed, no change flags	change_type: baseline, new, changed, unchanged, removed
Three formats, three encodings, three identifier styles	One schema, one EVSE ID format (DE*ABC*E123 → DEABCE123)

History starts the first time you run the actor: the first Delta run returns everything as baseline; every later run shows what changed since the previous one. Runs where no source has been republished finish in seconds and cost only the start fee.

How to use EU EV Charging Graph

1. Try it: keep the pre-filled input (snapshot, Spain, 100 rows) and click Start. It takes about a minute.
2. Get the whole inventory once: choose Delta (or Snapshot), select the countries you need and set Max items to 0. The first run returns every charge point as baseline.
3. Monitor weekly: schedule the same input every week. France and Spain refresh daily, Germany monthly. Use the same State key (or the same filters) to keep one stream. With an explicit State key the filters are still part of the stream, and changing the countries or the one-row-per-station switch starts a new baseline.
4. Instant bulk copy: Preloaded returns the latest full DE+FR+ES snapshot prepared by Rastriq without any waiting, at a lower price per row (EVSE level, no change history). At launch, if that snapshot is not published yet, the run falls back to the live snapshot mode and is billed as evse-record.

The actor always reads the complete source of the countries you select (about 300 MB in total for all three); filters only decide which rows are returned. That is what makes removed reliable.

Input

Field	Description
mode	snapshot (every charge point in scope, flagged against your last run), delta (default when empty: only new / changed / removed; first run = full baseline), preloaded (ready-made latest snapshot).
countries	DE, FR, ES. Empty = all three.
includeConnectors	true (default): one row per EVSE. false: one row per station with evse_count, the highest power_kw and the union of connector types.
operators	Optional. Operator name (matched on the normalised name) or 5-character EVSE operator prefix such as DEENB.
minPowerKw	Optional. Minimum power in kW (e.g. 50 for DC fast charging).
postalCodes	Optional. Postal codes or prefixes (28, 10115, 75).
excludeFlags	Optional. Leave out rows carrying these quality flags (for example missing_evse_id, duplicate_evse). By default nothing is filtered.
allowMassChange	Advanced. In Delta, a run where more than 10 % of a country's records are new, changed or removed stops with status: mass_change (nothing returned, state untouched) because that usually means a broken source file. Tick it only if you expect such a change.
maxItems	Stop after N rows (0 = no limit). A truncated first run does not save state.
stateKey	Name of your delta stream. By default derived from countries, level and filters, so different setups keep separate histories.

Output

One flat row per charge point (or station). Main fields:

Group	Fields
Keys	record_key (source|evse_id), evse_id, evse_id_raw, operator_id, operator_siren (France), source_record_id, station_id, country, source (bnetza, irve, nap_es)
Operator	operator_name_raw, operator_name_norm
Place	station_name, address, postal_code, city, region, lat, lon
Charging	connector_type (type2, ccs, chademo, domestic…), connector_type_raw, power_kw, status, commissioning_date, payment_methods (common vocabulary: app, card, cash, contactless, credit_card, debit_card, free, free_with_registration, online, pay_per_use, plug_and_charge, rfid, other), payment_methods_raw (as published), opening_hours (OSM style, e.g. Mo-Fr 08:00-18:00; Sa 09:00-12:00), opening_hours_raw (the original text when it cannot be interpreted; opening_hours is then null)
History	first_seen_at, last_seen_at, change_type, changed_fields, previous_values, source_updated_at, source_url
Quality	quality_flags: missing_evse_id, invalid_evse_format, duplicate_evse, coords_out_of_country, missing_coordinates, power_outlier, operator_from_developer, operator_possibly_natural_person, operator_name_redacted, location_coarsened; for repeated EVSE IDs, every copy carries duplicate_evse and only the extra copies also carry duplicate_evse_secondary

Example (Spain, changed power):

{
  "record_key": "nap_es|ESALLEGO0000191",
  "level": "evse",
  "evse_id": "ESALLEGO0000191",
  "evse_id_raw": "ES*ALL*EGO0000191",
  "country": "ES",
  "source": "nap_es",
  "operator_name_raw": "Allego Spain S.L.U",
  "operator_name_norm": "allego",
  "operator_id": "ESALL",
  "station_name": "Allego RAFAEL ALMENAR S.A.",
  "postal_code": "46014",
  "city": "Xirivella",
  "lat": 39.462288,
  "lon": -0.405479,
  "connector_type": "ccs;chademo",
  "power_kw": 50.0,
  "payment_methods": ["app", "credit_card", "rfid"],
  "opening_hours": "24/7",
  "change_type": "changed",
  "changed_fields": ["power_kw"],
  "previous_values": { "power_kw": 22.0 },
  "quality_flags": []
}

Example (Germany, first run, real record). evse_id is null because the register publishes no EVSE ID for this point, so the row is flagged missing_evse_id and keeps a fallback key:

{
  "record_key": "bnetza|1009242|1",
  "level": "evse",
  "evse_id": null,
  "evse_id_raw": null,
  "country": "DE",
  "source": "bnetza",
  "source_record_id": "1009242",
  "source_point_id": null,
  "station_id": "1009242",
  "evse_count": null,
  "evse_ids": null,
  "operator_name_raw": "Shell Deutschland GmbH",
  "operator_name_norm": "shell",
  "operator_id": null,
  "operator_siren": null,
  "station_name": null,
  "address": "Kiesgräble 2",
  "postal_code": "89129",
  "city": "Langenau",
  "region": "Baden-Württemberg",
  "lat": 48.49339,
  "lon": 10.10783,
  "connector_type": "ccs",
  "connector_type_raw": "DC Fahrzeugkupplung Typ Combo 2 (CCS)",
  "power_kw": 150,
  "status": "operational",
  "commissioning_date": "2023-07-25",
  "payment_methods": [
    "contactless",
    "credit_card",
    "debit_card",
    "online",
    "rfid"
  ],
  "payment_methods_raw": [
    "Kreditkarte (NFC)",
    "Debitkarte (NFC)",
    "Onlinezahlungsverfahren",
    "RFID-Karte"
  ],
  "opening_hours": null,
  "opening_hours_raw": null,
  "source_url": "https://data.bundesnetzagentur.de/Bundesnetzagentur/DE/Fachthemen/ElektrizitaetundGas/E-Mobilitaet/Ladesaeulenregister_BNetzA_2026-09-01.csv",
  "source_updated_at": "2026-09-01",
  "first_seen_at": "2026-10-12T15:59:53Z",
  "last_seen_at": "2026-10-12T15:59:53Z",
  "change_type": "baseline",
  "changed_fields": [],
  "previous_values": null,
  "quality_flags": [
    "missing_evse_id"
  ]
}

What the sources contain (measured 12 Oct 2026)

Country	Source	Size	Rows
Germany	Bundesnetzagentur Ladesäulenregister (CSV, monthly)	56 MB	about 116,000 stations, 209,000 charge points. Only about 65,000 points carry an EVSE ID in the register; the others get a fallback key and the flag missing_evse_id.
France	Base nationale des IRVE (data.gouv.fr, daily)	159 MB	about 225,000 rows, 168,000 distinct EVSE IDs; about 57,000 IDs are repeated by different producers (flag duplicate_evse).
Spain	DGT National Access Point, DATEX II (daily)	83 MB	about 12,000 sites, 35,500 charge points.

A full DE+FR+ES run returns about 470,000 rows.

EU EV Charging Graph input parameters

Parameter	Type	Default	Description
mode	string	delta when empty (pre-filled: snapshot)	snapshot, delta or preloaded.
countries	array of strings	empty = all three (pre-filled: ["ES"])	DE, FR, ES.
includeConnectors	boolean	true	true: one row per EVSE. false: one row per station with evse_count, the highest power_kw and the union of connector types.
operators	array of strings	empty = all	Operator name (normalised match) or 5-character EVSE operator prefix such as DEENB.
minPowerKw	number (min 0)	empty	Minimum power in kW. Points without a published power are excluded when this is set.
postalCodes	array of strings	empty = all	Full postal codes or prefixes (28, 10115, 75).
excludeFlags	array of strings	empty (nothing excluded)	Leave out rows carrying these flags: missing_evse_id, duplicate_evse, duplicate_evse_secondary, invalid_evse_format, coords_out_of_country, missing_coordinates, power_outlier, operator_possibly_natural_person. Excluded rows are not billed.
maxItems	integer (min 0)	0 = no limit (pre-filled: 100)	Stop after this many rows.
stateKey	string	empty	Name of your delta stream.
allowMassChange	boolean	false	Delta safety net, see the table above.

Pre-filled input (Spain, 100 rows):

{"mode": "snapshot", "countries": ["ES"], "maxItems": 100}

Weekly delta for fast chargers in Germany only:

{"mode": "delta", "countries": ["DE"], "minPowerKw": 50, "excludeFlags": ["missing_evse_id"], "maxItems": 0}

💰 How much does it cost to use EU EV Charging Graph?

EU EV Charging Graph uses pay-per-event pricing. Prices per 1,000 events on the BRONZE tier:

Event	Price
actor-start	$0.01 per run (4 GB memory)
evse-record (one row in Snapshot mode, or the first Delta baseline)	$0.50 per 1,000 rows
evse-change (one new / changed / removed row in Delta mode)	$2.00 per 1,000 rows
evse-record-preloaded (one row of the ready-made snapshot)	$0.30 per 1,000 rows

Apify plan discounts apply to these prices: Free plan +15 %, Silver -15 %, Gold / Platinum / Diamond -30 %.

Real-cost examples

• Spain: about 35,500 charge points is roughly $17.75 as a first baseline.
• Germany + France + Spain: about 470,400 rows is roughly $235 (about $141 as preloaded).
• Cleaner extract: with excludeFlags set to missing_evse_id and duplicate_evse_secondary the same three countries leave about 268,000 rows, roughly $134. Dropping every copy of repeated IDs (duplicate_evse) leaves about 209,000 rows, roughly $104.
• Weekly delta: you pay only for the changed rows, plus the start fee.
• No source republished: the run ends in seconds with no per-row charge, only the Actor start fee.

Worked example (BRONZE tier): a 1,000-row snapshot is 1,000 x $0.0005 = $0.50, plus the Actor start (4 GB = 4 x $0.0025 = $0.01): $0.51. A weekly delta that returns 100 changed points is 100 x $0.002 = $0.20 plus $0.01: $0.21. The same 1,000 rows as preloaded cost 1,000 x $0.0003 = $0.30 plus $0.01: $0.31.

Use countries, minPowerKw and excludeFlags to pay only for the rows you need, and run a small test (Max items 100) first.

Sources and licences

• Germany: Bundesnetzagentur, Ladesäulenregister, licence CC BY 4.0. Attribution: "Bundesnetzagentur.de".
• France: Base nationale des Infrastructures de Recharge pour Véhicules Électriques (IRVE), data.gouv.fr, Licence Ouverte / Open Licence (Etalab). Attribution: source data.gouv.fr – Base nationale des IRVE, with the date of the extraction. data.gouv.fr has announced that this dataset will be withdrawn on 31 December 2026 and replaced by the national consolidation published by the Point d'Accès National (transport.data.gouv.fr). If the source disappears, the actor stops with a clear message instead of returning partial data.
• Spain: DGT, Punto de Acceso Nacional (NAP-ES), electric charging points in DATEX II. Reuse terms of the NAP dataset have not been verified by us; check the licence shown for the dataset on nap.dgt.es before redistributing the data commercially. Source: DGT – NAP.

You are responsible for including the attributions above when you republish the data. EU EV Charging Graph only adds normalisation, flags and change tracking.

Personal data and redaction

• France: the source also contains contact e-mails and phone numbers of developers and operators. They are never read or returned.
• Germany (and any source): some registered operators are private individuals or sole traders. When an operator name looks like a natural person (a heuristic based on the name pattern and sole-trader markers such as "e.K." or "Einzelunternehmen"), the actor always redacts it, and this cannot be switched off: operator_name_raw, operator_name_norm and station_name are null, and address is null if it contains that name. address is also null, and the coordinates are rounded to 3 decimals (about 100 m) because the site may be a private home. The row is still returned with its EVSE ID, power, connectors, postal code, city and EVSE-derived operator_id, and carries the flags operator_possibly_natural_person, operator_name_redacted and location_coarsened. The name is dropped while the file is parsed, so it is never stored in your delta state or in the logs.
• E-mail addresses and phone numbers that some German operator names carry glued to them are stripped from every text field.
• The heuristic prefers over-redaction to leaks (some company names are redacted too), but it can still miss some names and can flag a few real companies; if a row you expected to name an operator is redacted, the flag tells you why. Use excludeFlags with operator_possibly_natural_person if you prefer to leave those rows out altogether.

FAQ

Is it legal to use this data? The three registers are official open data published for reuse; see the licences above. Check the Spanish terms and your own obligations (attribution, personal data) before commercial redistribution.

Why do some German rows have no evse_id? The register does not require an EVSE ID for every charge point. Those rows keep a stable fallback key (bnetza|<station id>|<position>) and the flag missing_evse_id.

Why do I see the same EVSE ID twice? Different producers publish the same ID (typical in France). All copies are kept, flagged duplicate_evse, with distinct keys.

What happens when a source does not change? In Delta mode the run ends in seconds with status: source_unchanged and no results are charged.

What if a source loses more than 10 % of its rows? The delta is aborted (source_anomaly), the previous state is kept and nothing is marked as removed. The same happens with mass_change when more than 10 % of a country's records change at once, and with source_format_changed when a file no longer has the expected columns or structure.

How do I keep the cost down? Set countries, minPowerKw, postalCodes and excludeFlags to return only the rows you need (excluded rows are not billed), start with Max items 100, and use Delta for repeat runs: unchanged charge points are free.

Why are some fields null? Fields the source does not publish stay null: in the real German record above evse_id, operator_id, station_name and opening_hours are null. Check quality_flags for the reason where one exists.

Related Actors

• EU Pharma Sites Graph – EU pharma manufacturers and distributors from EMA EudraGMDP.
• Google Maps Scraper - Places - Google Maps places data.
• Idealista Scraper - real estate listings in Spain, Portugal and Italy.

Built by Rastriq.