🇩🇪 Energieeffizienz-Experten — German Energy Adviser Leads

Germany's official register of the experts who sign off federal energy-efficiency subsidies — with the email address.

The Energieeffizienz-Expertenliste für Förderprogramme des Bundes, run by the dena (Deutsche Energie-Agentur) for the federal economics ministry, is the gate every German subsidy goes through. A homeowner cannot claim a BAFA Energieberatung, a KfW Effizienzhaus loan or a BEG Einzelmaßnahme grant unless a listed expert signs it off. So this register is, in practice, the national list of the people who decide which heat pump, which insulation system and which window goes into a subsidised German renovation.

This Actor turns it into a lead list — one row per expert, with the contact details and the full matrix of which funding programmes that expert is actually accredited for.

---

📊 The numbers, measured on 2026-09-16

Every figure below was counted against the live site on the day this Actor was built, from 213 records captured across 8 regions, all three register lanes, and pages ranging from the first to page 401 of 676. Nothing here is an estimate, and the offline validator recounts all of it on every run.

Field fill — counted over 179 distinct experts

Field	Fill		Field	Fill
name	100.0%		email	95.0%
street	100.0%		phone	87.7%
postalCode	100.0%		website	65.4%
city	100.0%		organisation (firm)	83.8%
accreditedProgrammes	100.0%		education	98.3%
programmes (full matrix)	100.0%		businessTypes	86.0%
email or phone	97.2%		selfDescription	50.3%

"Filled" excludes "", null, "N/A" and "-". 213 parsed rows collapsed to 179 distinct experts, so 34 of them were returned by more than one of the searches — deduplication is on by default and a dropped duplicate is never charged.

How big each search is

Search (from 10115 Berlin, 250 km)	Experts	Pages
Wohngebäude (residential)	3,467	232
Nichtwohngebäude (companies + municipalities)	1,556	104
Anlagen und Prozesse (industry)	537	36

Other regions, residential lane, 250 km: Stuttgart 10,131 · Köln 9,002 · München 6,687 · Hamburg 5,127 · Leipzig 4,916 · Dresden 2,908. Radius matters more than anything else: around 10115 Berlin, 5 km = 228 experts, 20 km = 774, 250 km = 3,467.

---

🎯 What makes a row here different from Google Maps

Maps can tell you a German engineering office exists. It cannot tell you that this person may sign a KfW Effizienzhaus application but not a KFN new-build one. This register can, because it publishes a per-expert checkbox matrix, and this Actor reads every box — ticked and unticked:

"accreditedProgrammes": [
  "Energieberatung für Wohngebäude",
  "Effizienzhaus (KfW)",
  "Einzelmaßnahmen",
  "Wärmedämmung",
  "Denkmal und besonders erhaltenswerte Bausubstanz",
  "Klimafreundlicher Neubau (KFN) und Wohneigentum für Familien (WEF)"
],
"programmes": [
  { "scope": "residential", "category": "Energieberatung",
    "programme": "bafa", "label": "Energieberatung für Wohngebäude", "accredited": true },
  { "scope": "residential", "category": "Sanierung",
    "subCategory": "Bundesförderung für effiziente Gebäude: Wohngebäude",
    "programme": "kfw-single", "label": "Einzelmaßnahmen", "accredited": true },
  { "scope": "residential", "measureOf": "kfw-single",
    "label": "Heizung", "accredited": false }
]

An unticked box is recorded as "accredited": false, not dropped — so "who is accredited for Heizung but not yet doing KFN new-builds" is a query you can actually run on the export. The matrix runs from 7 to 29 programme rows per expert, 15 on average.

servesResidential / servesNonResidential / servesIndustry summarise the same thing as three booleans. In the captured sample, 42.5% of residential-lane experts also hold non-residential accreditations and 17.3% hold industrial ones.

Who buys this: heat-pump and insulation brands recruiting Fachpartner (Viessmann, Vaillant, Stiebel Eltron, Bosch), Energieberater software vendors (Hottgenroth, ZUB, EVEBI), renovation lead networks, and anyone selling into the German BEG subsidy chain.

---

📧 The email column is the whole product, and it is a trap

dena hides the addresses. Not by splitting them across tags — the "@" is not a character at all. It is an <img src=".../at.svg"> inside a <span class="print-color">, so the raw text of an email block reads:

roland.orgel        ro-ingenieure.de

Any scraper that takes the element's text and runs an email regex over it finds zero addresses on a page where 95% of the records publish one, reports SUCCESS, and hands you a contact list with no contacts. This Actor puts the "@" back before the tags are stripped, and then checks itself:

• 202 of the 213 captured records publish an email block; 201 reconstructed into a valid address (99.5%).
• The 1 that did not (ebfe.ebfe@info) is malformed on dena's own site — no TLD. It is returned as null rather than shipped, because a broken address in your CRM is worse than an empty field.
• If dena ever changes that markup, the run fails rather than delivering a silently blank email column. RUN_SUMMARY.emailReconstruction carries the counts on every run.
• 169 of the 170 addresses in the sample are distinct, so these are per-expert mailboxes, not one chamber-of-commerce inbox repeated.

---

⚙️ What you control

Input	What it does
Register lane	residential / non-residential / industry. Three separate searches, run in turn.
Postcodes or place names	"10115" or "Berlin" — the site's own box takes both (verified: "Berlin" / 20 km → 778 experts).
Radius	5, 10, 20, 50, 100 or 250 km. These six are the only values the site offers; anything else is refused before a request is made.
Funding programme	The register's own checkboxes (BAFA, KfW Effizienzhaus, Denkmal, KFN, QNG, DIN 18599, DIN 16247 …). Measured: no filter 3,467 → bafa 3,024 → kfn_wg 3,173 → both 2,788, i.e. they combine as AND.
Single measures	Wärmedämmung / Fenster und Türen / Heizung / Lüftung — the heat-pump filter. Residential lane only.
Surname or firm	"Müller" within 250 km of Berlin → 39 of 3,467.
requireEmail / requirePhone / requireContact	Filter before delivery, so an excluded row is never charged.
Deduplicate	On by default. Overlapping radii are matched on name + postcode + street + firm.
Include the full matrix	Turn off for a slim CRM-sized row that keeps only the ticked labels.
maxExperts	Hard cap on rows delivered and therefore charged.

---

💸 Pricing

$0.90 per 1,000 experts ($0.0009 per row). No run-start fee. You are charged once per row delivered to your dataset, and:

• duplicates from overlapping radii are dropped before delivery and never charged,
• rows excluded by your email/phone filters are never charged,
• a search that matches nobody costs nothing — no start fee means a zero-result run is free,
• a run that fails mid-way has charged only for the rows already in your dataset, and RUN_SUMMARY says exactly how many that was.

For comparison, German business-directory scrapers on this Store that sell a plain listing row charge $0.001–$0.004. Nobody else scrapes this register.

---

⏱️ Speed, and the throttle we measured

The register throttles by IP, and it does it politely:

10 result-page requests inside ~51 seconds, then HTTP 429 with an honest Retry-After: 60. Proven IP-scoped, not session-scoped: a brand-new session on the same IP was throttled too. Honouring the header recovered cleanly on the first try.

There is no CAPTCHA, no ban, no escalation — and the 429 body is 1.5 MB of ordinary TYPO3 HTML, so a scraper that only looks at the response size sees a fat, healthy-looking page. This Actor sends every request on a fresh Apify proxy session, i.e. a different IP, so the per-IP budget is not the ceiling. With no proxy configured at all it paces itself to 8 page requests per minute instead and says so in the log.

• 15 experts per page. Each page is 287–395 KB and takes 3.7–7.1 s.
• At the default 3 parallel requests that is roughly 675 experts a minute.
• Verified 2026-09-16 that the site answers HTTP 200 through Apify RESIDENTIAL, through the datacenter pool and through BUYPROXIES94952 alike. RESIDENTIAL/DE is the default because the pages are ~300 KB — the size at which a datacenter proxy starts truncating a body mid-stream while still returning HTTP 200 — and because a throttle scoped to the IP wants the larger pool.

---

🧯 Honest limits — read these before you buy

1. There is no nationwide search. 250 km is the largest radius the site offers, and an empty location makes it answer HTTP 200 with its 141,923-byte homepage and no results at all. National coverage means tiling postcodes and deduplicating the overlap. This Actor refuses an empty location up front rather than charging you for a search it never ran.

2. We do not know the national de-duplicated total, and will not guess it. The largest single query observed is 10,131 (Stuttgart, 250 km, residential). Regional tiles overlap heavily — 34 of 213 captured rows were duplicates — so the national unique figure is well below the sum of the tiles and has not been measured.

3. The register publishes no expert ID. No profile URL, no numeric id, no data attribute — checked across all 318 captured records. Rows are deduplicated on name + postcode + street + firm. Two genuinely different people with the same name at the same address would merge; an expert who moves office will look like a new row in a later run.

4. The three lanes are three corpora, not three views. A record found through the residential search still carries its full accreditation matrix for all three scopes — but only the lane you search decides which experts come back. Run all three if you want the union.

5. selfDescription is only on half the rows (50.3%), and dena states plainly that it does not verify it: "Diese Selbstangaben werden von der dena nicht überprüft."

6. Fields that do not exist here, and are therefore not advertised: no expert ID, no registration or listing date, no rating or review, no company register (Handelsregister) number, no VAT ID, no coordinates, no distance-from-centre value, no photo. The site does not publish any of them on the result page, so this Actor does not pretend to.

7. robots.txt. Quoted verbatim, as fetched on 2026-09-16:

   User-agent: *
   Disallow: /typo3/
   Disallow: /fileadmin/
   Disallow: /uploads/
   Disallow: /index.php
   Disallow: /?id=
   Disallow: /fuer-private-bauherren/finden-sie-experten-in-ihrer-naehe/
   Disallow: /login
   Disallow: /aussteller/suche-experten
   Disallow: /profil-anlegen
   

   The residential lane's results path is on that list. The non-residential (/fuer-unternehmen-und-kommunen/…) and industry (/anlagen-prozesse/…) paths are not. AhrefsBot, SemrushBot and MJ12bot are disallowed site-wide. Decide for yourself what that means for your use of the data.

8. This is personal data about named individuals — many of these experts are sole traders (20.5% of the captured addresses are free-mail: gmx, web.de, t-online, gmail). GDPR applies. Germany's UWG also restricts unsolicited commercial email to businesses. Use it for recruiting partners and for research, and honour opt-outs.

---

📤 Output

One row per expert:

{
  "name": "Dipl.-Ing. (FH) Hanna Full",
  "organisation": null,
  "email": "kontakt@hannafull.de",
  "phone": "015237060357",
  "website": "https://www.hannafull.de",
  "street": "Schwartzkopffstraße 8",
  "postalCode": "10115",
  "city": "Berlin",
  "addressFull": "Schwartzkopffstraße 8, 10115 Berlin",
  "education": ["Technisches/ naturwissenschaftliches Studium"],
  "businessTypes": ["sonstige Branche"],
  "selfDescription": "Hanna Full ist Ingenieurin im Bereich erneuerbare Energien …",
  "servesResidential": true,
  "servesNonResidential": false,
  "servesIndustry": false,
  "accreditedCount": 6,
  "accreditedProgrammes": ["Energieberatung für Wohngebäude", "Einzelmaßnahmen", "Wärmedämmung",
                           "Fenster und Türen", "Heizung", "Lüftung"],
  "programmes": [ "…the full ticked/unticked matrix…" ],
  "foundViaLane": "residential",
  "searchLocation": "10115",
  "searchRadiusKm": 5,
  "resultPage": 1,
  "resultRank": 2,
  "sourceUrl": "https://www.energie-effizienz-experten.de/…",
  "scrapedAt": "2026-09-16T…"
}

Umlauts and ß survive end to end (Schwartzkopffstraße, Wohngebäude) — the transfer is checked for completeness before a single record is read from it.

RUN_SUMMARY (key-value store) reconciles the run: what each search said was there versus what was read, rows delivered (== rows charged), duplicates dropped, rows excluded by your filters, field fill counted on the rows this run actually delivered, how many email blocks reconstructed, and every non-200 that was retried.

---

🛡️ What this Actor refuses to do

• Report a partial region as a complete one. If a search says 3,467 experts and the walk reads materially fewer without hitting your cap, the run fails.
• Treat a non-200 as the end of the data. It is a transport failure; it is retried on a fresh IP, and if it still fails the run fails and tells you how many rows were already charged.
• Parse a body that did not arrive whole. Pages are fetched uncompressed on purpose so the declared Content-Length can be checked against what arrived, and a page missing its closing </html> is treated as truncated rather than short.
• Confuse "no experts here" with "we could not read it". A genuine 0 Einträge is reported and charged nothing; the homepage-instead-of-results response fails the run.
• Bill for a page it was served twice. Every page is checked against the "Seite X von Y" label it came back with, so a stale pagination token can never quietly re-deliver page 1 fifteen times.

---

Source: energie-effizienz-experten.de, the Energieeffizienz-Expertenliste operated by the Deutsche Energie-Agentur (dena) on behalf of the German federal government. This Actor reads only the register's public search results. It is not affiliated with dena.