diff --git a/helpers.py b/helpers.py
index 1615bf60..3e061d6c 100644
--- a/helpers.py
+++ b/helpers.py
@@ -1294,19 +1294,34 @@ def serp_website_lookup(company_name):
             for result in data["organic_results"][:5]:
                 url = result.get("link", "")
                 title = result.get("title", "")
-                snippet = result.get("snippet", "")
-                if url and isinstance(url, str) and url.lower().startswith(("http://", "https://")) and not any(bad_domain in url.lower() for bad_domain in blacklist):
-                    normalized_url = simple_normalize_url(url)
-                    if normalized_url != "k.A.":
-                         normalized_company = normalize_company_name(company_name)
-                         domain_part_normalized = normalized_url.replace('www.', '').split('.')[0]
-                         title_lower = title.lower()
-                         snippet_lower = snippet.lower()
-                         domain_name_match = domain_part_normalized in normalized_company
-                         name_in_result_text = normalized_company in title_lower or normalized_company in snippet_lower
-                         if domain_name_match or name_in_result_text:
-                              logger.info(f"SERP Lookup: Website '{normalized_url}' aus Organic Results fuer '{company_name}' gefunden.")
-                              return normalized_url
+                if not url or not isinstance(url, str) or not url.lower().startswith(("http://", "https://")) or any(bad_domain in url.lower() for bad_domain in blacklist):
+                    continue
+
+                normalized_url = simple_normalize_url(url)
+                if normalized_url == "k.A.":
+                    continue
+
+                # --- NEUE, ROBUSTERE VALIDIERUNGSLOGIK ---
+                normalized_company_name = normalize_company_name(company_name)
+                # Zerlege den Firmennamen in signifikante Tokens (mind. 3 Zeichen)
+                company_tokens = {token for token in re.split(r'\s+', normalized_company_name) if len(token) >= 3}
+
+                if not company_tokens:
+                    continue # Wenn keine sinnvollen Tokens vorhanden sind, überspringen
+
+                domain_part = normalized_url.split('.')[0].lower()
+                title_lower = title.lower()
+
+                # Prüfe, ob mindestens ein signifikantes Token in der Domain oder im Titel vorkommt
+                token_in_domain = any(token in domain_part for token in company_tokens)
+                token_in_title = any(token in title_lower for token in company_tokens)
+
+                # Wir akzeptieren die URL, wenn ein Token in der Domain ODER im Titel ist.
+                # Eine Übereinstimmung in der Domain ist ein stärkeres Signal.
+                if token_in_domain or token_in_title:
+                    logger.info(f"SERP Lookup: Website '{normalized_url}' aus Organic Results fuer '{company_name}' gefunden (Token-Match).")
+                    return normalized_url
+                # --- ENDE DER NEUEN LOGIK ---
 
         logger.info(f"SERP Lookup: Keine passende Website fuer '{company_name}' gefunden nach Pruefung KG und Top Organic Results.")
         return "k.A."