diff --git a/esphome/components/web_server/web_server.cpp b/esphome/components/web_server/web_server.cpp
index d5ded2a02c..1242db57ff 100644
--- a/esphome/components/web_server/web_server.cpp
+++ b/esphome/components/web_server/web_server.cpp
@@ -46,70 +46,58 @@ static const char *const HEADER_CORS_REQ_PNA = "Access-Control-Request-Private-N
 static const char *const HEADER_CORS_ALLOW_PNA = "Access-Control-Allow-Private-Network";
 #endif
 
-UrlMatch match_url(const std::string &url, bool only_domain = false) {
-  UrlMatch match;
-  match.valid = false;
-  match.domain = nullptr;
-  match.id = nullptr;
-  match.method = nullptr;
-  match.domain_len = 0;
-  match.id_len = 0;
-  match.method_len = 0;
-
-  const char *url_ptr = url.c_str();
-  size_t url_len = url.length();
+// Parse URL and return match info
+static UrlMatch match_url(const char *url_ptr, size_t url_len, bool only_domain) {
+  UrlMatch match{};
 
   // URL must start with '/'
-  if (url_len < 2 || url_ptr[0] != '/')
+  if (url_len < 2 || url_ptr[0] != '/') {
     return match;
+  }
 
-  // Find domain
-  size_t domain_start = 1;
-  size_t domain_end = url.find('/', domain_start);
+  // Skip leading '/'
+  const char *start = url_ptr + 1;
+  const char *end = url_ptr + url_len;
 
-  if (domain_end == std::string::npos) {
-    // URL is just "/domain"
-    match.domain = url_ptr + domain_start;
-    match.domain_len = url_len - domain_start;
-    match.valid = true;
+  // Find domain (everything up to next '/' or end)
+  const char *domain_end = (const char *) memchr(start, '/', end - start);
+  if (!domain_end) {
+    // No second slash found - original behavior returns invalid
     return match;
   }
 
   // Set domain
-  match.domain = url_ptr + domain_start;
-  match.domain_len = domain_end - domain_start;
+  match.domain = start;
+  match.domain_len = domain_end - start;
+  match.valid = true;
 
   if (only_domain) {
-    match.valid = true;
     return match;
   }
 
-  // Check if there's anything after domain
-  if (url_len == domain_end + 1)
-    return match;
+  // Parse ID if present
+  if (domain_end + 1 >= end) {
+    return match;  // Nothing after domain slash
+  }
 
-  // Find ID
-  size_t id_begin = domain_end + 1;
-  size_t id_end = url.find('/', id_begin);
+  const char *id_start = domain_end + 1;
+  const char *id_end = (const char *) memchr(id_start, '/', end - id_start);
 
-  match.valid = true;
-
-  if (id_end == std::string::npos) {
-    // URL is "/domain/id" with no method
-    match.id = url_ptr + id_begin;
-    match.id_len = url_len - id_begin;
+  if (!id_end) {
+    // No more slashes, entire remaining string is ID
+    match.id = id_start;
+    match.id_len = end - id_start;
     return match;
   }
 
   // Set ID
-  match.id = url_ptr + id_begin;
-  match.id_len = id_end - id_begin;
+  match.id = id_start;
+  match.id_len = id_end - id_start;
 
-  // Set method if present
-  size_t method_begin = id_end + 1;
-  if (method_begin < url_len) {
-    match.method = url_ptr + method_begin;
-    match.method_len = url_len - method_begin;
+  // Parse method if present
+  if (id_end + 1 < end) {
+    match.method = id_end + 1;
+    match.method_len = end - (id_end + 1);
   }
 
   return match;
@@ -1759,7 +1747,11 @@ bool WebServer::canHandle(AsyncWebServerRequest *request) const {
   }
 #endif
 
-  UrlMatch match = match_url(request->url().c_str(), true);  // NOLINT
+  // Store the URL to prevent temporary string destruction
+  // request->url() returns a reference to a String (on Arduino) or std::string (on ESP-IDF)
+  // UrlMatch stores pointers to the string's data, so we must ensure the string outlives match_url()
+  const auto &url = request->url();
+  UrlMatch match = match_url(url.c_str(), url.length(), true);
   if (!match.valid)
     return false;
 #ifdef USE_SENSOR
@@ -1898,7 +1890,10 @@ void WebServer::handleRequest(AsyncWebServerRequest *request) {
   }
 #endif
 
-  UrlMatch match = match_url(request->url().c_str());  // NOLINT
+  // See comment in canHandle() for why we store the URL reference
+  const auto &url = request->url();
+  UrlMatch match = match_url(url.c_str(), url.length(), false);
+
 #ifdef USE_SENSOR
   if (match.domain_equals("sensor")) {
     this->handle_sensor_request(request, match);