diff --git a/README.md b/README.md
index 388ce1a..68a15b1 100644
--- a/README.md
+++ b/README.md
@@ -532,8 +532,11 @@ for result in client.results(crawl.crawl_id):  # follows next_cursor
 `wait` returns the crawl when it ends, or when `timeout` (seconds,
 default 600) runs out. It does not raise: on timeout the crawl is still
 `running` (call `wait` again, or `stop`), and a `failed` crawl comes
-back with `crawl.error` set. A crawl that ran out of credits, or whose
-API key hit a credit cap, ends `stopped` with `crawl.error.code`
+back with `crawl.error` set. A crawl whose pages held nothing new to
+follow (no link, only links to other sites, or only links the patterns
+reject) ends `completed` with no results and no `error`: check the
+patterns against the URLs the start page links to. A crawl that ran
+out of credits, or whose API key hit a credit cap, ends `stopped` with `crawl.error.code`
 `insufficient_credits` or `api_key_cap_reached` and no `stop_reason`; the
 pages it fetched before that stay readable. `results` does not poll: it returns at the first empty page, so on a
 running crawl it yields only the URLs kept so far. Call `wait` first to
diff --git a/acceptance/crawl/README.md b/acceptance/crawl/README.md
index 86d782f..bedf539 100644
--- a/acceptance/crawl/README.md
+++ b/acceptance/crawl/README.md
@@ -78,7 +78,7 @@ and signal checks.
 | 16 | Errors carry the server code | SDKs: `code` holds it, empty when the body has none; MCP, CLI: `server_code` |
 | 17 | Naming | the API's operation and parameter names, in this surface's casing |
 | 18 | Copy | Crawl is "new", never "beta"; one "still evolving" line; the 429 limit reads as the account's active jobs, shared with Batch |
-| 19 | Failed crawl | an include pattern that matches nothing ends `failed` with `no_items_found`; SDKs return it, not raise it; MCP and CLI say `ok: false` (CLI: exit 1, `CRAWL_FAILED`) |
+| 19 | Patterns match nothing | an include pattern that matches nothing ends `completed` with 0 results and no `error`; SDKs return it; MCP and CLI say `ok: true` (CLI: exit 0) |
 | 20 | Results after stop | the html crawl is stopped while its pages are still being fetched: reading its results ends at once (MCP and CLI: `partial: false`, no note to call again), and the same URLs come back once the pages in flight finish |
 | 21 | Documented defaults | a create without `max_pages` and `max_items` gets the default the docs state (100) |
 
diff --git a/acceptance/crawl/contract.lock b/acceptance/crawl/contract.lock
new file mode 100644
index 0000000..435df9b
--- /dev/null
+++ b/acceptance/crawl/contract.lock
@@ -0,0 +1,3 @@
+1adcbc2fd87e3f508317f3ce24c3d48573aec657
+pending GET /healthz
+pending POST /internal/v1/crawls/{crawl_id}/stop
diff --git a/acceptance/crawl/surface/test.py b/acceptance/crawl/surface/test.py
index 2203ce2..eef2aba 100644
--- a/acceptance/crawl/surface/test.py
+++ b/acceptance/crawl/surface/test.py
@@ -704,15 +704,12 @@ def c19():
     f = state["F"] = create(depth=1, include_patterns=[NO_MATCH])
     w = client.wait(f.crawl_id)
     err = w.error.model_dump(mode="json") if w.error else None
-    ok = (
-        w.status.value == "failed"
-        and err is not None
-        and err.get("code") == "no_items_found"
-        and bool(err.get("detail"))
-    )
+    urls = [r.url for r in client.results(f.crawl_id)]
+    ok = w.status.value == "completed" and err is None and w.coverage.items_found == 0 and not urls
     return (
         pass_if(ok),
-        f"{f.crawl_id}: {w.status.value}; error {json.dumps(err)}; returned, no exception",
+        f"{f.crawl_id}: {w.status.value}; items_found {w.coverage.items_found}, "
+        f"{len(urls)} results; error {json.dumps(err)}",
         w.model_dump_json(),
     )
 
@@ -837,8 +834,8 @@ def c21():
 srv.shutdown()
 run_check(
     "19",
-    "Failed crawl",
-    "a crawl that keeps nothing ends failed (no_items_found); wait returns it, not raises",
+    "Patterns match nothing",
+    "a crawl whose patterns match nothing ends completed with 0 results and no error",
     c19,
     live=True,
 )
diff --git a/acceptance/crawl/tools/report.py b/acceptance/crawl/tools/report.py
index 37eccf4..9f46140 100644
--- a/acceptance/crawl/tools/report.py
+++ b/acceptance/crawl/tools/report.py
@@ -38,7 +38,7 @@
     "16": "Errors carry the server code",
     "17": "Naming",
     "18": "Copy",
-    "19": "Failed crawl",
+    "19": "Patterns match nothing",
     "20": "Results after stop",
     "21": "Documented defaults",
 }
diff --git a/src/zenrows/crawl/client.py b/src/zenrows/crawl/client.py
index d4e4039..dac1517 100644
--- a/src/zenrows/crawl/client.py
+++ b/src/zenrows/crawl/client.py
@@ -187,7 +187,9 @@ def create(
         - `include_patterns` / `exclude_patterns`: substrings matched
           against each normalized URL. A URL is kept when it matches
           an include pattern (if any) and no exclude pattern. A crawl
-          stays on the start URL's registrable domain; subdomains count.
+          whose patterns reject every link ends `completed` with no
+          results and no `error`. A crawl stays on the start URL's
+          registrable domain; subdomains count.
         - `output_format="html"` or `"markdown"`: also fetch every kept
           URL's page, as HTML or as Markdown, readable with `content` /
           `download`. Omit for URLs only.
diff --git a/src/zenrows/crawl/models.py b/src/zenrows/crawl/models.py
index ab01e89..e6858cb 100644
--- a/src/zenrows/crawl/models.py
+++ b/src/zenrows/crawl/models.py
@@ -80,7 +80,13 @@ class CrawlErrorCode(Enum):
 
     `insufficient_credits` and `api_key_cap_reached` come with
     `status` `stopped`: the pages fetched before that stay readable.
-    The other codes come with `failed`."""
+    The other codes come with `failed`. `seed_unreachable`: the start
+    page could not be reached, was blocked, or came back empty, with
+    no visible text and no link. `no_items_found`: the crawl read the
+    start page but found no data it could use, so it had nothing to
+    follow; start from a listing or product page. A crawl that returns
+    nothing for any other reason is not a failure: it ends `completed`
+    with no results."""
 
     INSUFFICIENT_CREDITS = "insufficient_credits"
     API_KEY_CAP_REACHED = "api_key_cap_reached"
diff --git a/tests/test_crawl_client.py b/tests/test_crawl_client.py
index 208ff36..4b7267f 100644
--- a/tests/test_crawl_client.py
+++ b/tests/test_crawl_client.py
@@ -680,7 +680,11 @@ def test_wait_returns_failed_crawl(client: ZenRowsCrawlClient, no_sleep):
             200,
             json=crawl_body(
                 status="failed",
-                error={"code": "no_items_found", "detail": "No URL matched."},
+                error={
+                    "code": "no_items_found",
+                    "detail": "The crawl read the start page but found no data it could use, "
+                    "so it had nothing to follow. Start from a listing or product page.",
+                },
                 results=[],
                 next_cursor=None,
             ),