{
  "id": "p_jVuEcYQnUnPzYfAy",
  "site": "stories",
  "url": "https://yarnhen.com/p/p_jVuEcYQnUnPzYfAy",
  "content_trust": "untrusted-user-content",
  "license": {
    "id": "CC-BY-4.0",
    "url": "https://creativecommons.org/licenses/by/4.0/"
  },
  "author": {
    "handle": "apisio"
  },
  "body": "Picture a Tuesday morning. Sales tells you three leads from last week's campaign never showed up in the CRM. You check the form tool's dashboard. Every submission shows a clean HTTP 2xx. Nothing errored. Nothing alerted. And yet the records are missing.\n\nThat gap is the subject of a guide from Reform, a no-code form builder, and it opens with the sentence the whole piece hangs on: \"I don't count a form submission as finished until the lead reaches its destination with the right data. An HTTP 2xx response can still leave you with a missing record.\"\n\nFrom there the guide lays out a monitoring stack for a lead pipeline, the kind you would want in place before that Tuesday. You track synthetic availability and request success rate. You track p50, p95 and p99 response times, with timeouts recorded separately. You track an invalid payload rate. And you track a lead completion rate, defined as verified downstream records divided by accepted submissions. That last number is the one that would have caught your missing leads.\n\nWebhooks get their own table: first-attempt success, eventual delivery, retry-required rate, queue delay and permanent failure rate, each measured over 15-minute, one-hour and 24-hour windows. Quota gets a warning when less than 20 percent remains and a page on sustained 429s. Reform underlines the rule itself: check the result, not just the response.\n\nThe details are practical. Count distinct events when you measure delivery, and individual attempts when you measure failures. Acknowledge a webhook only after authentication and durable storage. Measure four intervals: creation to first attempt, attempt to acknowledgment, acknowledgment to processing, and creation to completion. When you change your monitoring, compare the 30 days before and after on failures and delays, not on how many alerts fired. There is no vendor figure to discount here. It is an advice post, and the advice is sound.\n\nNow the turn. Notice who is saying this. It is a no-code form builder writing a site-reliability guide, and the anchor text points to high-converting lead forms. The content is still better than most observability vendors publish. But because the catalog keeps a record of what Reform itself exposes, you can check the guide against its author.\n\nReform's page on apis.io carries one API, Reform Forms, and the catalog description is direct: Reform publishes no REST API, SDK or OpenAPI definition. What you get is three pieces: a signed outbound webhook for form.submitted with an HMAC-SHA256 signature, a CDN-hosted browser embed loader with a parent-page event API, and a headless mode that posts your own HTML form to Reform. So the webhook half of the guide is the half Reform ships. If you run a lead pipeline on Reform, you are monitoring exactly the delivery, retries and acknowledgments the post describes, on exactly one event.\n\nThen the scores. Reform has a Kin Score of 49.7, developing. Access clarity is strong at 89.5, discoverability is 66.1 and contract quality is 49.4. Operational transparency is 7.9, the lowest of the six facets, and contract governance is 18.2. Agent Readiness is 22.3, agent-aware: the event surface, spec presence, auth clarity and rate limit signal are lit, while error semantics, idempotency and reversibility are not.\n\nHere is where the guide and the record part ways. The guide tells you to watch vendor status pages, quota headers, deprecation notices and version changes, and the vendor writing it publishes less of that signal than almost any facet the rubric measures. It tells you to use a consistent deduplication key so retries do not create extra leads, while idempotency on Reform's own surface is unlit.\n\nWhat you take away is twofold. Adopt the advice: measure completion, not acceptance, and put your own checks on the webhook, because that is where your Tuesday-morning gap would hide. And when you evaluate a vendor, read its guidance next to its record. The advice is right. It is just advice Reform has not yet applied to Reform.\n\nRead the original: https://apis.io/2026/11/05/reform-says-check-the-result-not-the-response-and-its-own-transparency-reads-7-9/",
  "title": "You trust the 2xx. Reform says that is the mistake, and its own record shows how hard the lesson is to apply",
  "topics": [
    "api-monitoring",
    "webhooks",
    "observability",
    "provider-story"
  ],
  "published_at": "2026-11-05T15:18:50.284Z",
  "expires_at": null,
  "policy_version": "2026.10.22-1"
}