SiteWarden

Three most likely causes of death, ranked by probability. They compound: #1 makes #3 easier, and #2 speeds up #1.

---

Cause #1: The pain was too rare to renew (probability ~45% of failure cases)

What happened. The pilot found a few broken forms, and agencies signed up after hearing the stories. Then months passed with no alerts. Agencies asked, "What are we paying for? WP Umbrella already tells us the site is up." The monthly report said "0 failures" three months running, and agencies read it as "nothing to fix", not "we're watching". Month-3 retention was 68%. New sign-ups came from scares, and left after quiet quarters. MRR stalled at ~$1.5k.

Early warning signs

• Pilot: fewer than 1 real failure per 100 forms over 30 days.
• Fewer than 1 in 4 accounts gets a real (not false) alert in their first 90 days.
• Under 60% of agencies forward the report to clients.
• Cancellation reasons say "nothing ever happened" or "we have [other tool]".

Fixable? Partly. You can make quiet months visible (leads-delivered counts, "tested 30 times" proof, MX and email-auth warnings that catch problems before they break). You can't make forms break more often. If failures are genuinely rare, this is an insurance product with insurance churn, and the ceiling is low. That's why day 30 counts failures before anything is built.

---

Cause #2: False alarms and broken tests destroyed trust (probability ~30%)

What happened. Headless tests ran into reCAPTCHA and Cloudflare Turnstile, and failed randomly. Some agencies didn't install the companion plugin, so tests went through the real form and a law firm's receptionist got "TEST — ignore" emails every morning. A plugin update changed a field name and tests failed across 40 sites overnight at 2 a.m. The founder, working ~15 hrs a week, saw it 9 hours later. Two agencies posted about it in a WordPress agency group. And one weekend the checker itself went down, missing a real failure at a client that then lost a week of leads.

Early warning signs

• False-alarm rate above 5% of alerts in any week.
• More than 10% of monitored forms tested without the companion plugin.
• Any day with missed test runs across more than one region.
• Support tickets that start "your test email is annoying the client".

Fixable? Yes, if designed in from day one:

• the plugin allowlist as the default path, with passive silence monitoring for the rest;
• a retry from a second region before any alert;
• tagged test entries that never reach the client's inbox;
• a second-provider watchdog on your own runners;
• a 7-day response duty (condition 3).

It's much harder to fix after the reputation damage, because agencies talk to each other.

---

Cause #3: An incumbent added it, or clones raced to the bottom (probability ~25%)

What happened. Oh Dear added a "form check" to its HTTP checks, since it already supports POST requests with payloads. WP Umbrella added "form health" to its €1.99 bundle, which covered 80,000 sites overnight. Gravity Forms shipped a built-in delivery test, the way Solspace Freeform already has for Craft CMS. Meanwhile, AI-built "form monitor" tools launched every month at £0.29 per form. The wedge became a checkbox.

Early warning signs

• A changelog or roadmap post from Oh Dear, WP Umbrella, ManageWP, Gravity Forms or WPForms mentioning form tests.
• A new Product Hunt or "UptimeRobot alternative"-style launch focused on forms every quarter.
• Trial-to-paid falling while trials hold steady (buyers comparing prices).

Fixable? Partly. You can't stop an incumbent adding a check. You can own what a checkbox won't do:

• inbox and CRM arrival, not just a successful POST;
• the failure-reason library from your own data (what broke, why, how fast it was fixed), published quarterly;
• the agency report that clients actually read;
• fast onboarding from WP Umbrella, ManageWP and MainWP exports.

If an incumbent bundles the full job for free, stop and sell the customer base or the plugin.

---

Honourable mentions (less likely, still real)

• Founder-market fit gap. Without agency experience or agency contacts, you'll build what seems logical, and never get the 20 calls. (Condition 1.)
• Mail-provider changes. Gmail or Outlook tightening rules for forwarding or catch addresses could break arrival checks. Keep more than one arrival method.
• Liability event. A client blames the agency, and the agency blames you, for lost leads during an outage of your own. Terms must cap liability at fees paid, and the product must never say "guaranteed".
• Time-boxing failure. At ~15 hrs/week, the reliability duty eats selling time. If support and incidents exceed 4 hrs/week before $3k MRR, the hours don't fit the product.