How One Broken Form Cost an Agency Months of Leads

What actually happened, and the lesson underneath it

One broken form cost an agency roughly one to two months of leads because the failure was invisible: submissions were captured but never delivered to the client, while the campaign dashboards stayed green. On PPC Live The Podcast, Optidge founder Danny Gavin described a client, an autism therapy provider, whose Google Ads were performing well, clicks rising, cost per lead strong, yet no enquiries were landing. The leads existed. They just never arrived.

This is the trap that catches good agencies. Everyone was watching the metrics that move, impressions, clicks, conversion events, and nobody was watching the one thing that pays the bills: did a real human on the client's side actually receive the enquiry? When Gavin's team found the gap, they did the hard, correct thing. They told the client immediately, exported every submission still sitting in the database, and handed back the recovered leads. The financial loss stung less than the feeling of having let down someone he knew personally.

The operational lesson is blunt: a generated lead and a delivered lead are not the same thing, and most reporting only proves the first.

Why broken forms fail silently for weeks

Most website problems announce themselves. A 500 error throws an alert. A broken checkout spikes a support queue. A form that stops delivering does the opposite, it produces nothing, and nothing is the hardest signal to notice. The page still loads, the submit button still animates, the visitor still sees a thank-you message. The break sits downstream, in the part nobody looks at.

There are only a handful of places a lead actually dies, and each one looks fine from the front end:

Because the symptom is absence, detection lag is brutal. Teams write off two weeks of zero enquiries as a slow patch, a seasonal dip, or a bad audience, exactly the rationalization that let this case run for a month or more.

Why your analytics said everything was fine

Here is what most guides on this topic miss: your conversion tracking and your lead delivery are two separate pipes, and they break independently. A Google Ads or GA4 conversion fires the moment the thank-you page loads or a submit event triggers. That happens before the email, the webhook, or the CRM write ever runs. So your platform can confidently report a healthy cost per lead while not a single lead has reached a human.

That is the dangerous bit. The number that looks most trustworthy, conversions, is precisely the one that lies in this scenario. It measures intent to submit, not successful receipt.

If your only proof a lead exists is a conversion event, you have proof that someone clicked submit, not proof that anyone on your side can call them back.

The fix is to monitor the gap between the two pipes. When form-submission events stay flat or rise but actual delivered enquiries fall toward zero, that divergence is the alarm. Watching organic and paid landing pages in Sentinel SERP makes that easier, a page can hold its rankings, clicks, and impressions while its real-world outcome quietly collapses, and seeing traffic stay strong against a lead count that has cratered is often the first honest hint that the problem is the form, not the market.

The form-monitoring checklist that stops silent leaks

You cannot manually test every form every day, so build the checks into your process. This is the routine that turns a one-month disaster into a one-hour incident.

1. Test the full path, not the page. Submit a real test lead and confirm it lands in the inbox and the CRM, with the right routing and owner. Submission is not delivery.
2. Schedule synthetic submissions. Use an uptime or form-monitoring service to fire a test entry on a schedule and alert if it does not arrive at the destination, not just if the page returns 200.
3. Alert on volume anomalies. Set a threshold so a sudden drop in delivered enquiries pages someone. Absence should trigger an alarm, not a shrug.
4. Re-test after every change. Theme update, plugin update, consent-banner tweak, tag deploy, DNS change, anything that touches the site gets a form test before you call it done.
5. Check spam and quarantine weekly. Real leads hide in junk folders and CRM spam queues. Whitelist your form's sending domain with SPF, DKIM, and DMARC.
6. Reconcile two sources monthly. Compare conversion-event counts against actual leads received. A widening gap is your early-warning system.

None of this requires a developer in 2026, no-code form-monitoring tools now handle synthetic submissions and end-to-end delivery checks out of the box. The barrier was never tooling. It was remembering that the job is not finished when the form is built.

Build these checks into the cadence you already run. Most agencies have a monthly reporting ritual, so bolt the reconciliation onto it: the same hour you pull rankings and traffic, pull the delivered-lead count and put the two numbers side by side. A page that holds its position and clicks while its lead total slides toward zero is not a market problem, it is a plumbing problem, and that side-by-side view is what surfaces it in minutes instead of months. The teams that never lose a month of leads are not luckier, they simply made delivered the metric they report on, not submitted.

What to do the moment you discover the leak

When you find a gap like this, the next hour defines the relationship. Danny Gavin's response is the template, and it works because it inverts the instinct to minimize.

• Tell the client immediately. They will find out eventually; the only question is whether you controlled the conversation. Honesty is the only viable response to a real mistake.
• Recover what you can. Check the form plugin's local database, server logs, the CRM trash, and email archives. Many platforms store every submission even when delivery fails, that is how Gavin's team exported the lost leads and handed them back.
• Hand over everything, fast. Give the client the full recovered list with timestamps so they can follow up, even weeks late. A late call still beats no call.
• Show the fix and the new safeguard. Explain the root cause and the monitoring you have added so it cannot recur silently. The safeguard is what rebuilds trust.

The clients who stay are rarely the ones who never see a mistake. They are the ones who watch you catch it, own it, and close the hole. A broken form is an operational failure. Hiding it is a relationship failure, and only one of those is fatal.