- May 12, 2026
- 3 min read
Why I Dropped Formbricks From My Lead Funnel
Three weeks after turning Formbricks surveys into code and putting a validator in front of an agent, I decided to take Formbricks out of my most important funnel: the free website audit.
Not because Formbricks is bad. Because it was the wrong shape for that job.
How the audit funnel worked
My website audit tool started the simple way: a visitor enters their site and email, the audit runs, and the lead is appended to a CSV file. Then I added Formbricks: each lead was also posted to a Formbricks survey through its responses API, so I’d have a dashboard. Leads went to two places on every audit.
Writing down what I actually needed from leads made the problem obvious.
Four reasons it didn’t fit
1. Survey-shaped data isn’t CRM-shaped data. Formbricks stores responses, keyed by survey. The questions I ask of leads look like this: “score under 50, site on Wix, no call booked yet.” That’s a query across leads, audits and bookings, which a survey response store makes awkward.
2. I already owned the form. The audit modal was my own component, with its own inputs, states and animations. Formbricks’ biggest value is its form builder, logic jumps and hosted survey pages. I wasn’t using any of it.
3. Routing was five lines of code. I’d half-assumed qualification logic was a reason to keep a survey tool. It isn’t:
if (form.budget === "0-2k" || form.timeline === "exploring") {
showSoftPitch();
} else {
showBookCall();
}
4. Two stores drift. CSV plus Formbricks meant every lead was written twice, and when one write failed quietly, the two disagreed. With hardcoded IDs and an extra HTTP call to a third party on every audit, Formbricks was also one more thing that could fail outside my control.
And why not just the CSV?
The CSV had its own problems as the main store:
- No safe concurrent writes. Two audits at once can tear a row.
- No queries. Following up meant grep and eyeballing.
- No relations between a lead, its audits and its bookings.
- No hooks. Nothing can send an email when a row appears without polling the file.
The CSV is good at one thing: portability. Every tool imports it. So it stayed, but as a nightly export, not as the place leads live.
What replaces it
- PocketBase as the lead store: a single self-hosted binary with SQLite, an admin UI and a REST API. Leads and audits as related records, queryable, with a dashboard I control.
- A CSV export of the leads collection, nightly, as the disaster-recovery copy.
- Routing in the form itself, in the few lines above.
One store of record, one backup, no third party in the request path. That was the decision; moving the audit tool onto it was the work of the following weeks.
When Formbricks is the right call
This isn’t “never use Formbricks”. It’s very good when:
- People who don’t code edit the forms. Its builder is the product.
- You run real surveys: NPS, in-product feedback, multi-step research with logic, where per-question drop-off analytics matter.
- You have no backend and want hosted forms that just work.
My audit funnel was none of those: a developer-owned form whose answers are really lead records.
What I’d keep from the Formbricks phase
The “as code” habit. Forms as files, a draft-then-publish lifecycle, and a validator that checks what an agent produces all outlived the tool. They apply to any form, wherever its answers end up.
Takeaways
- Choose the store by the questions you’ll ask of the data, not by the form.
- If you already own the form, a form builder adds a dependency, not value.
- One store of record, one export. Two live stores drift.
- Survey tools are for surveys. Leads are records.