The short answer
Ask for revenue, amount requested, time in business, whether statements are available, and complete contact details — then verify the phone with a one-time code. Every extra field cuts completion and raises quality, so the page is not optimised for conversion rate: it is optimised for the number of records that clear qualification, which is a different maximum.
The standard advice is to reduce fields. That advice comes from e-commerce, where every completion is worth the same. Here it is wrong: a completion from a merchant doing $8,000 a month is worth nothing, and the field that would have removed them was the one you deleted.
The fields that earn their place
| Field | Why it stays |
|---|---|
| Monthly revenue band | The primary qualifier, and it self-selects |
| Amount requested | Separates real intent from browsing |
| Time in business | The floor most funders enforce |
| Statements available? | Predicts whether the deal can move |
| Business and full name | A file, not a first name |
| Phone, verified by code | The difference between a lead and a row |
| Documents and the follow-up cadence |
The one-time code is the last step
The six-digit code is the final gate. The merchant fills in the application, the code goes to the mobile number they gave, and until they enter it the application does not submit and the record does not exist. That is how our own funnel is built. It filters mistyped and fake numbers, VoIP and landlines before they become records, and because it sits after the qualifying fields, nobody is asked to verify a phone for an application that would have been disqualified anyway. Pair it with an email verifier — a third-party check that the address is real and active — and spam runs at one to three percent of submissions.
Which steps earn their drop-off and which just lose merchants is its own question: good friction versus bad friction sets out both lists. One step that failed the test is the bank statement upload, which is why the field above asks whether statements are available rather than asking for the files.
Two thank-you pages, not one
Where the application ends is a tracking decision as much as a copy decision. A submission that clears every gate goes to one URL; a submission that fails goes to a different one. If both land on the same page the pixel cannot tell them apart, so neither can Meta, and the campaign spends the next month finding more of whoever failed. It is the first item on the ten-step build for that reason.
What kills quality
- Rate promises in the headline. The page inherits whatever the ad selected for, and a rate-led page compounds it.
- Optional revenue fields. Optional means unanswered means unqualified.
- Free-text revenue instead of bands. "Good" is not a number you can screen on.
- No thank-you state that sets an expectation. A merchant who does not know a call is coming does not answer it.
- One thank-you page for everyone. It costs nothing to build the second one, and without it the tracking on the whole funnel is guesswork.
The last one is cheap and skipped constantly. A confirmation screen saying somebody will call shortly measurably improves the contact rate on records the page just produced.