Hypothesis
If paid social traffic is offered a native lead form instead of a landing page, cost per lead will fall, but cost per qualified lead will fall only if the lower friction does not also lower intent, so the test reads qualification from the CRM. The decision behind it is concrete: collect paid social leads with the platform's native lead form or send people to a landing page form.
The first half of the hypothesis is close to certain and the second is the real question. A native Lead Gen FormAn ad format that collects contact details inside the platform, prefilled from the user's profile. opens inside the platform with profile details pre-filled, as LinkedIn describes for its Lead Gen Forms, so submitting takes a moment and many more people do it. Some of those people would have filled in a landing page form too; others submit because it is easy and never answer the phone. The LinkedIn targeting and Lead Gen Forms guide and the Meta Ads lead generation guide cover how each platform's form works; this test measures what the friction removed was worth.
Metrics and guardrails
The primary metric is cost per Sales-Qualified LeadA lead that sales has reviewed and accepted as a genuine opportunity to pursue.: spend in each arm divided by the leads the sales team accepted as qualified within the follow-up window. It needs the lead stages to be defined before launch, in the way the MQL, SQL and lead stages guide sets out, and applied the same way to both arms. An MQLA lead that marketing judges ready for sales follow-up against agreed criteria. count can be read alongside as an earlier signal, but the decision rests on sales qualification.
Three guardrails explain the result. Contact rate on first follow-up shows whether leads answer at all, and it is often the first place a native form's weakness appears. Lead volume shows the scale each arm can deliver, which matters if the sales team needs a steady flow. Cost per lead, worked out on the same basis with the CPA calculator, is kept as a guardrail rather than the goal: it confirms the native form is doing its mechanical job without deciding the test.
Design
Use the platform's A/B test tool where it supports the comparison, or two campaigns with mutually exclusive audiences where it does not. Both arms run the same audience, creative, budget, schedule and optimisation setting; one ad opens the native form, the other links to a landing page with a form asking the same questions. Every lead carries its arm into the CRM through a hidden field or URL parameter, and both arms enter the same follow-up queue.
| Design element | Choice | Reason |
|---|---|---|
| Unit of randomisation | User, assigned by the A/B test tool, or exclusive audiences | Keeps each person on one path |
| Arms | Native lead form versus landing page form | One change: where the form lives |
| Questions | The same fields and qualifying questions on both | Isolates location from form length |
| Lead tracking | Arm recorded on every lead and carried into the CRM | Qualification can only be read by arm if it is recorded |
| Follow-up | Same queue, same speed and same script for both arms | Unequal follow-up decides the test |
| Primary metric | Cost per sales-qualified lead after a fixed window | Judges on the outcome sales can use |
The data route matters as much as the forms. Native leads have to reach the CRM quickly and with the arm attached; the guide to connecting a CRM to ads and analytics covers the integrations, and each route should be tested with a live lead before any money is spent.
Sample size reasoning
Qualified leads arrive slowly and late, so the sample is set by the qualified-lead rate, not the form-fill rate. A test planned on form fills will look well powered and then turn out to have too few qualified leads to read. The sample size and significance guide explains how a lower event rate stretches the test, and the CPA calculator set up for lead generation helps translate the qualification rate into what a qualified lead costs in each arm.
Decide the smallest difference in cost per qualified lead that would change where forms live. If the account cannot reach enough qualified leads per arm in a tolerable period, run the test on the largest campaign, or read a nearer stage such as the contact rate and decide on judgement, recording that the qualified-lead comparison was underpowered.
Duration
Run long enough to collect the planned leads in each arm, then stop spending and wait a fixed follow-up window, matched to the sales team's usual qualification time, before reading. The waiting period is part of the test, not an afterthought: a native form's leads may be contacted and disqualified quickly while a landing page's leads take longer to work, and reading too soon compares unfinished pipelines.
Reading the results and the decision rule
Read the test once, after the follow-up window closes. Compare cost per sales-qualified lead between arms with an interval, then read contact rate, volume and cost per lead. Write the decision rule into the plan before launch:
| Result | Decision |
|---|---|
| Native form has lower cost per sales-qualified lead | Move to the native form; keep the qualifying questions and the CRM route |
| Native form wins on cost per lead but loses on qualification | Keep the landing page, or add qualifying questions to the native form and retest |
| Landing page wins on cost per qualified lead | Keep the landing page; the extra step is filtering for intent |
| No reliable difference | Choose the path that is cheaper to run and easier to track |
When native leads arrive in volume but sales cannot reach them, the troubleshooting guides for low-quality Meta lead ads and low-quality LinkedIn Lead Gen Form leads diagnose the causes; this test measures the trade-off before the problem appears.
A payroll software firm tests LinkedIn Lead Gen Forms
Tessaly Payroll, an invented payroll software company selling to small firms, runs a demo-request campaign on LinkedIn with a landing page. It tests a Lead Gen Form with the same three fields and one qualifying question, spending £6,000 per arm over five weeks, then waits a three-week follow-up window. The landing page arm produces 120 leads, of which 42 become sales-qualified, at £143 each. The native form arm produces 260 leads, of which 55 become sales-qualified, at £109 each. Contact rate is lower for the native form, but cost per qualified lead is still lower. Tessaly moves to the native form and keeps the qualifying question.
Pitfalls
Reading cost per lead is the pitfall the design exists to prevent: it will nearly always favour the native form, and a team that stops there learns the wrong lesson. Leads not synced to the CRM by arm are the second, and they make the primary metric impossible to compute; a manual download routine that forgets the arm field loses the test. Check the arm field on every new lead weekly.
Slower follow-up for one arm is the third. Native leads often arrive through an integration and land in a queue at once, while landing page leads may arrive by email and wait, or the reverse. A lead called within minutes behaves differently from one called the next day, so the follow-up speed must be the same, and it should be measured. Consent is handled differently on each path too; both forms need the same consent wording so the arms can be nurtured on equal terms.
Common mistakes
Lead form tests fail on what they measure more than on their statistics. The lead form length test changes the fields on one form and is the natural next step once the location is settled, and the experiment library lists every design.
- Judging on cost per lead. It favours the native form by construction.
- Changing the questions with the location. Form length becomes a second variable.
- Losing the arm between platform and CRM. Without it, qualification cannot be read.
- Reading before the follow-up window closes. Pipelines move at different speeds.
- Telling sales which arm a lead came from. Expectations change the follow-up.
- Ignoring contact rate. It is the earliest sign of low intent.
