# AI lead magnet planning worksheet

Use one copy per client concept. Keep the first version narrow enough to test with real inputs in a week or two, even if the polished implementation takes longer.

## 1. Business fit

- Client:
- Offer the lead magnet should lead toward:
- Intended audience:
- Decision the visitor is trying to make:
- Useful result we can provide before a sales call:
- Natural next step after the result:
- Why this is better than a static PDF for this decision:

## 2. Result contract

Complete this sentence:

> After providing __________, the visitor receives __________, which helps them decide __________.

- What the result includes:
- What the result explicitly does not include:
- Assumptions shown to the visitor:
- Confidence or uncertainty language:
- When the experience must recommend human review:

## 3. Inputs

| Input | Why it is needed | Required? | Sensitive? | Validation or allowed range |
|---|---|---:|---:|---|
|  |  |  |  |  |
|  |  |  |  |  |
|  |  |  |  |  |

Data-minimization check:

- Could we produce a useful preview before asking for contact details?
- Can any free-text field become a constrained choice?
- Are we asking for information only because sales might want it later?
- How long will submissions be retained, and who can delete them?

## 4. Logic map

- Deterministic rules or calculations:
- Approved source material for retrieval:
- Generative tasks:
- Prohibited claims or topics:
- Fallback when the model or API is unavailable:
- Output checks before a result is shown:
- Test cases for extreme, empty, malicious, and ordinary inputs:

## 5. Result page

- One-sentence answer:
- Three prioritized recommendations:
- Evidence or assumptions displayed:
- One action the visitor can take now:
- CTA that matches the result:
- Email or downloadable version:
- Shareable summary, if appropriate:

## 6. Qualification without interrogation

Choose no more than three signals that improve either the result or the follow-up.

1. Signal:
   - How it changes the result:
   - How it helps follow-up:
2. Signal:
   - How it changes the result:
   - How it helps follow-up:
3. Signal:
   - How it changes the result:
   - How it helps follow-up:

## 7. Accessibility, privacy, and safety

- Every control has a persistent label:
- Instructions and errors are available in text:
- Keyboard-only path tested:
- Result is understandable without color:
- Privacy notice and consent language reviewed:
- Sensitive data excluded or protected:
- Prompt-injection and hostile-input tests completed:
- Model output is treated as untrusted until validated:
- High-stakes advice is excluded or sent to a qualified reviewer:

## 8. Measurement plan

Define the funnel without assuming that more form fills always means a better lead magnet.

| Stage | Event | Quality check |
|---|---|---|
| Start | Visitor begins | Traffic source is relevant |
| Complete | Visitor receives a result | Result generated successfully |
| Continue | Visitor takes the next step | CTA matches the result |
| Qualified | Lead meets agreed criteria | Criteria were defined before launch |

- Baseline to compare against:
- Review date:
- Signals for revising questions:
- Signals for retiring the lead magnet:

## 9. Ownership and handoff

- Business owner:
- Content/source owner:
- Technical owner:
- Approved model or API:
- Monthly operating-cost guardrail:
- Person who reviews failed generations:
- Update cadence for source material:
- Export and deletion process:
- Shutdown fallback:

## Scope summary

- Smallest useful first version:
- Features deliberately postponed:
- Dependencies:
- Acceptance criteria:
- Launch checklist owner: