The event a campaign buys shapes the customers it attracts. Paying for signup gives creators quick feedback and builders more volume, but can reward accounts that never receive value. Paying for activation moves closer to product success, yet becomes unfair when the milestone is unstable, invisible, or slow. The right choice is the earliest event that reliably predicts enough customer value to support the CPA.
Define signup precisely
A signup might mean an email submitted, a verified account, an accepted invitation, or a new workspace with required company details. These are not equivalent. If signup is payable, specify identity verification, geography, business eligibility, existing-user handling, duplicate-account checks, and whether free consumer emails qualify. Avoid requiring customer information that the marketplace does not need.
Signup works best when account creation has genuine friction, historical signup quality is stable, and the builder can detect duplicates without broad surveillance. It is risky when disposable accounts are easy, the free tier has little intent, or a long onboarding funnel separates signup from value.
Find the activation milestone
Activation should describe the first moment a qualified user receives the product's core value, not a sequence chosen merely because it is easy to log. Interview retained customers and inspect their early behavior. Candidate milestones might be publishing a report, completing a generated asset, inviting a teammate, or running a workflow with real data.
Prefer one milestone or a small conjunction that is understandable outside the product team. “Reached activation score 0.72” is difficult to contract around unless inputs and versions are visible. “Connected one data source and published the first dashboard within fourteen days” gives creators and reviewers a concrete definition.
Score the candidate events
| Dimension | Question |
|---|---|
| Value | How strongly does the event predict retained contribution? |
| Reliability | Can the backend produce it consistently and idempotently? |
| Speed | How long must a creator wait for feedback and approval? |
| Influence | Can creator education reasonably affect completion? |
| Integrity | How cheaply can the event be fabricated or duplicated? |
Score with evidence, not adjectives. Use cohort conversion, time-to-event, event loss, retry behavior, and abuse history. Record uncertainty when the sample is small. The event with the highest total is not automatically correct if it fails a hard requirement such as privacy, auditability, or acceptable payout delay.
Compare the economics
A signup CPA must be lower when only a fraction of signups activate. If an activated customer supports a $70 acquisition ceiling and half of eligible signups activate, the equivalent signup ceiling begins near $35 before accounting for differences in fraud, support, and retention. Do not pay the activation rate for a signup merely to make the offer appear competitive.
Model the funnel by creator cohort. A campaign can produce fewer signups but a higher activation share, which may justify a strong CPA and continued partnership. Compare retained value after the same maturity period. Recent activation cohorts should not be credited with retention they have not yet had time to demonstrate.
Set deadlines and state transitions
Activation needs a completion window. Without one, creators can wait indefinitely and campaign liabilities remain open. Choose a period aligned with normal onboarding behavior, such as seven, fourteen, or thirty days. Explain whether the event must occur before campaign pause and how an event received late is treated.
Show pending signup, activated, approved, rejected, reversed, and paid states separately. A creator should know whether the customer is still inside the activation window without seeing private activity. The builder should know the maximum open liability before refunding unused campaign funds.
Test the event contract
Create cases for a new qualified user, existing user, duplicate account, missing verification, activation just inside and outside the deadline, repeated event, reversed action, and account later refunded. Send them through the real conversion endpoint and compare product, campaign, and payout states.
Ask someone outside the product team to decide each case from the written definition. If their answers differ from the implementation, revise the words or the code. Never resolve the mismatch by telling creators that the internal meaning is “obvious.”
Use a staged policy when evidence is weak
A new product may begin with verified signup under manual approval because activation instrumentation is not ready. That can be acceptable as a bounded pilot if the lower CPA, approval policy, and temporary nature are explicit. Record activation for analysis even when it is not yet payable.
Move to activation after enough journeys show the event is reliable and connected to value. Consider a two-part structure only if the additional complexity is worthwhile: a small qualified-signup amount and a larger activation amount. Make sure retries cannot pay both stages twice and creators can reconcile the two obligations.
Review after launch
Track signup-to-activation rate, median time to activate, event delivery failures, duplicate and rejection rates, creator-level variation, retained behavior, approval time, and appeals. Investigate product or integration changes before attributing every shift to creator quality.
Change the payable event prospectively with a new campaign version. Do not redefine past conversions after seeing results. The event is ready when users can reach it naturally, engineering can report it reliably, creators can understand it, and finance can connect it to a defensible payout.
Use a conversion-event scorecard
Score each candidate event on four dimensions before choosing it: closeness to customer value, ease of reliable measurement, speed of feedback for creators, and resistance to manipulation. Signup is fast and easy to observe, but it may be far from value. Activation is usually stronger, but only if the product has one stable action that genuinely predicts retention. Payment is closest to revenue but can make feedback too slow for an early campaign.
Write the event as a contract an engineer and creator can both understand. “Activated account” is too vague. “A new referred workspace connects one data source and publishes its first report within fourteen days” is inspectable. Define exclusions such as existing users, test accounts, duplicate identities, refunds, prohibited geographies, and events received outside the attribution window.
Before launch, send synthetic events through the same endpoint that production uses. Confirm the referral code, idempotency key, event time, external user identifier, and approval state appear correctly. Then inspect five real journeys manually. A creator should be able to see whether an event is pending, approved, rejected, or reversed without learning sensitive customer details. Revisit the definition when onboarding changes; otherwise the campaign may keep paying for a milestone that no longer reflects value.
Primary sources for Signup vs Activation: Pick the Conversion Event That Reflects Value
The sources for Signup vs Activation: Pick the Conversion Event That Reflects Value were reviewed on July 15, 2026. Check the publisher for revisions and confirm which requirements apply to the campaign, audience, platform, and jurisdiction.