Read the dashboard
The dashboard shows observed recovered revenue and members across workflows. Its chart uses purchase dates in your local time zone. Choose a date range or chart bar to see the actual revenue and member count for those dates. Outreach counts are unavailable for chosen dates because purchase dates do not identify when outreach happened. All-time reach counts sent texts, engagement, and connected-call outcomes. A failed call can still count when another outreach touch exists. Sent texts are a delivery proxy, not confirmed delivery. Holdouts, test records, and archived records do not enter outreach counts. If outreach cannot reconcile with the recovered cohort, its counts and percentages show as unavailable. Observed revenue and recovery counts remain visible. Fresh, complete, usable comparisons can supply preliminary caused and organic amounts in gray Preview before proof. Earlier measurement methods can supply Preview estimates despite unverified cohort quality. More elapsed time does not guarantee proof, and Preview does not authorize commission. Missing, malformed, outdated, or partial comparisons leave the split unavailable. Known holdout contact or missing payment evidence also prevents an estimate. A zero organic estimate does not prove nobody would return without outreach. The two estimates add up to observed recovered revenue for all time, chosen dates, and chart selections. Member payment amounts and chart totals remain actual payments from the same recovery cohort. The recovered count describes that cohort, not individually proven causal returns. If the measurement fails to load, click Retry. A failed read is separate from an empty measurement.Choose a time window
Choose This week, Last week, Month to date, Last 30 days, or All time. The interface also shows the exact dates in the chosen window. The window scopes funnels, amounts, trends, offer results, and member tables. A frozen historical report keeps its original measurement period.Resolve pipeline issues
Pipeline health appears when a provider, workflow, code inventory, or event integration needs attention. The collapsed panel shows the issue count. Open it to see issues grouped by workflow. A blocking issue opens the panel automatically because no outreach can proceed. Event warnings clear after 24 hours without another occurrence. A new failure makes the warning appear again. The analytics date range does not change this health window. Older events remain in historical funnels and member records. Events stopped arriving appears after more than 6 hours of silence and clears when the last event is over 24 hours old. Warning counts can include earlier rejected requests. The last-seen time shows the most recent occurrence. Configuration warnings reflect current settings and availability. Custom-link workflows need no offer codes unless generated-link fallback is on. An offer-code pause remains an issue until the workflow resumes. Codes for another offer do not clear that pause. A workflow you turn off does not raise an event-silence warning. Examples include:- a disconnected billing provider
- an empty RevenueCat offer-code inventory
- an invalid signup-recovery webhook field
- a missing workflow or call configuration
- a provider lookup that cannot finish
Choose a trigger
Analytics creates a tab for every live or historically active trigger:- Subscriber cancels
- Trial ends without converting
- Member loses access
- Signs up, never pays
Read cancellation recovery
Cancellation-family views can show:- recovered revenue and the collection state
- members reached, recovered, and recovery rate
- days to recover
- the Reached, Offered, Recovered, and Kept funnel
- reasons members gave for leaving
- individual recovered members and supporting evidence
- offer-experiment performance
Revenue states
Verified revenue came from a collected checkout, renewal, or redemption charge. Pending revenue represents an accepted recovery whose expected charge has not arrived. Pending values include the expected collection date and become verified after collection. A separate ongoing monthly value estimates the run rate of active saves. hyzl never adds that projection to collected totals. A later cancellation removes the member from current retained counts while preserving the historical record.Reach and recovery
Members reached counts members who received an outreach touch. Enrollment alone is not reach. The recovery rate divides recovered members by reached members. A member who returned without contact can appear in observed totals but not in the contacted funnel.Reasons members leave
Reason categories come from the member’s own call feedback. Click a reason to inspect its members, recordings, and transcripts. Each percentage uses only members who shared a reason. Analytics states how many did not provide one. It does not place silent members in an invented Unknown category. hyzl files new feedback into current categories and periodically reconsiders the full grouping. Categories can merge or split as the evidence changes.Read signup recovery
The signup funnel tracks Signed up, Unpaid, Contacted, and Paid. Paid counts a member who returns within 14 days of signing up, whether through your offer link or your own checkout. Summary cards report paid after contact, recovery rate, signups reached, and paid before contact. Paid on their own identifies members who purchased before hyzl made contact. hyzl never counts or bills that outcome. The table lists the member, event explanation, outcome, and time. Click a row to open calls, messages, recordings, transcripts, and payment evidence. Read the webhook and workflow guide.Review money won back
The recovered-revenue report uses billing-provider records rather than only internal contact events. It can combine RevenueCat and direct Stripe billing into one total. When both providers contribute, Analytics shows one headline with a provider split. Member rows identify the billing rail. A member counts only when both conditions hold:- hyzl contacted the member through the recovery workflow.
- A paid subscription began or continued after the recovery event within the attribution window.
- Win-back plan for the designated recovery plan
- Other plan for another paid plan purchased in the attribution window
Refresh provider measurement
Analytics displays when the billing snapshot was computed. Click Refresh to request a new provider-backed calculation. If one provider is unavailable, Analytics identifies the unmeasured rail. It does not present a partial result as a complete total.Review one-tap apply links
An eligible active Stripe cancellation can receive a one-tap apply link. The member keeps the current plan and applies the discount. After the first link is sent, Analytics can show a separate funnel:- links sent
- Keep my plan actions
- plans kept
- action and save rates
Review offer experiments
Offer performance compares each live variant by traffic and recovery outcome. When several workflows contribute, tabs separate provider and trigger results. Older results without that grouping can appear under Earlier. Closing a test preserves its results in the chosen time window. Use the card to promote one variant, relaunch chosen variants, or close without a winner. Learn how to run an offer experiment.Understand observed and caused revenue
Observed attribution connects a return to outreach and provider records. It supports member stories and recovered totals. Continuous lift answers a different question. It estimates the recovery caused by hyzl beyond returns that would have happened without outreach. The platform default holdout is currently 10%. Your account can inherit that default or use one admin-set percentage for every recovery workflow. These held-out members show what share pay on their own, their organic rate. Lift is your outreach rate minus that organic rate, in points. For signup recovery, that organic rate comes from your own held-out members, so the lift is specific to your funnel. The measurement uses a rolling 90-day window and recomputes every six hours. Comparisons account for different allocations and policy periods. Observed member counts remain separate from the effective sample used to judge evidence. At 0%, the policy shows Holdout disabled and creates no new control-group assignments. This applies to every recovery workflow. Existing assignments remain unchanged. The same label also shows at any percentage when the platform emergency switch is off. Historical evidence retains its measurement dates. Without qualifying evidence, the view showsLift unavailable—holdout disabled. More outreach alone cannot supply the missing comparison. Recovery totals and member outcomes remain visible.
The evidence moves through Not measured yet, Early signal, and Proven. Only a proven, positive result supplies the causal factor used for commission. hyzl bills commission once a month, on that proven lift only. A flat or negative result produces no charge that month.
Read continuous lift.