S2S postback tracking: an integration checklist
Preserve attribution IDs, handle duplicates and status changes, and prove delivery with a repeatable test.

Preserve identity from click to conversion
A server-to-server postback reports a conversion from one backend to another. It still needs an attribution link. Capture the platform's click or transaction ID at the agreed entry point and store it with the lead or order before later redirects, callbacks or CRM updates. A postback cannot recreate an identifier your system lost.
Create a field map for click ID, your business conversion ID, offer, event, status, amount, currency and event time. Distinguish commission from order value. Everflow documents click-time transaction IDs; Affise distinguishes clickid and action_id. Use each receiver's actual parameter names and encoding rules rather than assuming that the platforms share a universal schema.
Define events, status transitions and duplicates
Agree what submission, approval, rejection, hold and payment mean in this integration. Platform statuses do not automatically match your CRM stages. Specify which transitions create a conversion, update it or change payable value. Document late approvals, reversals and out-of-order messages; a received lead is not necessarily approved or payable.
Define a stable business event identifier and the receiver's deduplication rules. A retry of the same event should not add another commission; a legitimate repeat purchase must remain distinguishable. Test these cases explicitly. Do not treat a status update as a new sale, or silently discard a later rejection because its conversion ID already exists.
Make delivery observable and recoverable
Use authenticated HTTPS endpoints and keep secrets out of public URLs and routine logs. Confirm timeout, success-response and retry behavior with each sender and receiver. For your own delivery worker, persist the event, use bounded retries with backoff for transient failures, and surface exhausted attempts for reconciliation. Fix permanent validation errors before replaying.
Minimum useful logs include UTC receipt and event times, correlation and conversion IDs, offer, event, status, amount and currency, attempt number, HTTP response and validation outcome. Redact credentials and unnecessary personal data. HTTP delivery success alone does not prove attribution: inspect the recorded conversion and its status in the destination report.
Test the whole journey and reconcile
Test a valid click-to-lead-to-approval journey, missing and malformed IDs, duplicate delivery, status reversal, wrong offer or event, delayed conversion and a temporary endpoint failure. Compare one expected business event with one recorded result and the correct value. Reconcile source, tracker and CRM counts by the same cohort and time zone before live spending grows.
Affise and Everflow documentation supports the platform examples here. Sales Machine is a related private-demo lead CRM entry; its live integration behavior has not been verified for this guide. Ask to demonstrate identifier storage, status mapping, retries and exports using this checklist before treating any workflow as confirmed.