Tracker Installed, Postbacks Proven Before Handover

Of the last eleven tracker installations I was asked to audit, nine had conversions arriving at the tracker and never leaving it. The postback to the traffic source was missing, malformed or firing on the wrong event, so every optimisation algorithm buying on that account was bidding blind while the reports inside the tracker looked healthy. That is the defect this engagement exists to remove, and it is usually found in the first afternoon. What gets installed: One of Voluum, RedTrack, Binom or Keitaro — chosen after I see your traffic mix and your reporting load, not before. Voluum where the cost API integrations across native and social networks matter more than the licence saving. RedTrack where the buying is mostly platform-side and the team wants managed hosting. Binom or Keitaro on your own server where the report volume is heavy or the data cannot sit with a third party. Keitaro runs on Ubuntu 24.04 with ClickHouse behind the reports; Binom gets its own instance rather than sharing with anything that serves pages. If you already hold a licence, I will work with it rather than sell you a migration. Domains, which is where installs actually fail: The redirect domain and the landing domain are separate, registered separately, and neither is your brand domain. One flagged redirect should not take your site's reputation with it. Custom tracking domains get their own certificates, DNS at a provider that is not also your registrar of record, and no shared CDN account across offers. Direct linking is used where the traffic source penalises hops; the redirect chain is measured, and anything over 250 milliseconds of added latency is treated as a defect and fixed. The wiring: Click ID passed through to the offer and returned on the postback, tested with a real conversion rather than a test fire. Cost pulled by API where the source supports it, and reconciled monthly against the invoice by hand, because API-reported cost and billed cost disagree more often than anyone admits. Server-side conversion delivery to Meta CAPI and Google Ads offline conversions with the deduplication key set properly, so browser and server events do not double-count. For app traffic, AppsFlyer or Adjust postbacks mapped to tracker events, and SKAN and AdAttributionKit conversion values documented so the delayed and coarse values are read as delayed and coarse rather than as a drop. Verification before I hand over: A written test log: a live conversion on each source, each offer and each device class, with the timestamp at every hop. A deliberate broken-postback test so you can see what failure looks like in the reports before it happens for real. A reconciliation of one full week of tracker revenue against network statements, line by line, with the variance explained. I do not consider an installation finished because the dashboard is populated. Out of scope: Cloaking, split-testing systems designed to show reviewers a different page, and traffic distribution set up to evade a network's compliance check. Fraud detection beyond the tracker's own tools — if you need click-fraud analysis, that is FraudScore or a similar specialist and I will say so. Ongoing server administration; I will harden and document the box, but someone needs to own patching afterwards. Not for: A team running one offer on one source, where the tracker costs more in maintenance than the reporting is worth. And anyone who wants the install without the reconciliation week — that week is the part that finds the errors, and the rest is configuration anybody can do.