Who it is for
Product analytics leads, mobile engineers, and growth managers who already have a tracking plan — or a board report that pretends they do — and who suspect the current iOS, Android, or web build no longer sends what that plan describes. Typical clients sit in the Klang Valley, Penang, Johor, and Singapore, and send exports to the Seremban office rather than asking us to sit in their stand-up.
The result
A dated findings document that lists each mismatched event, the property that failed, the source (iOS, Android, web, or a specific OEM build), and the report that currently trusts the broken field. The last section ranks the first ten fixes by how much they distort conversion or funnel counts, not by how many events they touch.
Scope
We sample production payloads across the date range you name, usually fourteen to thirty days. We read parser rules where you can share them: required properties, allowed types, and drop behaviour. We compare those rules with the screens that should fire, using your current information architecture, not last year’s sitemap.
Included: payload sampling, schema comparison, a written findings list, a one-hour findings call, and a short appendix of events that look unused. Excluded: rewriting production code, standing in a release train, overnight monitoring, and cleaning the full taxonomy. Those last two belong to taxonomy cleanup or a later retainer, quoted separately.
Who does the reading
Harith Osman leads the payload pass. Amina Rashid reviews the findings against the reports you actually show leadership. Farah Idris collects the export and books the call. Work is prepared at Tingkat 1, Jalan Tun Dr Ismail in Seremban; the call is online unless you prefer to travel to Negeri Sembilan.
How the weeks run
- Intake: sources, date range, whether debug builds must be excluded, and which funnels you cannot afford to misread.
- Sample collection: you send an export (CSV or JSON lines) with a source column. We reject mixed debug and production files until they are split.
- Validation pass: each high-traffic event is checked for presence, type, allowed values, and whether it still matches the screen name in the current build.
- Findings workshop: one hour, findings document sent the working day before.
Most catalogues of a few hundred names finish in two to three weeks from a usable sample. Larger multi-app catalogues take longer; we say so at intake rather than discovering it in week two.
What to prepare
A production export with timestamps, event name, properties, and a source field. The current tracking plan if it exists, even if it is a neglected spreadsheet. A list of the five reports that would cause an argument if the count moved. Access to a staging build is useful and not required.
Constraints
We do not accept samples that include personal identifiers beyond what the event already carries for product analytics. We do not validate ads network postbacks or warehouse models in this audit. If the parser lives in a vendor console we cannot see, you export the rules; we do not ask for unrestricted admin seats.
Price basis
From RM 7,800. The figure moves with the number of distinct event names, the number of sources, and whether parser rules arrive as a document or must be inferred from drops. Rates are informational; there is no checkout on this site. See how quotes are built.