# Break-map record. One record per incident. Copy, fill, keep with the evidence. # Verdicts: pass, fail, not_checked, not_applicable. A verdict without evidence is not_checked. # Never paste personal data, credentials or tokens into this file. incident: "" # short name, e.g. "GA4 sign_up drop after signup form rebuild" opened: "" # YYYY-MM-DD owner: "" # person running the investigation question: complaint_as_stated: "" # what the person said, word for word report: "" # report, exploration, dashboard tile or query metric: "" # e.g. "event count, sign_up" scope: "" # user, session or event scope; key event or not date_range: "" # completed days only timezone: "" # of the report and of the independent source filters: "" # report filters, segments, comparisons independent_source: "" # system that commits the outcome, and its query or export expected_ratio: "" # baseline band of analytics / independent, previous 28 comparable days (same weekday type, holidays left out) observed_ratio: "" # same ratio for the incident days first_change_date: "" # first of 2 consecutive completed days outside the band, or of 3 out of 5; a single day is "watch" sizing: direction: "" # under or over (above the band, or above 100% for a once-per-outcome event) expected_from_baseline: "" # independent count x baseline ratio, per day observed: "" # analytics count per day gap: "" # absolute gap per day: expected minus observed (under), observed minus expected (over) unique_id_coverage: "" # events with an outcome ID: distinct analytics IDs found in the backend / backend count, baseline and incident events_per_id: "" # analytics events / distinct IDs found, baseline and incident; above baseline or coverage down with the ratio steady = offsetting errors segments_checked: [] # e.g. ["method", "page_path", "browser", "consent state", "region", "channel"] where_the_gap_sits: "" # the segment(s) holding the gap, with numbers guards: "" # mix shift, independent-source definition, volume per bucket, recent days still processing (GA4 48 hours, Google Ads conversion lag) containment: # drafted when the gap is outside the band for 2+ completed days and the conversion feeds bidding, audiences or budget rules; not a fix proposed_on: "" # YYYY-MM-DD, usually the day of sizing, before the proof actions: [] # e.g. data exclusion (click dates, account time zone, start covering conversion delay, fixed end), provisional annotations, automations to pause approved_by: "" # owner of each action; nothing is applied by the agent test_traffic: test_identities: "" # pattern with a backend test flag, e.g. qa-signup-NN@example.com, one fresh identity per run, excluded from both sides ad_hosts_blocked: "" # ad conversion hosts blocked in the test browser, unless the ad platform is the layer under test developer_filter_state: "" # Active, Testing, Inactive or none, read before the first test test_event_log: [] # time (with timezone), journey and test identity of each run; required if the filter is not Active changes_in_window: [] # releases, web and server container versions with publish times, container upgrades, CMP and vendor scripts, CDN and gateway routes, ads, GA4 property change history, export, model, GTM and GA4 release notes journey_tested: "" # one journey, one event, one destination comparison_journey: "" # a journey that should pass at the suspected layer layers: - layer: link_and_redirect verdict: "" evidence: "" - layer: consent verdict: "" evidence: "" - layer: tag_manager verdict: "" evidence: "" - layer: browser_request verdict: "" evidence: "" - layer: server_container verdict: "" evidence: "" - layer: platform_processing verdict: "" evidence: "" - layer: export verdict: "" evidence: "" - layer: warehouse_model verdict: "" evidence: "" - layer: report verdict: "" evidence: "" first_failing_layer: layer: "" layers_above_all_pass: "" # yes or no, with the evidence references comparison_journey_passes: "" # yes or no, with evidence share_of_gap_explained: "" # at least 80% of the absolute gap, otherwise keep walking; "not measurable" with no independent count repeated_runs: "" # with no independent count: failing journey fails 3 of 3 across two browsers, comparison passes 3 of 3 cause: "" # the mechanism, in one sentence fix: owner: "" change_draft: "" # link to the diff, change list or version note approved_by: "" approved_on: "" downstream_changes_avoided: "" # what was proposed downstream and not done retest: journey_rerun_on: "" layer_results: "" # pass at every layer to the report ratio_after: "" # at least 2 completed days after the processing window inside_baseline_band: "" # yes or no gap_statement: untracked_window: "" # from and to, with timezone estimated_missing: "" # analytics events, not customers affected_destinations: [] # reports, audiences, ad platforms bidding on imported conversions backfill_decision: "" # usually none for analytics; Google advises against ad conversion backfill, and any approved one waits 5+ days after the exclusion ad_data_exclusion_proposed: "" # campaigns, click-date window in the account time zone, applied date; see containment annotations_proposed: [] context_updates_proposed: [] # martech-context/ entries proposed, not written: decisions.md, tracking-plan/, baseline band