Search Console, GA4, task research, or business records?

Four evidence sources answer four different questions

Search Console describes how a site appeared and received clicks in Google Search under its reporting rules. GA4 describes measured sessions and events after entry under its configuration. Task research observes whether representative people can complete a defined information job. Business records capture confirmed operational outcomes such as subscriptions, support cases, refunds, or commissions under separate attribution rules. Compare them with an evidence-source matrix containing observation stage, unit, coverage gaps, latency, segmentation, explanatory depth, manipulation risk, and decision fit. None is the ‘single source of truth’ for the entire organic journey. GOV.UK’s success-measure discipline supports choosing measures from the outcome, while the tools’ own documentation supplies counting boundaries.

Evidence: Google Search Console Help; Google Analytics Help

Set the matrix around one unanswered question

The matrix should identify the missing evidence for a live decision, not reward the source with the most polished dashboard.

Evidence: Google Search Console Help; GOV.UK Service Manual

Search Console: closest to search exposure, bounded after the click

Use Search Console when the question concerns impressions, clicks, query-page relationships, devices, countries, search appearance, or changes in reported average position. It is valuable for identifying which pages and queries moved and for separating exposure from later behavior. Its limits include aggregation, privacy filtering, data delay, and definitions that do not map one-to-one to analytics sessions. It cannot tell whether a reader completed the page’s task or made a purchase. In the matrix, it scores high for search-stage diagnosis and low for on-site explanation. Preserve filters and date ranges, and avoid reading average position as a stable rank experienced by every user.

A search observation should generate a hypothesis for another source when the decision depends on comprehension or business outcome.

Evidence: Google Search Console Help; Google Search Console Help

GA4: configurable landing behavior with implementation dependence

Use GA4 when the question concerns measured entry pages, session behavior, defined events, continuation paths, and configured attribution. Its flexibility is also its vulnerability: consent, tags, filters, URL normalization, cross-domain setup, event triggers, and model changes alter what the report means. The landing-page dimension helps connect an entry with events, but it does not recreate Search Console clicks or reveal unmeasured users. In the matrix, GA4 scores high when event definitions are stable and tied to a reader job. It scores poorly when the implementation changed or the team relies on generic engagement as a quality verdict.

Attach the event specification to every decision. A chart without its trigger and exclusions is not reproducible evidence.

Evidence: Google Search Console Help; Google Analytics Help

Task research: strongest for reasons, weakest for population estimates

Use task research when the question is whether people understand a result promise, locate evidence, distinguish options, or choose an appropriate next step. A small session can reveal why a route or answer fails when analytics shows only an exit. Realistic prompts, representative recruitment, neutral facilitation, and preservation of contradictory cases matter. The source does not estimate market demand or causal effect from a few participants. In the matrix, it scores high for comprehension and accessibility insight, with higher cost and bounded coverage. GOV.UK’s usability and success-measure practice supports connecting tasks to service outcomes rather than showcasing an interface.

Combine task findings with platform segments when deciding how widespread a problem may be, but keep the evidence types visibly separate.

Evidence: Google Analytics Help; GOV.UK Service Manual

Business records: confirmed operations with attribution distance

Use business records when the decision depends on a completed operational event: confirmed subscription, merchant approval, refund, commission, support escalation, or documented rejection. These records may be closest to value and risk, but they often contain sparse volume, delayed reversals, channel attribution rules, and factors outside the content. An affiliate commission credited to a last click cannot prove that one page created the decision; a refund can reveal product mismatch without identifying the original search promise. In the matrix, business records score high for outcome confirmation and low for diagnosing earlier reader experience by themselves.

Define verification status, reversal window, and ownership. Do not promote provisional or model-assigned revenue as a settled content result.

Evidence: Google Search Console Help; Google Search Console Help; Google Analytics Help

Choose the source nearest the unresolved decision

Fill one matrix row for the decision in front of the team. If exposure changed, begin with Search Console; if measured landing behavior diverged, inspect GA4 definitions; if the answer appears confusing, observe tasks; if value or harm must be confirmed, reconcile business records. The next action is to select one primary source and one countercheck, then state what neither can prove. Limits remain for every option: missing data, latency, configuration, recruitment, and attribution. Do not merge incompatible units into an opaque composite score. Stop collecting when the evidence supports a reversible choice or shows the decision is not worth further cost. Revisit the source choice when definitions or risk change. The comparison succeeds when it prevents a dashboard from answering a question it was never designed to address.