Adobe Analytics Business Practitioner Professional cheat sheet

A KPI requirement is complete only when it specifies metric, dimension breakdowns, time grain, segment scope, and data source.

Business requirements answer what and why; functional requirements answer how the tool implements it (variables, events, rules).

Success criteria should be written as measurable thresholds tied to a baseline period, not as qualitative goals.

Stakeholder interviews should capture decision cadence (daily, weekly, quarterly), because it sets report latency and granularity.

A solution design reference (SDR) is the master document mapping every business question to variables, events, and classifications.

Scope sign-off should list explicit out-of-scope items to prevent mid-project variable creep.

Requirements for calculated metrics must state the numerator, denominator, and how division-by-zero and unattributed traffic are handled.

Acceptance testing criteria should be defined before implementation starts, referencing expected values from the SDR.

Design variables against a naming convention that encodes scope and purpose, e.g. page:type, evt:video:start, prop:internal:campaign.

Choose an eVar allocation model per business need: most recent for content affinity, original value for first-touch acquisition, linear for multi-touch credit.

Set eVar expiration to match the question: visit for in-session behavior, purchase event for conversion attribution, never for persistent user attributes.

Use props for high-cardinality, hit-scoped, pathing-relevant values and listVars for delimited multi-value fields.

Success events should be countable business actions; do not create an event for something derivable from an existing metric.

Merchandising eVars bind a value to a product at the moment of interaction; they require product syntax in the merchandising variable.

Data layer design should push values before the analytics beacon fires, with a documented event-driven trigger contract.

Multi-suite tagging shares one beacon across report suites; suite-specific variables must be documented in the SDR to avoid leakage.

Adobe Experience Platform Launch (Tags) fires the Analytics extension via a rule; the beacon is an image request to the tracking server with query-string parameters.

The s.t() call sends a page view; s.tl() sends a link/event call and requires the linkType argument (o, d, or e).

s.tl() does not increment page views and does not process pageName unless explicitly set and permitted.

The doPlugins function runs before the beacon is assembled, making it the correct place for derived variable logic.

Use s.getQueryParam or s.getValOnce in doPlugins for campaign and referral capture, guarding against repeat counting.

The AppMeasurement library version determines available features; upgrade paths and deprecations belong in the implementation notes.

Debugging uses the Experience Cloud Debugger or browser network panel to inspect the raw query string before it reaches processing.

Cross-domain tracking requires the visitor ID to be passed between domains via the adobe_mc parameter or a shared tracking server configuration.

Traffic variables are limited by report suite settings; exceeding the enabled eVar or event count silently drops data.

Classification rules map raw values to friendly groupings; unclassified values fall into 'None' or 'Unspecified' in reports.

VISTA rules and processing rules alter data at collection; document them because they are invisible in the interface.

Data retention and full-data processing settings determine historical depth; changing them does not retroactively reprocess past hits.

Alerts on sudden metric drops or 'Unspecified' spikes catch broken tagging before reporting cycles.

PII must not be captured in query strings or variables; governance review should include a variable-by-variable PII audit.

IP address and geolocation settings are report-suite level; obfuscation choices affect bot filtering and geo reports.

A variable retirement plan prevents orphaned dimensions; deprecated variables should be disabled only after downstream reports are migrated.

A hit segment evaluates each server call; a visit segment evaluates the whole visit; a visitor segment evaluates the full visitor history.

Segment containers are hierarchical; a visitor container that contains a hit container requires the hit to occur within the visitor scope.

Sequential segments require the defined order of hits or visits and can specify time limits between steps.

Segment stacking combines logic with AND; mutually exclusive segments should be built to partition traffic cleanly.

Segment preview shows estimated population and can validate logic before saving to a shared library.

Fallout analysis uses a defined touchpoint sequence and reports the percentage proceeding between steps.

Attribution IQ applies to dimensions in Analysis Workspace; the chosen model changes credit distribution without changing raw data.

Audience publishing to Experience Cloud requires the report suite to be linked to the organization and an appropriate data source.

Freeform tables support breakdowns up to the row limit; increasing rows beyond the limit requires a different visualization such as a dimension report.

The 'Unspecified' row aggregates values with no classification or no captured value; investigate before excluding it.

Calculated metrics combine existing metrics with operators; they can include segments and are shared via the metric library.

Report suite totals and segment totals can differ due to attribution and data latency; document the expected variance.

Data latency for most dimensions is minutes while some (e.g., geolocation) can lag longer; dashboards should state refresh cadence.

Virtual report suites apply a segment at the suite level and can have their own calendar and processing timezone.

Panels and visualizations inherit date ranges from the panel unless overridden at the visualization level.

Scheduled projects deliver in multiple formats; PDF and CSV have different row and breakdown limits.

Start with a question-to-metric mapping; every insight should trace to a metric, a segment, and a time comparison.

Compare period-over-period and segment-over-segment to separate trend from mix shift.

Anomaly detection flags values outside the expected range based on historical seasonality; it is a signal, not a conclusion.

Contribution analysis ranks dimension values by their share of a change in a metric between two periods.

Correlation between two metrics does not establish causation; validate with a controlled test or a segment-based natural experiment.

Segment overlap analysis reveals whether two segments measure the same population, which affects incremental lift estimates.

Recommendations should state the expected metric impact, the measurement plan, and the owner.

Distinguish data quality issues from genuine performance changes before reporting a decline as a business result.