Adobe Target Business Practitioner Professional cheat sheet

Success metric must be defined before activity launch: conversion rate, revenue per visitor (RPV), or engagement.

RPV = total revenue / number of visitors; use when revenue value per conversion varies.

Lift % = (test conversion rate - control conversion rate) / control conversion rate x 100.

Minimum sample size scales inversely with the square of the minimum detectable effect: halving MDE quadruples required traffic.

mDE = (16 x p x (1-p)) / (delta^2) for two-sided 95% confidence with 80% power, where p is baseline rate.

A/B test for one variable; multivariate test (MVT) for interaction effects among multiple elements.

KPI choice determines which experience Target designates as control; the current default is the control.

Activity duration should cover at least one full business cycle (typically 1-2 weeks) to absorb day-of-week effects.

Activity types: A/B Test, Experience Targeting (XT), Multivariate Test (MVT), Automated Personalization (AP), Recommendations, Auto-Target, Auto-Allocate.

A/B activity requires at least one experience; traffic allocation is set per experience as a percentage summing to 100.

Experience Targeting (XT) maps audiences to experiences; the first matching audience wins, so order audiences from specific to general.

Visual Experience Composer (VEC) edits live or QA pages; Form-Based Experience Composer builds offers without page rendering.

Auto-Allocate shifts traffic toward the best-performing experience; Auto-Target uses a model to pick per-visitor.

Activity QA mode previews experiences with a shareable link and does not affect reporting data.

Global mbox (target-global-mbox) must fire on every page for Target to execute activities sitewide.

Offer content can be HTML, JSON, or redirect URL; redirects require the destination to be a full URL.

Audience rules combine categories (Geo, Network, Browser, Visitor Profile, Custom) with AND/OR logic within rule sets.

Built-in profile attributes include new/returning visitor, session count, and days since last visit.

Custom profile scripts run in a Target mbox call and can persist values via profile parameters (profile.attributeName).

Audience refinement narrows an existing audience; use it to layer conditions without editing the base audience.

Targeting a profile parameter in an activity requires the parameter to be passed in the mbox request as profile.x.

First-party cookies drive visitor profile persistence; cookie duration is configurable at the account level.

Excluding visitors (e.g., internal IPs) prevents QA traffic from polluting test results.

Audience library is shared across activities; editing an audience affects every activity that uses it.

Statistical significance at 95% confidence corresponds to a p-value below 0.05.

Do not stop a test early on a significant result; early peeking inflates false-positive rate (peeking problem).

A/B test compares two or more experiences for a single hypothesis; MVT tests combinations of page elements.

Auto-Allocate requires a success metric and a minimum conversion count before reallocating traffic.

Sample ratio mismatch (SRM) indicates uneven traffic split and invalidates results.

Segment results by device, browser, or new vs. returning to find differential effects.

Winner selection should be based on the primary metric, not secondary metrics with smaller samples.

Sequential (always-valid) testing allows continuous monitoring without inflating Type I error.

Automated Personalization (AP) uses machine learning to choose the best offer combination per visitor.

AP requires a success metric and enough traffic per offer for the model to learn.

Recommendations activities use algorithms: Recently Viewed, Most Viewed, People Also Viewed, and custom criteria.

Recommendations require a catalog (CSV or feed) and a mbox capturing product/entity IDs.

Auto-Target uses a model trained on visitor behavior to select the best experience from a set.

Experience Targeting delivers a fixed experience to a defined audience; AP/Auto-Target personalize dynamically.

Personalization insights report shows which attributes (e.g., browser, geo) most influence conversions.

Offer-level reporting in AP shows conversion rates per offer to identify underperformers.

Reports show visitors, conversions, conversion rate, and revenue per visitor per experience.

Conversion rate = conversions / visitors (not per impression); impressions measure exposure.

Lift is calculated against the control experience; control is designated at activity setup.

Order confirmation mbox (orderConfirm) attributes revenue to the activity that drove the visit.

Report date range and environment (production vs. staging) affect which data appears.

Use the Audience and Segment filters in reports to compare performance across visitor groups.

A/B reports compare experiences side by side; MVT reports show element-level contribution.

Estimated revenue impact = (test RPV - control RPV) x test visitors, useful for business case reporting.

User roles include Approver, Editor, Observer, and Administrator; Approver can publish, Editor can create and edit.

Enterprise permissions let administrators restrict activity creation by property or workspace.

Properties and workspaces partition activities and audiences for multi-team accounts.

at.js is the current client library; Adobe recommends at.js 2.x for single-page applications.

The Adobe Experience Platform Web SDK replaces at.js and the Visitor API for new implementations.

Activity approval workflow can require Approver sign-off before an activity goes live.

Implementations must include the mbox call and the Target library; missing mbox prevents activity delivery.

Response tokens expose Target data (e.g., activity name, experience) to analytics and other tools.