Behavioural and Intent Marketing Data Termimology
PDF download form at the bottom of the page.
Account-Level vs User-Level Intent
Summary: Distinguishes signals aggregated to an organization (account) from signals attributable to individual users.
Use Cases:
-
Use account-level spikes for ABM
-
Respect privacy by activating at cohort/account level
Anomaly Detection & Outlier Suppression
Summary: Techniques to detect and dampen unnatural spikes from crawlers, syndication bursts, or tagging errors.
Use Cases:
-
Apply IQR/z-score rules to cap extremes
-
Flag broken tags when spikes exceed historical bounds
Attribution Window
Summary: Look-back period during which outcomes (conversions, visits) can be credited to prior exposures or interactions.
Use Cases:
-
Choose 7/14/30-day windows to align with buying cycle
-
Compare brand vs. performance campaigns under different windows
Baseline & Seasonality Normalization
Summary: Adjustments that account for expected traffic cycles so surges reflect genuine interest, not seasonality.
Use Cases:
-
Use 13-week rolling baselines
-
Remove holiday effects before scoring
Bot/IVT Filtering
Summary: Identification and removal of non-human or invalid traffic (spoofing, automation, data center IPs) from behavioral datasets.
Use Cases:
-
Exclude bot traffic from surge calculations
-
Protect downstream attribution and modeling from IVT bias
Cohorting (Time/Behavior)
Summary: Grouping users/accounts by shared start dates or behaviors to analyze trajectories consistently.
Use Cases:
-
Measure post-exposure engagement by cohort
-
Compare weekly cohorts for campaign lift
Consent Signals (GPP/TCF)
Summary: Framework strings that communicate user consent and regional requirements for data use.
Use Cases:
-
Honor per-purpose consent in scoring pipelines
-
Drop or anonymize events without valid consent
Content Affinity
Summary: Statistical relationship between audiences and topics/categories based on observed consumption patterns.
Use Cases:
-
Build topic-based audiences for prospecting
-
Prioritize sponsorships around high-affinity themes
Context Classification (NLP)
Summary: Use of NLP/ML to classify content, entities, and sentiment for topic and intent inference.
Use Cases:
-
Map articles to IAB/GARM taxonomies
-
Exclude sensitive contexts from intent models
Data Retention / TTL
Summary: Policies that define how long raw events and derived intent scores are stored and considered valid.
Use Cases:
-
Expire intent after 30–90 days unless renewed
-
Define TTLs per region and purpose
De-anonymization Guardrails
Summary: Policies and controls that prevent re-identification of individuals from aggregated or cohort-level behaviors.
Use Cases:
-
Apply k-anonymity thresholds before delivery
-
Only expose cohort metrics above minimum size
Deduplication & Sessionization
Summary: Combining multiple events into sessions and removing duplicates to avoid inflated engagement.
Use Cases:
-
Stitch rapid repeat events into one session
-
Drop duplicate pings from auto-refresh
Event Stream
Summary: Time-stamped user actions (page views, clicks, searches, plays) captured for analytics and activation.
Use Cases:
-
Power real-time personalization with event feeds
-
Trigger scoring when users view key content
Identity Stitching (Cross-Device & Anonymous IDs)
Summary: Resolving anonymous IDs, cookies, and device identifiers into a stable person/household/account entity.
Use Cases:
-
De-duplicate reach across devices
-
Tie onsite behavior to account-level surges
Intent Qualification & Thresholds
Summary: Rules that define when a score is high enough to count as ‘in-market’ for activation.
Use Cases:
-
Set topic thresholds by segment size
-
Avoid false positives with minimum persistence rules
Noise Filtering & Smoothing
Summary: Moving averages and filters that reduce volatility and noise in short time-series.
Use Cases:
-
Use 7-day MA before surge detection
-
Apply EWMA to stabilize daily scores
Onsite vs Offsite Intent
Summary: Differentiates first-party behavior on owned properties from third-party signals across the web.
Use Cases:
-
Prioritize onsite for high precision
-
Use offsite to expand early-funnel discovery
Panel vs Census Behavioral Data
Summary: Panel = sampled users with weights; Census = near-exhaustive event coverage.
Use Cases:
-
Blend panels to calibrate census bias
-
Use census for activation, panels for insight
Pathing & Sequence Modeling
Summary: Analyzing event order (Markov paths, sequence models) to understand journeys and predictors.
Use Cases:
-
Identify paths that precede conversions
-
Inform content sequencing and retargeting
Recency/Decay
Summary: Time-weighted logic that reduces the value of older behaviors relative to fresh signals.
Use Cases:
-
Use half-life decay in intent scoring
-
Down-weight seasonally stale topics
Sampling & Weighting
Summary: Downsampling and statistical weights used to manage volume and bias in large streams.
Use Cases:
-
Keep real-time pipelines performant
-
Re-weight oversampled sources to population share
Scroll Depth & Time-in-Content
Summary: Attention proxies measuring how far users scroll and time actively spent with content.
Use Cases:
-
Filter out bounced visits from scoring
-
Optimize placements to content with deeper engagement
Source Quality Scoring
Summary: Per-source or per-publisher quality metrics based on fraud risk, coverage, freshness, and noise.
Use Cases:
-
Route budgets to high-quality sources
-
Exclude sources with persistent IVT flags
Supply Path Transparency
Summary: Documentation of where behavioral signals originate and how they are transformed, joined, and scored.
Use Cases:
-
Prefer sources with auditable provenance
-
Compare vendor pipelines to assess risk/quality
Surge Scoring
Summary: Relative increase in topic consumption versus a baseline for an account, cohort, or region.
Use Cases:
-
Alert SDRs when an account spikes on a topic
-
Shift budget to surging themes mid-flight
Taxonomy Governance & Versioning
Summary: Processes to evolve topic definitions while maintaining backward-compatible reporting.
Use Cases:
-
Version topics quarterly with changelogs
-
Map old topics to new for trend continuity
Time-on-Page (Active Time)
Summary: Measures of engaged time that exclude idle background tabs and inactive windows.
Use Cases:
-
Favor active-time over raw dwell
-
Detect autoplay or idle inflation
Topic Taxonomy
Summary: Controlled vocabulary that maps content into standardized topics for intent modeling and reporting.
Use Cases:
-
Normalize disparate publisher tags
-
Roll up granular topics into reportable themes
Unique Users & Identity Keys
Summary: Definition of uniqueness and the identifiers used (cookies, MAIDs, logins) for counting and capping.
Use Cases:
-
Report reach using a stable key hierarchy
-
Prevent double-counting across domains/apps
Viewability Gate / Exposure Eligibility
Summary: Rules that only count exposures that met minimum viewability before attributing outcomes.
Use Cases:
-
Require 50%/1s for display; 2s for video
-
Reduce bias from non-viewable impressions
Windowing (Event vs Processing Time)
Summary: Streaming concepts distinguishing when an event happened from when it was processed.
Use Cases:
-
Use event-time windows for accuracy
-
Handle late events with watermarks
