Unit
- Attribution definition: Marketing attribution assigns credit to touchpoints along the customer journey that influence a conversion. Without attribution, marketers cannot know which channels or campaigns drive results, leading to misallocated budgets and missed opportunities.
- Touchpoint types: A touchpoint is any interaction a prospect has with a brand, such as a display ad view, email open, website visit, or paid search click. Each touchpoint plays a role in moving the prospect toward a conversion, but the influence varies by timing and context.
- Single-touch models: These models assign 100 percent of conversion credit to one touchpoint. First-click gives all credit to the initial interaction, while last-click credits the final touchpoint before conversion. They are simple but ignore the full journey.
- Multi-touch models: These distribute credit across multiple touchpoints, offering a more balanced view. Common examples include linear (equal credit), time-decay (more credit to recent), and position-based (40-20-40 split). Multi-touch models require robust data tracking.
- Attribution window: The time period during which touchpoints are considered for attribution. Common windows range from 7 to 90 days. Shorter windows favor recent touchpoints; longer windows include earlier interactions that may have built awareness.
- Cross-device attribution: Customers interact across devices, creating challenges in stitching together a single journey. Solutions include deterministic matching (login-based) and probabilistic matching (behavioral patterns), but privacy regulations increasingly limit these approaches.
- Conversion types: Not all conversions are equal. Micro-conversions (email signups, content downloads) indicate interest; macro-conversions (purchases, subscriptions) represent primary goals. Attribution models can be tuned for different conversion values.
- Fraud and noise: Bot traffic, viewability issues, and ad fraud can distort attribution data. Marketers must filter out invalid interactions and apply minimum engagement thresholds to ensure only genuine touchpoints receive credit.
- Decay factor: In time-decay models, the decay factor determines how quickly credit diminishes for older touchpoints. A linear decay reduces credit proportionally with time, while exponential decay drops faster. Choosing the right decay factor aligns attribution with actual customer behavior and purchase cycles.
- Recency over frequency: Time-decay attribution prioritizes recency over frequency. A touchpoint that occurs close to conversion gets more credit, even if earlier touchpoints were more numerous. This model works well for short sales cycles where decision momentum builds quickly.
- Model calibration: To implement time-decay effectively, teams must calibrate the model using historical conversion data. Testing different decay curves (e.g., 7-day half-life vs. 30-day half-life) helps identify which weighting best predicts real outcomes. Regular recalibration ensures the model stays accurate as customer behavior shifts.
Campaign Case Study
Campaign: Latitude SaaS Free Trial (fictional)
Objective: Increase free trial signups for a project management tool by 30 percent in Q3.
Target Audience: Mid-market operations managers aged 30-45, tech-savvy, located in North America.
Channels: LinkedIn ads, Google search (non-branded), industry blog partnerships, email nurture.
Results: Last-click attribution showed search as the primary driver (70 percent of signups). However, multi-touch analysis revealed that LinkedIn content and blog posts initiated 60 percent of journeys, earning them significant assist value. The team reallocated 20 percent of search budget to top-of-funnel content, improving overall conversion rate by 18 percent.
Quick Revision
- Attribution quantifies the contribution of each touchpoint to a conversion, enabling data-driven budget decisions.
- Single-touch models are simple but incomplete; multi-touch models provide a holistic view of the customer journey.
- Attribution windows, cross-device tracking, and conversion definitions critically affect model outcomes.
- Filtering fraud and noise is essential for accurate measurement.