Attribution Modeling

While GA4’s model is a step forward from last-click, our approach offers several crucial advantages: (1) We analyze non-converting paths to identify journey barriers, (2) We calculate custom conversion windows tailored to your business rather than using fixed windows, (3) We provide flexible brand traffic treatment options beyond just removing direct traffic, (4) We incorporate on-site behavior quality in our model, and (5) We use advanced Markov Chain methodology for more accurate value distribution.

For basic models, we recommend at least 3 months of data with a minimum of 1,000 conversions. For more sophisticated machine learning models, 6+ months of data and 5,000+ conversions will provide greater accuracy. However, we can begin implementation with whatever data you currently have available and refine the model as more data accumulates.

Yes. We can incorporate offline conversion data through various methods including CRM integration, unique tracking codes, or call tracking systems. This allows us to connect offline sales back to the digital touchpoints that influenced them, creating a truly omnichannel attribution view.

Our approach emphasizes first-party data collection, server-side tracking, and statistical modeling to mitigate the impact of cookie restrictions and privacy regulations. We design attribution systems that balance detailed measurement with compliance, focusing on patterns and aggregated insights rather than individual-level tracking where necessary.

While the underlying Markov Chain model continuously improves with new data, we recommend a formal review of your attribution model quarterly to account for changes in marketing channels, customer behavior, and seasonal factors. For businesses with rapid changes in their marketing mix, monthly calibration may be appropriate.

Our model recognizes that direct traffic, branded paid search, and branded organic search often represent the same customer mindset at a similar stage in the purchase journey. We give you the flexibility to either merge these sources (treating them as homogeneous brand traffic) or separate them for more granular analysis. This approach provides clearer evaluation of non-brand marketing efforts by removing potential distortions caused by brand recognition.

Yes. We can push attribution data back to major advertising platforms through API integrations, enabling automated bid management and campaign optimization based on our more accurate attribution model rather than the platform’s native attribution.

Our hybrid model considers not just the traffic source but the quality of engagement during each session. We analyze factors like bounce rates, progression through the purchase funnel, time on site, and specific micro-conversions to weight each session’s contribution to the eventual conversion. This approach prevents giving undue credit to sources that generate low-quality traffic.

Yes. We implement cross-device tracking solutions that connect user journeys across multiple devices and browsers. Our probabilistic matching techniques can identify likely cross-device journeys even when deterministic matching isn’t possible, providing a more complete view of the customer path to purchase.