Attribution Model Template for Product Analytics

Start with position-based (40% first-touch, 40% last-touch, 20% distributed across middle touches). It gives appropriate weight to both the channel that introduced the user and the channel that converted them, without ignoring middle touches. Compare it to last-touch as a baseline. The [RICE Calculator](/tools/rice-calculator) can help you prioritize which attribution improvements to tackle first if you have limited analytics engineering resources.

For B2B SaaS, 30 days is a good default. Enterprise products with longer sales cycles may need 60-90 days. Consumer products often use 7-14 days. Set the window to match your typical time-to-conversion. Pull a distribution of days-to-convert from your data and set the window at the 90th percentile.

Cross-device attribution requires user identity resolution (usually through login events). If a user visits on mobile, then converts on desktop, you need a shared user ID to connect the sessions. Most analytics platforms support this via authenticated user tracking. If you do not have login-based tracking, your attribution will undercount multi-device journeys and overcredit single-device ones.

Slack messages, private Discord channels, and DMs do not pass referrer headers. This traffic shows up as "direct" in analytics. Accept that 20-40% of your traffic may be unattributable. Instead of trying to track everything, focus on making your trackable channels consistent. Survey new signups ("How did you hear about us?") to supplement analytics data.

Yes. In-app referrals, sharing features, and viral loops are channels that deserve the same attribution treatment as paid and organic. Tag in-app sharing events with UTM parameters or custom referral codes so they appear in your attribution model alongside marketing channels. ---