You check your attribution model by navigating to Tools & Settings in the top right of Google Ads, selecting Conversions under the Measurement section, then clicking into any specific conversion action. The attribution model is listed in the conversion settings, typically defaulting to Last click for most accounts unless you've changed it. Google Ads offers six attribution models: Last click (credits the final ad interaction), First click (credits the first), Linear (spreads credit evenly), Time decay (more credit to recent clicks), Position-based (40% to first and last, 20% distributed to middle), and Data-driven (uses machine learning, available only with sufficient conversion volume). Data-driven requires at least 3,000 ad interactions and 300 conversions within 30 days per conversion action, so most smaller accounts stick with rule-based models. If you want to compare how different models would perform before committing, use the Attribution reports under Tools & Settings > Measurement > Attribution. This shows how conversions would be credited under each model using your actual campaign data. It's read-only analysis, not a setting change, which makes it safe for testing assumptions. Ottawa SEO typically runs Last click for lead gen clients who care about bottom-funnel performance and switch to Data-driven once accounts hit the volume threshold, usually around $8,000–$12,000 monthly spend in competitive verticals. For ecommerce with longer consideration cycles, Position-based makes sense to credit both discovery and closing interactions. The model you choose directly affects conversion counts in your reports and influences automated bidding decisions, so changing it mid-campaign can cause temporary performance swings while Smart Bidding recalibrates. One thing to watch: conversion action settings override campaign settings. You set attribution per conversion action, not globally, so if you track multiple goals like form fills and phone calls, each can use a different model. Check all your conversion actions individually rather than assuming one setting applies everywhere.