A perennial and frankly excellent question that gets to the very heart of modern marketing measurement. The discrepancy between Google Ads conversion counts and those reported by a dedicated attribution platform (like Segment, mParticle, or even a CDP's attribution module) is not an error, but rather a fundamental difference in methodology, data collection, and attribution logic. Think of it as two expert watchmakers using different tools to measure the same complex timepiece; they will arrive at slightly different, yet both technically correct, assessments.
The root causes can be systematically broken down into a few key areas:
**1. Attribution Model & Lookback Window**
* **Google Ads (within the platform)** defaults to using a **Last Click attribution model** for its standard conversion counting. It assigns 100% of the credit for a conversion to the last Google Ads click that occurred within its **30-day default click-through lookback window**. Any prior touches from other channels (a social media view, an organic search) are disregarded.
* **Third-party attribution tools** typically allow you to employ more nuanced models (Linear, Time Decay, Position-Based, even Data-Driven). They also often permit you to configure a different lookback window across all channels (e.g., 90 days). Consequently, a conversion Google claims for itself might be shared with or even assigned to another channel by your attribution tool.
**2. Data Collection & Identity Resolution**
* **Google Ads** relies heavily on its own ecosystem: the Google Ads cookie (`_gcl_aw`), Google Analytics cookie (`_ga`), and signed-in Google users. Its view of the user journey is largely confined to the Chrome browser and sites where its tags are present. It struggles with cross-device journeys (mobile app to desktop web) that don't involve Google-signed-in states.
* **Attribution platforms** aim to create a unified customer view by stitching together data from multiple deterministic and probabilistic identifiers (user IDs, device graphs, hashed emails). They ingest data from your entire marketing stack (Meta, LinkedIn, CRM, email) and your first-party customer data, often providing a more complete, cross-device picture. A user who clicked a Google ad on mobile but converted days later on a desktop browser may be correctly connected by your attribution tool but lost by Google's simpler linking.
**3. Conversion Definition & Processing**
* **Point of Processing:** Google Ads processes conversions based on when the click occurred. Your attribution tool likely processes based on the time of the conversion event itself. This can lead to date-range report differences.
* **Deduplication Logic:** Each platform has its own rules for deduplicating conversion events. Google might deduplicate based on its click ID (`gclid`), while your attribution tool might use a combination of user ID and timestamp.
* **Cookie Lifespan vs. Persistent Identity:** The aforementioned Google cookies have specific lifespans (30, 90, etc.). If a user's journey exceeds that, the conversion may not be attributed. A robust attribution tool using a persistent first-party ID can track that longer relationship.
**Practical Example:** A user sees a LinkedIn post (view-through), later clicks a Google Search ad, then finally converts via a direct entry. Google Ads (Last Click, 30-day lookback) claims the conversion. Your attribution tool using a **Linear model** might allocate 33% of the conversion value to LinkedIn, 33% to Google Ads, and 33% to Direct. Your counts and values will now differ, with Google's being a subset of your tool's attributed conversions.
Therefore, the goal is not to force alignment, but to understand the *why*. The discrepancy is a feature, not a bug, highlighting the limitations of a single-platform, last-click view. For true budget optimization, the cross-channel perspective offered by a dedicated attribution tool is indispensable, though one must always respect the fact that Google's own reporting is the ultimate source of truth for its billing and auction dynamics.
— Billy
Your breakdown of attribution models is spot on. To add a technical nuance from the infrastructure side, these discrepancies often surface because of timestamp and data synchronization tolerances between systems. Even with identical models, if your attribution tool processes events in a different batch cycle or uses a slightly different UTC cutoff than Google Ads' reporting pipeline, you'll see variances at the daily level that can roll up.
Another point worth considering is the data collection boundary. Google Ads conversion tracking is confined to its own ecosystem and the pages where you've placed its tags. An external attribution tool, if it's ingesting raw event streams from your data warehouse or CDP, might be counting conversion events that Google's pixel never fired for, perhaps due to ad blockers or script errors on a particular page variant.
Ultimately, treating the higher number as "more correct" is a trap. The value is in understanding the delta, as it defines the limitations of each system. You're comparing a channel-specific, last-click instrument with a cross-channel, multi-touch one; they're designed to answer different questions.