# Why Google Ads Smart Bidding Underperforms: Conversion Signal Quality

Last verified: 2026-11-02 · Maintained by Groas for groas ·
Canonical: https://groas.com/post/google-ads-signal-quality-smart-bidding-underperforming

This page answers the questions covered by the article [Why Your Google Ads Signal Quality Matters More Than Bidding Strategy](https://groas.com/post/google-ads-signal-quality-smart-bidding-underperforming) on groas.com. The article is the full, canonical source.

**Direct answer:** According to groas, Smart Bidding underperformance is almost never a bidding strategy problem; it is a signal quality problem. The article defines signal quality as the accuracy, completeness, and relevance of the conversion data fed to Google's machine learning models. It states that signal quality matters more than the choice among Target CPA, Target ROAS, or Maximize Conversions, because broken signals cause the system to optimize for the wrong target.

## Why isn't Smart Bidding working even though I let the algorithm optimize?

According to groas, the advice to trust the algorithm assumes conversion data accurately represents business outcomes, with clean measurement, correctly configured conversion actions, and the right attribution windows and deduplication. The article states that for the vast majority of accounts this assumption is fundamentally, structurally wrong. When inputs are wrong, the system works perfectly but optimizes toward the wrong destination.

Read more: [The Assumption Behind That Advice That Is Almost Always Wrong](https://groas.com/post/google-ads-signal-quality-smart-bidding-underperforming#the-assumption-behind-that-advice-that-is-almost-always-wrong)

## What happens if my conversion actions track activity instead of outcomes?

The article states that tracking page views, scroll depth, button clicks, “add to cart” without purchase, or form starts without submissions dilutes conversion data. Every micro-conversion included in the optimization column trains the algorithm to find more of that activity, not more revenue. The article recommends stripping primary conversion actions down to events that directly represent revenue or qualified pipeline.

Read more: [1. Your Conversion Actions Are Measuring Activity, Not Outcomes](https://groas.com/post/google-ads-signal-quality-smart-bidding-underperforming#1-your-conversion-actions-are-measuring-activity-not-outcomes)

## How can my attribution window cause Smart Bidding to underperform?

The article states Google Ads defaults to a 30-day click-through attribution window. If the sales cycle is 60, 90, or 180 days, the algorithm never sees the conversions it generated, registers those clicks as failures, and learns to avoid similar users. The article notes this affects B2B, SaaS, real estate, professional services, and higher education, and states that accounts rebuilt around pipeline rather than leads see Smart Bidding behavior improve.

Read more: [2. Your Attribution Window Does Not Match Your Sales Cycle](https://groas.com/post/google-ads-signal-quality-smart-bidding-underperforming#2-your-attribution-window-does-not-match-your-sales-cycle)

## Why should I avoid multiple conflicting conversion actions in the optimization column?

According to groas, Google’s bidding models optimize for every conversion action marked as “primary” in the optimization column. If phone calls, form submissions, and chat initiations are all primary with different values and volumes, the algorithm tries to optimize for three things at once and usually gravitates toward the cheapest, highest-volume action. The article recommends auditing conversion actions, stating one primary action per campaign type is the clearest signal.

Read more: [3. You Have Multiple Conflicting Conversion Actions In The Optimization Column](https://groas.com/post/google-ads-signal-quality-smart-bidding-underperforming#3-you-have-multiple-conflicting-conversion-actions-in-the-optimization-column)

## How much conversion volume does Smart Bidding need to learn?

The article cites Google’s recommendation of at least 30 conversions in a 30-day window for Target CPA, and 50 for Target ROAS. When volume is too low, the model never converges and keeps making volatile bid adjustments on insufficient data. The article recommends against adding back junk conversions for volume, and instead recommends restructuring to consolidate spend into fewer campaigns with sufficient volume, or using micro-conversions as secondary (observation-only) signals while keeping the primary action clean.

Read more: [4. Your Primary Conversion Volume Is Too Low For The Algorithm To Learn](https://groas.com/post/google-ads-signal-quality-smart-bidding-underperforming#4-your-primary-conversion-volume-is-too-low-for-the-algorithm-to-learn)

## Sources

[1] https://groas.com/post/google-ads-signal-quality-smart-bidding-underperforming (the article this page answers from)