Optimizing Google Ads toward form fills is the default. It is also, for most lead generation advertisers, a significant mistake. Smart Bidding learns from whatever signal you define as a conversion. If that signal is a form submission, the algorithm will get very good at generating form submissions — regardless of whether the people filling out those forms have any intention of buying.
A CRM feedback loop changes the optimization target. Instead of teaching the algorithm to find people who complete forms, you teach it to find people who become qualified leads, then opportunities, then customers. The mechanics are not simple, but they are repeatable. This guide covers the full implementation.
Why form fills are usually the wrong optimization target
Form fill volume correlates with ad spend. More spend generates more form fills. The problem is that the quality distribution of form fills is not uniform. Some campaigns generate high-volume, low-quality submissions: wrong geography, wrong company size, wrong intent, no budget. Others generate fewer submissions that convert into revenue at a high rate.
When Smart Bidding sees both types as identical "conversions," it will optimize across both equally. The result is an account that produces a predictable cost-per-form-fill while masking the fact that a large proportion of that spend generates zero pipeline. The feedback loop problem is invisible in Google Ads reporting but very visible in the sales team's CRM.
What a feedback loop actually looks like
Which sales stages are worth sending back
MQL
Marketing Qualified Lead is the first filter: a lead that meets basic targeting criteria (company size, industry, geography, job function) and has shown enough engagement to be worth sales attention. Sending MQL signals back into Google Ads gives Smart Bidding an early quality filter that is faster than waiting for downstream outcomes and covers more volume than deeper funnel stages.
SQL
Sales Qualified Lead has been reviewed by a sales representative and confirmed as a genuine prospect worth pursuing. SQL signals have higher quality signal value than MQLs because they reflect actual sales judgment, not just automated scoring. For accounts with sufficient SQL volume (30 or more per month), SQL can be a strong primary conversion signal.
Opportunity
An opportunity is a lead that has entered an active sales process: a meeting has been booked, a proposal is in progress, or a specific deal is being discussed. Opportunity signals are high-quality but lower volume, making them better suited as secondary conversion signals or value multipliers in value-based bidding strategies.
Closed won
The ultimate quality signal. Closed won revenue fed back into Google Ads via offline conversion import with revenue values enables true value-based bidding — the algorithm optimizes toward the characteristics of users who become customers at the highest revenue value. This requires the longest feedback loop but produces the most accurate optimization signal for accounts where it is viable.
How to choose the right primary conversion
| Primary conversion signal | When to use it | Minimum monthly volume needed |
|---|---|---|
| Form fill | Only when no downstream data is available | N/A — not recommended as sole signal |
| MQL | First step beyond form fills, available within 24–48 hours | 30+ MQLs per month |
| SQL | When sales qualification is fast and volume supports it | 30+ SQLs per month |
| Opportunity | High-value accounts with slower sales cycles | 10–20+ per month (use with secondary signal) |
| Closed won | Full value-based bidding where revenue data is available | 10+ per month (use with upstream signal) |
Match quality and data hygiene problems
The most common reason CRM feedback loops fail is not technical — it is data hygiene. GCLID values expire after 90 days, so leads that take longer than that to qualify cannot be imported. GCLID values must be stored correctly in the CRM from the moment of form submission; many CRM setups drop or overwrite this value during lead deduplication or record merging.
Before building the feedback loop, audit the GCLID storage in your CRM: what percentage of leads submitted via Google Ads have a stored GCLID value? If the answer is below 70 percent, fix the data capture first. A feedback loop built on 40 percent GCLID coverage is providing a biased sample to Smart Bidding, not a representative signal.
Lead generation example: B2B services
A B2B software consultancy runs Google Ads for enterprise leads. Their Google Ads account shows 150 conversions per month at a CPL of £85. On investigation, the sales team confirms that fewer than 30 of those 150 leads meet the minimum criteria to be worth pursuing. The rest are wrong company size, wrong industry, or just information-gathering with no buying intent.
After implementing GCLID capture, connecting HubSpot to Google Ads via the native integration, and setting up an MQL conversion action, the account has a feedback loop within six weeks. Smart Bidding initially shows a drop in raw form fill volume as it recalibrates, but after eight weeks the conversion rate from form fill to MQL improves from 20 percent to 34 percent. Total leads drop but qualified pipeline grows.
Lead generation example: high-ticket local services
A home renovation company tracks leads from initial inquiry through to signed contract. The average time from inquiry to signed contract is 21 days. They implement a Zapier trigger that sends a "Signed contract" conversion to Google Ads with the contract value whenever a deal is marked as won in Pipedrive. This takes four months to generate enough signal for Smart Bidding to act on, during which time they also send MQL signals as a secondary conversion. The combination produces a bidding strategy that gradually shifts budget toward the campaigns and keywords that generate the highest-value contracts, not just the highest volume of inquiries.
Ecommerce note: equivalent post-purchase quality signals
Ecommerce accounts have the equivalent challenge when optimizing for first purchase versus lifetime value customers. Sending back repeat purchase signals, high-margin purchase events, or subscription conversion data as offline imports enables value-based bidding that optimizes toward the characteristics of high-LTV customers rather than any first-time buyer.
Implementation roadmap
- Add hidden GCLID field to all Google Ads landing page forms
- Verify CRM is storing GCLID on all new lead records — check coverage rate
- Create conversion actions in Google Ads for each sales stage you will import
- Build the CRM trigger or integration that fires on stage changes and sends GCLID + conversion action to Google Ads
- Run the import for historical data (up to 90 days back) to seed the Smart Bidding model
- Set the downstream conversion action as the primary conversion; set form fill as secondary only
- Allow 6 to 8 weeks before evaluating bidding performance against the new signal
The reporting dashboard to build
- –Form fill volume by campaign
- –MQL rate by campaign (form fills that became MQLs)
- –SQL rate by campaign
- –Opportunity and closed won rate by campaign
- –Revenue or pipeline value attributed per campaign
- –Cost per SQL and cost per closed won (the metrics that actually matter)
This dashboard will often reveal that the campaign producing the most form fills has the worst MQL rate, and that a lower-volume campaign with different targeting produces leads that convert at three times the rate. That is the insight that justifies the entire feedback loop investment.