Lead generation accounts fail in one predictable way: the ad platform gets told that every form fill is good, so it becomes very efficient at finding more low-quality forms.
The fix is not mystical. You need a measurement stack that tells Google which leads actually matter — and that means moving significantly beyond basic thank-you-page conversion tracking.
What enhanced conversions for leads does
Enhanced conversions for leads improves how Google matches ad interactions to later lead outcomes by using first-party data more durably. It is a better setup than brittle workflows that stop at the form submit and rely on cookie-based matching that degrades over time.
For many lead gen advertisers, it should be treated as baseline infrastructure — not an advanced add-on. The setup effort is modest; the measurement improvement is significant.
Why basic form tracking creates fake efficiency
If you only import "form submitted," the bidder can optimise toward:
- –Spam submissions
- –Poor-fit leads who match traffic patterns but not buyer intent
- –Uncontactable leads with invalid contact details
- –Students, job seekers, or competitors filling forms
- –Accidental submissions from curious users with no intent to buy
That creates what looks like efficiency in the platform — low CPL, high conversion volume — while sales sees no improvement in pipeline.
The right lead quality stack
- Primary web conversion tracking (form submission event)
- Enhanced conversions for leads
- CRM integration or reliable offline conversion import
- A clear lead stage map tied to real business events
- Proxy values or weighted lead scoring for delayed revenue
- Feedback loops tied to qualified pipeline or closed revenue
What lead scoring should reflect
A useful scoring model should include some mix of: fit (industry, company size, geography), sales-readiness signals, product or service match, and actual downstream stage progression. Where possible, tie it to eventual revenue.
Do not build a scoring model that only reflects surface engagement — opens, clicks, or page views. Those signals are too weak to meaningfully guide Smart Bidding.
Real-world example
A legal services advertiser receives 200 leads in a month. On paper that looks fine. But after sales review:
- –60 are spam or bots
- –70 are outside the service area
- –35 are the wrong matter type
- –20 are uncontactable
- –15 are real consultations worth booking
If Google is optimising to all 200, the account is being trained on the wrong target. If qualified consultations are scored and imported back, bidding starts moving toward the right intent.
Implementation checklist
Step 1: audit the current conversion map
List every event currently being used as a Google Ads conversion action and rank them by actual business value. You may find low-value events — page views, time on site, micro-engagements — being treated as primary conversions.
Step 2: define the true optimisation event
Pick one event or a weighted set of events tied to real commercial outcomes. In most accounts, this means a qualified lead or a downstream CRM stage — not just the initial form fill.
Step 3: connect CRM stages
Map at minimum: lead, marketing qualified lead, sales accepted lead, opportunity, and customer. The further downstream you can import data, the better the model learns.
Step 4: assign values
If revenue is delayed, use proxy values based on historical close rates and average deal size. Even rough values are better than treating every conversion as equal.
Step 5: import consistently
Bad imports create bad model behaviour. Data hygiene matters — deduplicate, validate contact matching, and exclude test records.