- When AI Max activates on a lead-gen account, watch for contact rate dropping 6–11 points inside 14 days while the Google Ads CPL dashboard holds flat. That gap is the diagnostic, not CPL.
- AI Max for Search replaces traditional match types with AI-powered query matching (Google Ads Help). If Smart Bidding is trained on form-fills, it will buy researchers who fill forms but won’t pick up the phone.
- The fix is a 4-lever conversion-value rebuild: form-fill at $1, contacted lead at $25–40, sold-customer at true CAC ceiling (gross profit × close rate), offline conversions pushed daily.
- Smart Bidding typically needs around 2 conversion cycles or roughly 50 conversions to recalibrate after a bid-strategy change (per Google Ads Help). Skip the rebuild and cost-per-qualified-lead drifts upward while your dashboard looks fine.
- Your Google rep will not bring this up. The in-platform CPL view makes AI Max look free.
If your sales team is complaining about lead quality and your Google Ads dashboard says CPL is steady, the problem is usually the same one. AI Max turned on, kept the CPL number happy, and started training Smart Bidding on people who fill out forms but never answer the phone. Nobody warned you about this at onboarding.
The pattern on lead-buyer accounts is consistent. Form-fill conversion rate stays flat or ticks up. Contact rate, the percentage of submitted leads your team actually reaches, drops noticeably inside two weeks. CPL on the dashboard looks fine for the first few weeks. Then cost-per-qualified-lead quietly climbs while your training set fills up with researchers.
The rest of this piece is the operator-level fix: the diagnostic to pull tomorrow, the four conversion-value levers that re-anchor Smart Bidding to revenue, and the calibration ranges by vertical so you do not have to guess at the numbers.
AI Max Is Not a Match-Type Update. It’s a Conversion-Quality Event.
AI Max for Search replaces traditional match-type logic with AI-powered query matching, keywordless targeting, and asset optimization. Google’s own documentation describes it as a suite of features that uses Google AI to expand reach beyond your keyword list and optimize assets to user intent (Google Ads Help: About AI Max for Search). In plain English: the system decides which queries are similar enough to your business based on what Smart Bidding thinks is converting, not based on the words in your keyword list.
That sentence is the whole problem. If your bidding model is trained on form-fills, AI Max will go find more form-fills. Researchers fill forms. Comparison shoppers fill forms. People three months away from buying anything fill forms. The expanded matching is doing exactly what it’s told. It’s being told the wrong thing.
Contrast this with legacy broad match. Broad match still anchored on the keyword’s lexical core and used user signals as a modifier. AI Max relaxes that anchor. Search themes and brand controls become the constraint layer, and most lead-gen accounts have neither configured properly when migration completes. If you are mid-migration from DSA, our DSA retirement playbook covers the pre-flight audit.
The Form-Fill-to-Contact-Rate Gap Is Your Earliest Diagnostic
Stop watching CPL. It will lie to you for three weeks.
Watch this instead:
The math is unglamorous. Pull two numbers from the same date range:
- Form-fill conversion rate from Google Ads (conversions ÷ clicks on the campaigns running AI Max).
- Contact rate from your CRM or dialer (contacted leads ÷ submitted leads from those same campaigns).
If form-fill rate is steady and contact rate has dropped meaningfully, AI Max is pulling traffic that converts on-platform but evaporates downstream. The threshold for action is any drop you can attribute to the AI Max activation date rather than a seasonal or operational change.
How to pull the gap in under 30 minutes
- In Google Ads, segment your AI Max campaigns by week for the four weeks before and four weeks after activation. Export click and conversion data.
- In your CRM, pull lead submissions and contacted-status leads for the same weeks, filtered to the same source campaigns. UTMs or a GCLID join works.
- Compute contact rate (contacted ÷ submitted) weekly. Plot it against form-fill conversion rate.
- The break point will be visible. If it isn’t, AI Max is not your problem and you should look at your dialer, list age, or call-center capacity instead. Our outbound contact rate diagnostic covers the non-AI-Max causes.
If you have already wired offline conversions properly, this is a 10-minute pull. If you haven’t, our Enhanced Conversions for Leads troubleshooting walkthrough covers the upstream plumbing.
Why ecommerce AI Max playbooks don’t transfer
Every AI Max guide written for ecommerce tells you to trust the system and feed it value data. That advice is correct for ecommerce because the conversion is the revenue. For lead-gen, the conversion is a form, and the form is not the revenue. Follow ecommerce advice on a lead-gen account and you accelerate the problem.
The 4-Lever Conversion-Value Rebuild That Re-Anchors Smart Bidding to Revenue
This is the core fix. Four levers, in this order, with specific values. Skip one and the system finds a new way to optimize toward the wrong thing.
Lever 1: Form-fill at $1 (the volume-chasing kill switch)
Open your conversion actions. Find the form-fill action that Smart Bidding has been optimizing toward. Set its conversion value to $1.
Not zero. Zero removes the signal and Smart Bidding loses its event volume floor. $1 keeps the action live as a signal source but strips its weight in any value-based bid strategy like tCPA, tROAS, or Maximize Conversion Value (Google’s automated bid strategies that optimize toward a target cost or revenue ratio, per Google Ads Help).
This is the lever everyone resists. Your Google rep will tell you Smart Bidding needs more signal. What it needs is the right signal. Form-fills aren’t it.
Lever 2: Contacted lead at $25–$40 via offline conversion import
Create a new conversion action: Contacted Lead. Import it via offline conversion import (OCI) when your CRM marks a lead as reached and verified by a human (Google Ads Help: Upload conversions).
A practical starting range for the conversion value, anchored to typical contacted-lead economics in each vertical:
- Insurance (Medicare, ACA): $25–$30
- Final expense: $30–$35
- HVAC, plumbing, electrical, roofing: $35–$45
- Mortgage, B2B: $40–$60
These numbers do not need to equal the literal economic value of a contacted lead. They need to be high enough relative to $1 that Smart Bidding learns the contacted-lead query patterns are substantially more valuable than form-fill-only patterns. That ratio is what matters. Calibrate to your own funnel math, not the table.
Lever 3: Sold customer at true CAC ceiling, not gross order value
Third conversion action: Sold Policy or Booked Job or whatever your revenue event is. Import via OCI when the sale closes.
Value = gross profit per customer × close rate from contacted lead.
Not gross order value. Not LTV. Not whatever the CFO wants on the slide. The true ceiling Smart Bidding can profitably pay to acquire that event.
Example: HVAC install with $2,800 gross profit and a 22% close rate from contacted lead = $616 ceiling. Set the value at $2,800 and Smart Bidding will bid as if every contacted lead becomes an install. It doesn’t. The model overspends and CPqL climbs.
Skip this lever and the system will chase contacted leads from queries that produce phone-answerers who never buy. The contacted-lead value alone doesn’t filter for purchase intent.
Lever 4: Daily push cadence, tight latency
Offline conversion import is not a weekly job. Push contacted-lead conversions daily. Push sold-customer conversions daily as they close. The 14–21 day lag on sold events is fine: Google attributes them back to the original click as long as you are inside the conversion window.
The working rule: contacted-lead signals lagging more than 72 hours behind the click cause measurable Smart Bidding decay. The model needs fresh feedback on which queries produced contactable humans. Weekly batch uploads break that loop.
If your CRM cannot push daily, use a middleware layer. The legacy OCI endpoint is being replaced by Google’s Data Manager API, and the migration is a Smart Bidding event in its own right. Our migration playbook covers it if you have not started.
Search Themes, Brand Controls, and What Your Google Rep Skipped
The conversion rebuild is the load-bearing fix. These are the constraint layers that keep AI Max’s matching breadth from drifting back out.
Search themes. Treat them like guardrails, not keywords. Three to seven themes per ad group, narrow and intent-loaded. “Medicare Advantage enrollment 2026” beats “Medicare plans.” The narrower the theme, the tighter AI Max’s interpretation space.
Brand exclusions. If you don’t sell competitor brands, exclude them at the account level. AI Max will happily match competitor-brand queries to your ads if it thinks the intent is similar, and competitor-brand researchers are typically a weak contact-rate cohort.
Final URL expansion (FUE). Turn it off for lead-gen until you are confident in the conversion rebuild. FUE lets Google route traffic to any page on your domain, including blog posts and resource pages that aren’t built to convert. Unaudited FUE is a common source of low-quality form-fills on accounts that recently migrated.
Why didn’t your rep mention any of this? Because the in-platform CPL dashboard makes AI Max look like a free win. Reps are measured on activation and feature adoption, not on contact rate in your dialer. This isn’t malicious. It’s the incentive structure.
What Re-Stabilization Looks Like by Vertical
After the rebuild, Smart Bidding needs time to relearn. Google’s own guidance is that bid-strategy changes can take roughly 2 conversion cycles or around 50 conversions to recalibrate (Google Ads Help: About bid strategy status). In practice, with healthy daily volume, that is often 10–14 days. With thin volume, it stretches longer. Do not stack additional changes during the relearn window or you restart the clock.
The contacted-lead values below are starting points. Calibrate to your own funnel math.
| Vertical | Contacted-lead value | Why |
|---|---|---|
| Medicare / ACA | $25–$30 | High contact rates, tight margins, long persistency window |
| Final expense | $30–$35 | Lower contact rate, higher policy value, persistency risk |
| HVAC / plumbing / electrical | $35–$45 | High booked-job value, faster close cycle |
| Roofing / solar | $40–$55 | Long sales cycle, high ticket, fewer leads needed |
| Mortgage | $45–$60 | Longer offline lag, rate-sensitive close rate |
| B2B lead gen | $50–$75 | Long sales cycles, scarce qualified contacts, high ACV |
For vertical-specific deep cuts, the Medicare CPA piece on 9-month persistency-adjusted benchmarks, the HVAC cost-per-booked-install playbook, and the server-side conversion tracking schema for insurance lead buyers all assume the conversion-value rebuild is already in place. Run the rebuild first, then layer the vertical playbook on top.
FAQ
Should I just turn AI Max off?
For most lead-gen accounts, no. With DSA retiring and AI Max becoming the default surface for keywordless and inferred-intent traffic, opting out leaves you fighting for the smaller pool of literal-match queries. The conversion-value rebuild lets you keep the matching breadth while controlling what it optimizes toward. If you are under $25k/month spend with no offline conversion plumbing, opting out is defensible as a stopgap.
How long does Smart Bidding take to re-stabilize after the rebuild?
Google’s published guidance is that bid-strategy changes can take around 2 conversion cycles or roughly 50 conversions to recalibrate (Google Ads Help). For accounts with healthy daily volume (15–20 contacted leads per campaign per day), that often lands in the 10–14 day range. Lower volume extends the window. Do not make additional bid strategy changes during this period or you restart the clock.
What if my form-fill volume drops sharply after I set the value to $1?
That is the system doing what it should. You were buying form-fill volume. Now you are buying contacted-lead volume. Total form-fills can fall while contacted leads stay flat or rise. The dashboard CPL will look worse. Cost-per-qualified-lead, the only number that matters, gets better.
Does AI Max respect negative keyword lists?
Yes, but with caveats. Negatives still block the literal terms you list. AI Max can still match semantically similar queries that don’t contain your negatives. Build negatives around concept clusters, not single terms. The recent change to how Google reports AI-interpreted search terms makes negative-keyword visibility harder, which is its own problem worth a separate look.
What’s the minimum daily conversion volume for Smart Bidding to train at the contacted-lead tier?
Google’s published guidance is 30 conversions in 30 days for tCPA. In practice, for tROAS and value-based bidding on contacted leads, look for 15–20 contacted leads per day per campaign before the model converges cleanly. Below that, consolidate ad groups or campaigns to pool the signal.
How do I weight pay-per-call conversions vs. form conversions under AI Max?
Calls that hit your billable duration threshold (typically 60–120 seconds depending on vertical) should be treated as contacted leads and weighted at the $25–$40 range. Calls under threshold should be valued at $1 or excluded. Don’t mix call and form values in one conversion action. Separate them so you can see which channel AI Max is favoring.
Why does my Google rep keep recommending more conversion actions instead of reweighting?
Because more actions means more in-platform conversions, which makes the account look healthier on Google’s side. Reweighting reduces reported conversion volume in the short term. Stand firm. Fewer, better-weighted actions beat a long list of equally-weighted form events every time.
Run the Rebuild With a Team That’s Already Done It in Your Vertical
The rebuild is doable in-house. The cost of guessing wrong on contacted-lead values is a month of polluted bidding-model training data you cannot undo without another relearn cycle.
If you are buying leads or running pay-per-call in insurance, home services, mortgage, or B2B and your contact rate has slipped since AI Max activated, talk to our pay-per-call and lead-buying team. We’ll walk your account, pull your form-fill-to-contact gap, and tell you which of the four levers will move the most for your vertical and volume. Book a free strategy call with Elevarus and we’ll build a custom paid media plan for your business.
