AI Max Just Turned 1 and Inherited Your DSA Campaigns: The Three-Step Steering Sequence That Stops Junk Conversions

Article title on dark teal banner with green accents announcing AI Max's three-step steering sequence for DSA campaigns.

Share This Post

AI Max Just Turned 1 and Inherited Your DSA Campaigns: The Three-Step Steering Sequence That Stops Junk Conversions

TL;DR

  • Google’s “AI Max turns 1” announcement quietly closed the door on Dynamic Search Ads. Force-migration completes by September 2026 with no opt-out.
  • In Elevarus-managed accounts that migrated DSA campaigns voluntarily, 18–22% of post-migration spend landed on URLs that were explicitly excluded in the source DSA campaign. Inherited exclusions behave as suggestions until re-declared at the asset-group level.
  • Without a rebuilt signal hierarchy, AI Max relearns toward the cheapest fillable action inside 72 hours. CPL holds flat while booked-revenue-per-lead drops 20–35% in the accounts we’ve audited.
  • The three-step AI Max DSA migration sequence has to run in order: re-declare URL exclusions, lock the brand inclusion list, then demote form fills. Reverse the order and you reset the learning phase.
  • This is an AI Max for Search playbook, not Performance Max. Don’t confuse the two.

The AI Max Anniversary Post Was a DSA Sunset Notice in Disguise

Google’s “AI Max turns 1” post is being read as a feature update. It’s not. It’s the official close of the AI Max DSA migration window, and lead-gen advertisers spending $25k to $500k a month on Search are the ones who’ll feel it first.

Portrait checklist infographic in teal and green outlining AI Max DSA migration steps.
AI Max DSA migration — metrics and decision framework.

The headline that matters: Dynamic Search Ads will be force-migrated into AI Max for Search by September 2026. There’s no opt-out path. The voluntary window closes, and Google flips the campaigns for you.

If you’re still running DSA today, you have a fixed runway to rebuild your guardrails before the system inherits your campaign shell and starts spending against signals you never approved.

Here’s the trap. AI Max imports the DSA campaign, but it doesn’t inherit your guardrails the way operators assume. URL exclusions, page-feed boundaries, and brand-term carve-outs all get treated as soft suggestions until you re-declare them at the asset-group level. Add in a conversion hierarchy that still treats form fills as primary, and the system will optimize toward the cheapest fillable action it can find inside the first 72 hours of relearn.

This article is the three-step steering sequence that stops that, in the order it has to run.

What Actually Changed: Inherited Exclusions Are Suggestions Now

Three shifts came out of the anniversary post. Each one matters for lead-gen.

The September 2026 Deadline and What “No Opt-Out” Actually Means

Google confirmed that DSA campaigns will be migrated automatically if you haven’t moved them yourself. There is no toggle to stay on DSA. The voluntary window, where you can run the migration on your timeline with your steering controls in place before the system starts learning, closes when Google’s queue picks up your account.

The practical read: every week you wait shortens the time you have to instrument the migration properly. Force-migrated campaigns enter the relearn phase with whatever signal hierarchy was attached to the source DSA campaign, and that hierarchy was almost certainly built when form fills were a reasonable proxy for revenue. They aren’t anymore.

Why Inherited DSA Exclusions Don’t Behave Like Exclusions Anymore

This is the part that catches operators off guard. When AI Max inherits a DSA campaign, the URL exclusions, page feeds, and negative keyword lists carry over visually. They show up in the interface. You can see them. They look like they’re working.

They aren’t, fully. Across Elevarus-managed lead-gen accounts that migrated voluntarily, roughly 18–22% of post-migration spend landed on URLs that had been explicitly excluded in the source DSA campaign. The system treats inherited exclusions as suggestions, weighted inputs to the broader expansion logic, until they’re re-declared at the asset-group level inside the new AI Max structure.

We covered the pre-migration audit that catches this in our DSA retirement playbook, and Google’s own AI Max documentation describes the asset-group declaration requirement, though the language is buried.

Operator Note: If you pull a URL report 14 days after migration and see spend against pages you excluded in DSA, that’s not a bug. That’s the inheritance behavior working as designed. Re-declaring at the asset-group level is what turns a suggestion into a hard constraint.

AI Max for Search vs. Performance Max: Don’t Confuse the Two

A quick clarification, because the naming is genuinely bad. AI Max is a steering layer that sits inside Search campaigns. It replaces DSA and adds match-type expansion, asset generation, and URL-level optimization to traditional keyword campaigns.

Performance Max is a separate campaign type that runs across Search, Display, YouTube, Gmail, and Discover. Different surfaces, different controls, different bidding behavior. If you’ve read advice about PMax hybrid strategies, it doesn’t apply here. AI Max for Search is its own thing.

What It Means for Lead-Gen: AI Max Will Bid Toward the Cheapest Fillable Action

Here’s the part that breaks dashboards.

AI Max optimizes against whatever conversion signal is marked primary. For most migrated DSA campaigns, that’s still “form fill” or “phone click,” the same setup that’s been running for years. Within roughly 72 hours of migration, the system relearns. It finds the cheapest version of that action and bids harder toward it. Junk form fills. Sub-30-second hangups. Bot submissions that pass basic validation. Lead forms completed by people who clicked the wrong ad.

The dashboards lie to you because cost-per-lead holds or improves. Volume goes up. The Google Ads interface shows green. Meanwhile, your call center is reporting that contact rates are down, your insurance carriers are flagging quality issues, your HVAC dispatchers can’t reach the homeowner, your mortgage LOs are saying the leads aren’t qualified. Booked-revenue-per-lead drops 20–35% in the accounts we’ve audited, and it takes two to three weeks of CRM data to see it clearly.

This is why the maximum profitable CPL formula matters more than the CPL number itself.

Key Concept: Maximum profitable CPL = gross profit per customer × lead-to-sale conversion rate. If your lead-to-sale rate drops because AI Max is feeding you junk, your max profitable CPL drops with it. The CPL the dashboard shows can be “good” and unprofitable at the same time.

Why Sequence Matters: Demoting Conversions Before Fixing Exclusions Resets Learning

This is the operator-level insight most migration guides skip. The order of the steering sequence determines whether the relearn phase costs you a week or a month.

If you demote form-fill conversions to secondary before re-declaring URL exclusions and locking the brand inclusion list, the system spends the next 72 hours relearning bid patterns under the new signal hierarchy, but it’s still spending against the bad URLs the inherited exclusions didn’t actually block. You teach the system that those junk pages produce qualified calls. When you finally re-declare the exclusions, you’ve reset the learning phase entirely, because the system now has to unlearn what it just learned.

We’ve watched this cost accounts two to three additional weeks of stabilization spend. Fix the perimeter first. Then change the target.

The Three-Step AI Max DSA Migration Steering Sequence, In Order

Run these in sequence. Don’t skip a step. Don’t reorder.

Step Action Why It Comes Here
1 Re-declare URL exclusions and page feeds at asset-group level Converts inherited “suggestions” into hard constraints before any relearn
2 Lock the brand inclusion list Stabilizes brand vs. non-brand definition before bidding logic shifts
3 Demote form fills, promote qualified calls or booked appointments Targets revenue once the perimeter is clean

Step 1: Re-Declare URL Exclusions and Page Feeds at the Asset-Group Level

Open the migrated AI Max campaign. Go to the asset group, not the campaign-level settings. Re-enter every URL exclusion from the source DSA campaign. Re-attach the page feed with the same boundaries.

The visual duplication feels redundant, the exclusions are already showing in the campaign view, but the asset-group declaration is what converts them from suggestions into hard constraints.

If you have a long exclusion list, export it from the DSA campaign before you migrate, then bulk-upload at the asset-group level after. Don’t trust the inheritance.

Quick Win: After Step 1, run a placement and URL report. If you see spend against any re-declared exclusion in the next 48 hours, the declaration didn’t take. Re-enter manually.

Step 2: Lock the Brand Inclusion List Before the Relearn Triggers

AI Max’s match-type expansion will bleed brand spend into generic queries during the relearn window if you don’t lock the brand inclusion list. For most lead-gen accounts, brand traffic is the highest-converting and lowest-CPL portion of the account. Letting AI Max blur it into generic match expansion inflates blended CPL and gives the system bad signal about what “good” looks like.

Use the brand inclusion list to specify exactly which queries count as brand. Lock it. Set it before any conversion changes happen, because the system needs a stable definition of brand vs. non-brand before it relearns bidding patterns under a new primary signal.

Step 3: Demote Form Fills, Promote Qualified Calls or Booked Appointments as Primary

This is the step everyone wants to do first. It’s the last one for a reason.

In the conversion goals section, demote form-fill conversions to secondary. Promote a downstream signal as primary: a qualified call (90+ seconds with a confirmed intent), a booked appointment, or a CRM-stage conversion imported via offline conversion tracking or Enhanced Conversions for Leads. The signal you promote should represent money: booked revenue, not just contact attempts.

For pay-per-call accounts running through Ringba or Invoca, the qualified-call threshold matters. In our experience across insurance, home services, and mortgage, 90 seconds plus a duration-weighted value is the cleanest signal: HVAC LSA-style calls under 60 seconds qualify at under 10%, so the 90-second floor catches roughly 80% of bookable intent without over-filtering. For form-led verticals, a CRM-stage conversion (qualified, contacted, sold) imported daily is the durable fix.

The point is the same either way: AI Max can only optimize toward what you tell it to value. Tell it to value the action that produces revenue.

Key Stat: When form-fill conversions remain primary post-migration, AI Max relearns toward the cheapest fillable action inside roughly 72 hours. Demoting form fills and promoting a qualified-call or booked-appointment signal is the only durable fix we’ve seen across managed accounts.

What to Watch in the 72 Hours After the Sequence Completes

Three metrics, daily, for the first week after Step 3:

  • Revenue per lead (total revenue divided by total qualified leads). Should hold or improve. If it drops, the primary signal isn’t clean.
  • Qualified-call rate or qualified-lead rate. Calls over 90 seconds divided by total calls, or CRM-qualified divided by total leads. Should rise as the system relearns.
  • Asset-group spend distribution against the re-declared exclusion list. Should be near zero against excluded URLs. If it’s not, Step 1 didn’t take and you need to re-enter the exclusions.

If all three look right at day 7, the relearn worked. If revenue per lead is dropping while CPL improves, you have signal that looks like the form-fill trap, and the primary conversion needs to move further downstream.

FAQ: AI Max DSA Migration

Will my DSA campaign performance carry over after AI Max migration?

Visually, yes. Functionally, no. The campaign shell, exclusions, and page feeds appear in the new interface, but inherited exclusions behave as soft suggestions until re-declared at the asset-group level. Expect a 72-hour relearn phase regardless of how stable the source DSA campaign was.

What’s the actual deadline for AI Max DSA migration?

Google has confirmed force-migration of remaining DSA campaigns by September 2026. There is no opt-out. Voluntary migration before that date lets you control timing and instrument steering controls before the relearn phase begins.

Is AI Max the same as Performance Max?

No. AI Max is a steering layer inside Search campaigns and is the direct DSA replacement. Performance Max is a separate campaign type running across Search, Display, YouTube, Gmail, and Discover. Different surfaces, different controls, different bidding logic.

Can I keep form fills as my primary conversion if I have good lead validation upstream?

Only if your validation rejects bad leads before they fire the conversion event. If form fills fire on submission and validation happens later, AI Max will optimize toward submission volume regardless of downstream quality. The cleanest fix is firing the conversion only after validation passes, or promoting a CRM-stage signal as primary.

What happens if I run the three steps out of order?

If you demote conversions before re-declaring exclusions, the system spends the next 72 hours relearning bid patterns against URLs the inherited exclusions didn’t actually block. You teach AI Max that junk pages produce qualified actions. Re-declaring exclusions later resets the learning phase entirely, costing two to three additional weeks of stabilization spend.

How do I know the relearn phase is over?

Watch revenue per lead, qualified-call rate, and asset-group spend distribution daily. When all three stabilize for five consecutive days (typically day 7 to day 12 post-sequence), the relearn is complete. Bid changes after that point compound normally instead of triggering another reset.

Talk to Our Pay-Per-Call Team Before Google Migrates the Campaign For You

The sequence above assumes you have clean offline conversion tracking and a downstream signal worth promoting. A lot of lead-gen accounts don’t, which is exactly why CPL becomes the only visible metric and why AI Max’s relearn quietly burns money. If your stack still treats the form fill or the raw call as the conversion, the steering controls don’t have anything real to steer toward.

If you’re running paid acquisition in insurance, home services, mortgage, or B2B lead gen and your DSA campaigns haven’t been migrated yet, talk to our pay-per-call team before Google does it for you. We’ll audit the inherited DSA structure, build the conversion-signal hierarchy that maps to booked revenue (not form fills), and run the steering sequence inside the voluntary window, so the relearn happens on your terms, not the queue’s.

Ask about exclusive lead routing for your vertical and volume need. Book a free strategy call with Elevarus to build a custom paid media plan for your business.

Picture of SHANE MCINTYRE

SHANE MCINTYRE

Founder & Executive with a Background in Marketing and Technology | Director of Growth Marketing.