The Headline Everyone Wrote vs. the Story That Actually Matters
The coverage of Google Marketing Live 2025 and the twelve months that followed was mostly about the "Power Pack" — Google's name for Demand Gen, AI Max for Search, and Performance Max working as a single AI-led system. Every agency blog in Canada wrote some version of "everything you need to know about the Power Pack."
That coverage is not wrong. It is just not the story that matters most for a Western Canadian HVAC company, a Spruce Grove law firm, or a three-location dental group. The Power Pack is real, and we will get to the independent data on it below. But three smaller, quieter changes Google shipped between late 2025 and April 2026 are more likely to reshape your next invoice than anything the AI press release promised.
Those three changes are the March 1, 2026 ad-schedule budget pacing rewrite, the deprecation of call ads, and the April 15 announcement that Dynamic Search Ads will be auto-upgraded into AI Max starting September 2026. Each change is individually small. Together, they change how you budget, how you generate phone leads, and how you scope AI-led expansion.
This post walks through what each of those changes actually does, what the independent data says about Google's AI narrative, and the operational adjustments we made across our portfolio so the changes became our own decisions instead of Google's.
The March 2026 Budget Trap
On March 1, 2026, Google changed how budget pacing works for any campaign that uses ad scheduling. The headline spending limits are unchanged: still up to 2x the average daily budget on a single day, still up to 30.4x the daily budget in a month. What changed is that campaigns are now pacing proactively toward the full 30.4x monthly cap inside whatever schedule you have set, regardless of how many days that schedule actually runs.
The math matters more than the language. Take an Edmonton HVAC account running Monday to Friday at $100 per day. Before March, that campaign behaved roughly like a $2,200-per-month campaign: five days a week at the daily target. After March, the same campaign is now allowed to pace toward the full $3,040 monthly envelope, concentrated into those same five days. An office-hours B2B campaign scheduled Monday to Thursday at $100 per day jumps from about $1,700 to roughly $3,040. A weekends-only campaign at $100 per day moves from about $800 to around $1,600, constrained by the 2x daily cap rather than the monthly one.
Nothing changed in the account. No new campaign, no budget edit, no strategy change. The client just spent more.
The mitigation is mechanical. If monthly budget is the actual business constraint, set daily budget to monthly target divided by 30.4 and stop using ad scheduling as a soft brake. Where a campaign is genuinely time-bound, such as a spring paving push or a one-week promotion, campaign total budgets are the cleaner tool because they cap the full campaign duration rather than relying on averaged daily logic. The catch is that campaign total budgets are only available on new campaigns and cannot be retrofitted onto existing ones. Deciding which campaigns to rebuild versus which to reset daily budgets on is itself a scoped piece of operational work, and skipping it is how accounts quietly start overspending.
Call Ads Are Gone. Your Intake Probably Noticed.
Google announced in October 2025 that call ads were being deprecated. The ability to create new call ads was removed in February 2026. Existing call ads are scheduled to stop receiving impressions in February 2027. The replacement is a standard responsive search ad with call assets attached.
Practitioner reports in March and April suggest the UI rollout has been inconsistent, with some accounts temporarily regaining the ability to create or edit call ads. The direction is not in doubt. The operational impact is.
Call ads were mobile-only and explicitly designed to drive a phone call as the primary response. For phone-led verticals (HVAC, plumbing, restoration, legal, automotive service, medical) that format was doing real work. Responsive search ads with call assets preserve the ability to call, but the ad surface is wider, the landing page matters more, and the asset mix shifts clicks toward the website as well as the phone.
What that does to lead mix is predictable. A plumbing account that was previously getting 70% phone calls and 30% form submissions can find itself at 55/45 or 50/50 within a few weeks of the transition, with total lead volume flat or slightly up. Cost per lead does not necessarily change. Intake economics might. A phone-led business that trained its front office to close on the call now has more form submissions landing in an inbox that was an afterthought.
We treat the call-ad transition as an intake operations project, not a media change. Conversion reporting gets split so phone-call conversions and form-fill conversions track separately. The CRM or intake system gets a usable form-follow-up workflow. Landing pages get click-to-call buttons treated as primary, not secondary. Only then is the RSA plus call-asset replacement a clean swap. Clients who skipped the intake work and just changed the ad format saw the same lead count and worse conversion to booked work, because the form submissions were queueing up behind a process that was never built for them.
DSA Gets Auto-Upgraded to AI Max in September
On April 15, 2026, Google confirmed that campaigns using Dynamic Search Ads, automatically created assets, and campaign-level broad match settings will begin auto-upgrading to AI Max in September 2026. The upgrades are expected to conclude by the end of that month. New DSA campaign creation is being wound down in parallel.
For mature accounts, this is a bigger operational event than the announcement suggests. DSA was, for many advertisers, a controlled long-tail catch-all that ran alongside a disciplined keyword structure. It filled the gaps between the exact-match queries the planner had accounted for and the real variation in how people searched. AI Max is a different animal. It expands queries more aggressively, layers creative automation on top, and is built to find conversions outside of whatever keyword scaffolding you have erected.
Being migrated involuntarily into a more expansionist system is not the same as running a controlled AI Max test on your own schedule. The advertisers who will do well with the migration are the ones who ran AI Max as an opt-in experiment earlier, measured it against qualified-lead outcomes rather than topline click or conversion counts, and have a view on whether the expansion is incremental to their existing Search coverage or substitutional. The advertisers who will do badly are the ones who find out in late September.
The Power Pack Lift Numbers, Honestly Read
Google's own performance claims for the Power Pack are strong, specific, and mostly sourced from Google's internal data. AI Max typically produces 14% more conversions or conversion value at a similar CPA or ROAS; for non-retail campaigns still dominated by exact and phrase match, the figure rises to 27%. Performance Max users outside retail see an average 27% lift. Demand Gen was reported to deliver a 26% increase in conversions per dollar in 2025. Smart Bidding Exploration averaged 18% more unique converting search categories and 19% more conversions.
Those are useful directional signals. They are not a clean, like-for-like proof that the Power Pack outperforms a well-built legacy account in every vertical.
The most useful independent pushback came from Mike Ryan's analysis at Smarter Ecommerce, summarized by Search Engine Journal, covering more than 250 campaigns that adopted AI Max. Across that set, the median campaign gained 13% in conversion value, close to Google's claim. The median campaign also gained 16% in CPA. ROAS outcomes were dispersed enough that individual campaigns ranged from 42% above baseline to 35% below it.
Read carefully, that data describes AI Max as a volume-expansion layer rather than an efficiency improvement. Revenue grows, the marginal conversion costs more, and whether the total is a win depends on your margin structure and how rigorously you measure lead quality.
| Source | Conversion value | CPA | Notes |
|---|---|---|---|
| Google, AI Max launch (May 2025) | +14% | similar | Excludes retail |
| Google, AI Max (April 2026) | +7% | similar | Full feature suite vs. search-term matching alone |
| Google, Performance Max | +27% | similar | Non-retail |
| Google, Demand Gen 2025 | +26% per dollar | n/a | Internal year-end summary |
| SMEC independent, 250+ campaigns | +13% median | +16% median | ROAS range +42% to -35% |
This is not a refutation of Google's narrative. It is a qualification. For a retail advertiser on 40% gross margin, 13% more revenue at 16% higher CPA is probably fine. For a lead-generation account where the cost of a bad lead is not just wasted ad spend but wasted intake time, the same numbers are not fine.
The Canadian Wrinkle
Ads within AI Overviews have been live in English in Canada across mobile and desktop since the 2025 expansion. Ads above and below AI Overviews run in every market where AI Overviews themselves are available. That means AI-mediated search is not a future state Canadian advertisers are waiting on; it is already a live operating environment.
Two caveats matter for the kind of portfolio we run. First, Google excludes sensitive verticals, explicitly including finance and healthcare, from ads inside AI Overviews. For dental practices, optometry clinics, medical specialists, and financial advisors, that exclusion is not cosmetic. It means AI Overview inventory is not available to those accounts on the same terms as a paving contractor or a home-services business, and the generic "get into AI search" marketing narrative does not map cleanly onto their account.
Second, the old exact-match mental model no longer reaches all of the inventory. Google has clarified that exact and phrase match keywords are not eligible to trigger ads within AI Overviews. The match types that can trigger those placements are broad match, AI Max, Performance Max, Shopping, and DSA. Exact-match discipline is still useful for cost control and quality assurance. It is no longer sufficient on its own to access every valuable Google surface.
The practical implication is that a Canadian SMB account in 2026 needs a deliberate view on which match-type layers it runs, which AI-eligible layers it has explicitly tested, and which layers it has explicitly declined. "Default settings" is no longer a stable posture.
What the New Advertiser Job Looks Like
Put the three quiet changes and the Power Pack context together and the outline of the new advertiser job resolves. The scarce skill is no longer adjusting hundreds of bids or splitting match types into ever-finer ad groups. Google's AI is going to do that work, well or badly, with or without you.
What Google's AI cannot do on its own is define what a good conversion actually looks like for your business. It cannot import qualified-lead or revenue signals from a CRM you have not connected. It cannot tell whether a cheap conversion is a form fill from a real prospect or a returning customer who would have called anyway. It cannot structure budgets across Search, AI Max, Performance Max, and Demand Gen so those campaigns stop bidding against each other. It cannot feed itself higher-quality creative, better landing pages, or cleaner offline value signals.
Those are the jobs that actually move the numbers in 2026. Clean conversion actions. Offline value imports. Budget architecture that respects campaign boundaries. Structured creative and landing-page pipelines. Regular review of the channel-performance and asset-level reports Google added under advertiser pressure. Experimentation discipline that lets you test AI-led features on your own terms instead of inheriting their rollout.
The change is not from tactical to passive. It is from tactical to strategic. The agencies that understood that shift by mid-2026 are running portfolios that look operationally different from the ones that did not. That's where we have pointed the practice.
What We Changed Across the Portfolio
Across our active Google Ads accounts, the shift looked like a set of concrete, documented operational changes rather than a single strategy pivot. The work was mechanical.
Budget recalibration. Every ad-scheduled campaign was audited against true monthly spend intent. Where ad scheduling was being used as a soft brake (which, historically, it was on roughly half of the affected campaigns) daily budgets were reset to monthly target divided by 30.4 and ad scheduling was either removed or preserved only as a genuine day-part lever. For a handful of genuinely time-bound pushes, the legacy campaign was replaced with a new campaign using a total budget cap.
Call-ad migration with split reporting. For every phone-led account, call ads were replaced with responsive search ads plus call assets before the February 2027 impression sunset rather than at the deadline. Conversion tracking was split so phone-call conversions and form-fill conversions reported separately. Intake workflows were updated so form submissions stopped being an afterthought inbox.
Opt-in AI Max testing, on our own timeline. For accounts with enough conversion volume to produce a readable signal, AI Max was turned on inside a dedicated test campaign with explicit guardrails: brand exclusions, location intent controls, an experiment window, and a lead-quality criterion more strict than Google Ads' default. The point was to enter September 2026 with our own data on whether AI Max expansion was incremental or substitutional, not Google's.
Tighter conversion-action hygiene. Every account's conversion configuration was reviewed for duplicate actions, mis-scoped primary/secondary flags, and stale offline imports. Smart Bidding only performs as well as what you let it optimize toward, and a quiet tracking break is not something any amount of AI can fix from the inside. That is a recurring theme in our operations work and something we wrote up separately in Conversion Tracking Breaks Silently.
Weekly ops added channel and asset reports. Performance Max channel performance reports and the new asset-level reports were added to the standing weekly review, alongside the Search-term and placement reviews we already ran. The reports exist because advertisers pushed Google for them. Using them is how you get the value they were built to deliver.
The Short Version
The tactical advertiser's edge is shrinking. The strategic advertiser's edge is growing. The agencies winning in 2026 are not the ones that surrendered control to Google's AI fastest, and they are not the ones that clung hardest to exact-match nostalgia. They are the ones that accepted where the platform was going, then built stronger measurement, cleaner budget architecture, better creative feedstock, and tighter lead-quality feedback loops than their competitors.
Read more about our Google Ads management approach, our Edmonton PPC practice, and our broader work on AI marketing and conversion optimization. For related posts, see We Track Your Business Across 10 AI Platforms and Google Ads vs Meta Ads: Where to Spend Your First Dollar.