Google Ads in 2026: How AI Max, Performance Max, and Smart Bidding Are Changing PPC Forever
By Jonathan Alonso – founder, YellowJack Media, Orlando, FL
I’ve been buying Google Ads for 14 years and I can safely say: the next 12 months will re-wire every habit we ever built around keywords, bids, and creative swaps. In late 2025 Google quietly rolled out AI Max campaigns (internally called “Search AI Max”) to a handful of accounts. By January 2026 every MCC in North America had the option to run keyword-free Search. Let that sink in—keyword-free Search. The interface simply asks for a landing page, a daily budget, and ROAS or CPA goals. Everything else—match types, negatives, ad copy variants, sitelinks—is handled by Gemini 3, the same large-language model powering Bard and the latest Pixel phones.
Meanwhile Performance Max is no longer the awkward “catch-all” hybrid campaign we met in 2021. It’s now the engine behind 62 % of all Google ad clicks according to the Google Ads Blog (Feb 2026 update). And Microsoft isn’t sitting still—Bing’s Copilot Shopping campaigns shipped in March with product-level creative generation and the same auction-time bidding muscle Google has been perfecting since 2018.
I manage roughly PPC management accounts across e-commerce, SaaS, and local lead-gen out of our Orlando office. The goal of this post is to tell you what’s actually working, what’s hype, and how to re-budget before your Q4 planning meetings hit. Let’s dive in.
The Keyword Is Dead, Long Live the Keyword (AI Max)
AI Max feels like someone handed the keys to a self-driving Tesla and said, “Don’t touch the wheel.” You give Google a URL and a KPI—say, 4.2 ROAS—and within 24 hours the campaign is live across Search, Discover, Gmail, YouTube Shorts, and Maps. Google auto-generates headlines, long-headlines, sitelinks, callouts, images, and even 30-second YouTube cuts from your product feed. The creepy part? It works. Our first beta (a DTC skincare brand out of Winter Park) lifted total sales 37 % while cutting CPA 28 % compared to legacy Exact + Phrase + RSA mix.
But there are guardrails:
- Feed quality matters more than ever. If your titles or Gtin attributes are thin, Gemini hallucinates “best Anti-Aging Solution” when you actually sell vegan body wash. Bad creative hallucinations tank Quality Score faster than you can say “broad match.”
- Brand safety is still manual. You MUST upload brand negatives and switch on “Brand Exclusions” inside the AI Max settings or you’ll buy your own brand at $6 CPC.
- Learning periods are longer. Google recommends a 3-week runway and 60-conversion threshold before any tCPA or tROAS edits. Smaller budgets (<$150/day) need patience.
My rule: use AI Max when you have (1) 300+ SKUs, (2) clean feed, and (3) willingness to give 30-40 % of total spend to the machine. Otherwise, keep Hybrid Search running.
Performance Max 2.0—Asset Groups, Audience Signals & Brand Exclusions
If you read our earlier Google Ads best practices guide you know I was on the fence about PMax when it launched. Twenty-four months later it’s matured. Here’s what changed:
- Asset Group-level reporting: we can finally see which audience signal or product group drove the conversion. No more black box.
- Brand exclusions at campaign level: Google heard our prayers and lets you pick brand terms to push back to a Standard Shopping or Search campaign.
- Video funnels inside PMax: upload a 5-sec bumper, 15-sec vertical, and 30-sec skippable. Google stitches them into an AI narrative for high-intent audiences.
We structure PMax around business outcome, not SKU count. For a kayak rental client we run two asset groups: “Locals 25 mi” and “Tourist Hotels.” Each group gets unique creatives (locals want storage, tourists want sunrise tour photos). Conversions up 41 %, cost per booking down 19 % since the switch.
Smart Bidding x Gemini 3—The 2026 Secret Sauce
Most advertisers think Smart Bidding maxed out in 2023. Google’s internal docs say the Gemini 3 integration added 11 new context signals including:
- time since last YouTube view of brand content
- latent vector of landing-page sentiment (positive vs. negative review tone)
- ChatGPT-style query paraphrasing to map “near me” to geo radius
The upshot: for high-volume accounts we’re seeing tROAS targets hit 6-9 % closer to goal compared to 2024. For low-volume (<30 conv/Month) tCPA is still noisy. My workaround: consolidate campaigns, group similar products under one tROAS, and wait 14 days before tweaking. Patience is the new bid modifier.
From Manual to Intent-Based—How We Re-Train Teams
Old school PPC was keyword archaeology. New school is intent architecture. We rebuilt our Orlando PPC agency workflow around three pillars:
- Data Layer Integrity: GTM server-side, offline conversion import, and CRM matching decide if Smart Bidding even works. Garbage in, garbage out.
- Creative Velocity: Every account now has a 2-week sprint to ship 15 new image/text/video assets for each audience quadrant (prospect/retarget/lapsed). Gemini chooses winners, but humans feed it.
- Budget Agility: we move 20-30 % of spend between AI Max, PMax, and Advantage+ every Monday using Google’s new “Budget Heat-map” that predicts ROAS lift by channel for the next 7 days.
Junior analysts no longer mine Search-term reports—they validate exclusions, feed attributes, and creative sentiment scores. Senior strategists optimize business constraints (margin, LTV, seasonality) rather than CPC bids. The result: same headcount, 34 % more conversions YoY across our book of business.
Microsoft Advertising—Copilot Shopping Enters the Race
Bing still owns 7.2 % of U.S. search, but 15 % of disposable-income queries (think finance, travel, B2B SaaS). Their March 2026 Copilot Shopping update auto-creates lifestyle imagery and pricing overlays for every SKU, then pushes the ad into Bing Chat, Edge sidebar, and MSN. Early tests show 19 % lower CPA than comparable PMax for one of our software clients. Takeaway: mirror your PMax feed to Microsoft Merchant Center even if Bing spends <10 % of budget. It’s 30 minutes of work and incremental ROAS.
Meta Ads—Advantage+ Shops and AI Creative
Yes, this is a Google post, but Meta’s trajectory mirrors Google’s. Advantage+ Shopping (A+SC) is testing open-ended creative where you upload a product catalog and 5 headlines; Meta generates infinite UGC-style videos using AI actors. We’re seeing CPAs drop 25 % for fashion clients but creative fatigue at 14 days instead of 45. The play: pair Advantage+ retarget with PMax prospecting to cover the full funnel. Allocate 60 % Google, 30 % Meta, 10 % test (TikTok or YouTube).
Practical Budget Playbook for Q4 2026
Below is the exact budget split I’m pitching my clients this month. Adapt the percentages based on feed size, margin, and brand awareness:
- 40 % AI Max Search: only if (a) 500+ conversions/mo, (b) tROAS history within 15 % of goal, (c) feed scored “excellent” in Google’s Diagnostics.
- 25 % Performance Max: split into 2–3 asset groups. Exclude brand. Turn on YouTube for scale but cap CPM at $18.
- 15 % Standard Shopping (manual): protect top 20 % SKU’s margin. Manual CPC with enhanced ROAS for last-click attribution insurance.
- 10 % Microsoft Copilot Shopping: import feed, enable automatic image enhancements, set tCPA 10 % lower than Google as a carrot.
- 10 % Test Bucket: Meta A+SC, TikTok Shop, or YouTube Video View (for reach). Rotate monthly.
Meeting with a CFO who wants hard projections? Use Google’s newest Performance Planner—it already models AI Max and PMax lift curves separately. Add a 15 % efficiency buffer; the models are optimistic by default.
What This Means for Agencies vs. In-House Teams
If you run an agency:
- Re-position deliverables around feed & creative ops. Clients don’t care about keywords; they care about margin. Charge for feed clean-up and motion creative the same way you once charged for long-tail expansion.
- Standardize cross-channel attribution reporting or risk losing budget to in-house teams who promise “transparency.” Looker Studio + GA4 server-side is our minimum stack.
- Build a migration SLO: 30-day sprint to move any account >400 conversions from Hybrid to AI Max/PMax with rollback plan.
If you’re in-house:
- Keep at least one human who understands SQL and GA4 BigQuery exports. Machines break; when they do, you still need to answer why revenue dropped 18 % week-over-week.
- Use digital marketing consulting partners for quarterly audits, not daily management. The best results come from short, high-intensity engagements.
- Don’t fire your agency the day AI Max goes live. We saw a 14 % revenue dip in accounts that removed human oversight entirely. Instead, shift agency scope to incrementality testing and creative production.
Closing Thoughts—Embrace, Verify, Fortify
AI Max, PMax 2.0, Gemini-powered Smart Bidding—these aren’t shiny objects. They’re the new floor. Google is literally giving every advertiser a super-computer that writes ads better than most copywriters. Our job is to feed it truth (clean data), guide it with goals (business metrics), and guard it with guardrails (brand negatives, exclusion lists, and margin controls).
Keep testing, keep questioning, and—most important—keep a human in the loop. If you want to swap Orlando PPC notes, find me at LinkedIn or drop by our Orlando office. Happy optimizing.
Jonathan Alonso is founder of YellowJack Media, a boutique growth agency in Orlando specializing in PPC, paid social, and conversion rate optimization.
