Google Search Ads in 2026: How AI Max, Broad Match, and Smart Bidding Actually Work Together

Two numbers should reframe how you think about Google Search this year. Ecommerce search CPCs climbed roughly 33% over the last twelve months to around $2.61 a click, and when independent testers measured Google's new AI Max for Search at the account level, 84% of advertisers saw neutral or negative results. Read those together and the takeaway is uncomfortable: clicks got more expensive, and the shiny new feature meant to make them worth more mostly didn't, at least not the way the demo promised.

That doesn't mean AI Max, broad match, or Smart Bidding are broken. It means the money on Google Search isn't made anymore by finding cheaper clicks or flipping on the newest setting. It's made in the matching and bidding discipline underneath. Across the brands we run, spending more than $7M a year in ad budget across 100+ brands, the Google accounts that scale are the ones that understand how these three layers work together, and which levers to leave off. Here's the operator's version.

What actually changed on Google Search in 2026

The headline shift is AI Max for Search, which reached general availability on April 15, 2026. The most important thing to understand is what it isn't. It's not a new campaign type. It's an opt-in suite you layer on top of your existing Search campaigns, and it bundles three features. Search Themes let you feed Google intent signals in plain language. Final URL Expansion lets Google match a query to the most relevant page on your site instead of the one you specified. And AI-generated ad text writes and rewrites headlines and descriptions on the fly. Google is also folding Dynamic Search Ads into AI Max through 2026, so the old DSA workflow is quietly becoming the new one.

Two other changes matter for where your ads show up now. Paid search ads appear inside AI Overviews, the AI-written answer block sitting above traditional results that reaches around 2.5 billion people a month, and inside AI Mode, across more than 200 markets. Your Search ad isn't just competing for the top of a list of blue links anymore. It's competing to appear next to an AI-generated answer that may have already given the shopper most of what they came for.

The 2026 Google Search benchmarks every ecommerce brand should know

Before touching strategy, calibrate against the market. These are the figures your team will see quoted through the back half of 2026, pulled from cross-industry and ecommerce-specific data.

Metric	2026 figure	What it means
Cross-industry Search CPC	~$2.96 (up from $2.64 in 2025)	Auction pressure, AI Overviews, and Smart Bidding all push costs up
Ecommerce Search CPC	~$2.61, up ~33% YoY	The steepest cost increase is hitting retail specifically
Ecommerce Search CTR	~4.10%	Healthy; below ~3.5% usually signals weak ad relevance
Ecommerce conversion rate	~2.81% (retail often 3% to 5%)	Post-click; a lagging number is usually a landing-page problem
Share of spend on AI bidding	~78% of all Google Ads spend	Manual CPC is the exception now, not the rule
CPA improvement with AI bidding	~22% lower cost per conversion	Real, but it varies widely by account maturity

The pattern is the same one hitting every paid channel. Impressions and clicks cost more, competition is denser, and the platform's answer is to hand more of the decision-making to its own algorithms. Whether that helps or hurts you comes down to how well you feed and fence those algorithms.

How AI Max, broad match, and Smart Bidding fit together

These three get discussed as if they're competing options. They aren't. They're three layers of the same modern Search engine, and each does a different job.

Broad match is the reach layer

Broad match used to be the setting you were warned away from, the one that burned budget on irrelevant queries. That reputation is out of date. Paired with Smart Bidding, broad match is now how most sophisticated accounts run. 62% of advertisers using Smart Bidding use broad match as their primary match type, and Google's own data shows broad match delivers around 13% more conversions or conversion value at a similar CPA or ROAS when it runs with Smart Bidding rather than without it. The key phrase is "with Smart Bidding." Broad match without an algorithm filtering the auction is still the budget bonfire everyone remembers.

Smart Bidding is the filter

Smart Bidding is what makes broad reach safe. It sets bids at auction time, for each individual query, using signals no manual bid could weigh in real time. The strategy you pick decides what it optimizes toward. Maximize conversions and Target CPA chase volume at a cost, while Maximize conversion value and Target ROAS chase revenue quality. For ecommerce, the shift that pays is moving from Target CPA to Target ROAS, which Google measures at around 14% more conversion value at a similar ROAS. That's the same bid-to-margin logic behind profit on ad spend: you want the algorithm optimizing for the value of the sale, not the raw count of sales. If you sell products with very different margins, a blended CPA target quietly overpays for your cheap, low-margin winners and underbids the profitable ones.

AI Max is the expansion layer

The distinction that trips most teams up is simple. Broad match expands matching at the keyword level. AI Max expands matching at the campaign level, using signals from your whole account and website rather than from any single keyword. Run both together and you have a very aggressive matching engine that can pull in loose, exploratory queries fast. That's powerful when your tracking and negatives are tight. It's expensive when they aren't, because the system will happily spend to discover what doesn't work on your dime.

The honest part: why 84% of accounts don't win with AI Max yet

This is where the operator view diverges from the vendor pitch. Google's public claim is that turning on the full AI Max suite delivers about 7% more conversions or conversion value at a similar CPA, and that advertisers activating it typically see around 14% more conversions. But read the fine print. Google excludes retail from that 14% figure. For ecommerce specifically, one 250-campaign study found AI Max produced roughly 13% more revenue at a 16% higher CPA, with individual account ROAS swinging anywhere from 42% better to 35% worse. That isn't a rounding error. That's a coin flip with real money on it.

The deeper issue is incrementality, the same story as Performance Max cannibalizing branded search. When those independent testers measured at the account level instead of the campaign level, 84% saw neutral or negative results, because much of AI Max's "new" traffic was pulled from campaigns already capturing those buyers. The campaign dashboard looked better while the account didn't move. We've seen this exact pattern in audits: a fresh AI Max campaign posting a strong ROAS, sitting on top of an account whose total revenue is flat.

The 16% who genuinely win share a profile. They have precise conversion tracking with Enhanced Conversions enabled, solid negative keyword lists, at least 100 conversions a month feeding the model, an existing track record with broad match, and enough budget to absorb an expanded-matching learning period. If your account is missing three of those five, turning AI Max all the way on will probably cost you money before it makes you any. Fix the foundation first, then expand.

Brand controls: the levers that protect margin

The most underused part of the 2026 toolkit is the control surface Google added after AI Max shipped, and it's exactly where a disciplined account pulls ahead. These are the settings that decide whether the algorithm works for your P&L or against it.

Start with branded-search controls, which now have three settings: show ads on all relevant searches (the default), apply inclusion or exclusion rules for branded queries, or restrict delivery to unbranded searches only. That last one is the cleanest lever there is for isolating incremental, net-new demand, because it stops you paying Google to intercept people already typing your name. For text, you get up to 25 term exclusions and 40 messaging restrictions per campaign, written as natural-language rules rather than keyword lists, that constrain what Google's AI is allowed to say when it writes your copy. For any brand in a compliance-sensitive category, supplements, health claims, anything under MAP, those guardrails aren't optional. They're the difference between AI-written copy and a policy problem.

Then there's new-customer acquisition. Google's New Customer Acquisition goal uses your first-party customer lists to bid up net-new buyers, and the 2026 New Prospects Mode goes further, filtering out prior purchasers, brand searchers, and existing site visitors to isolate cold demand before optimization even begins. Feed it Customer Match data, which Google credits with a 5.3% conversion uplift, and you have a Search engine pointed at growth instead of at recycling buyers you already had. Judge that campaign on new-customer ratio and acquisition quality, not blended ROAS, or you'll misread it. It's the same retention-versus-acquisition math that decides whether any channel is actually growing your customer base.

How we'd split a Google budget in 2026

Put it together and the modern Google account isn't AI Max versus PMax versus Search. It's a deliberate stack. Google's own framework, and roughly how we'd start for an ecommerce brand before tailoring to margin and data, looks like this.

Layer	Job	Starting share of budget
Performance Max	Broad multi-channel reach across Shopping, YouTube, and Display	60% to 70%
AI Max for Search	High-intent search demand with more transparency and control	30% to 40%
Demand Gen	Top-of-funnel awareness and audience building	Fund from test budget

Treat those weights as a first draft, not a rule. The point of splitting PMax and AI Max for Search this way is control. PMax is the black box that goes wide, and AI Max for Search is where you keep visibility into search terms, brand settings, and ad text so you can actually manage incrementality. This kind of cross-channel budget modeling is core to how we run full-service ecommerce management. One more thing worth getting ahead of: Google is moving toward switching some of these AI features on by default later in 2026. Deciding deliberately now, with your tracking and controls in place, beats getting opted in automatically and finding out from your P&L.

The brands that win on Google Search in 2026 aren't the ones who turned everything on the day it launched. They're the ones who got measurement clean, kept broad match on a Smart Bidding leash, fenced their brand terms, and pointed the budget at new customers. If you want a team that runs Google and paid social that way, that's how we do it. If you'd rather see the numbers first, here's what that looks like across our portfolio, or tell us what you're working with and we'll tell you honestly whether AI Max is a fit for your account yet.