Google Ads in 2026: The Platform That Advertisers Built Their Skills On No Longer Exists

If you ran Google Ads campaigns in 2023 and stepped away for three years, you would barely recognize the platform you came back to. Broad match is the default for new campaigns. Exact match triggers on queries the advertiser never approved. AI writes ad copy, picks landing pages, and decides when to show ads — often inside a system with minimal transparency into why specific decisions were made. Performance Max campaigns now account for 45% of all Google Ads conversions, running across every Google surface simultaneously with limited human oversight.

We buy traffic through Google Search Ads across 9 LATAM markets. This is our daily working environment — not a platform we analyze from the outside. What follows is a breakdown of what Google changed, what the available data shows, and what we see working in accounts where real money is at stake. If you are also watching how AI reshaped the organic side of search, the paid side went through an equally radical transformation.

Manual Control Is Over — Here Is What Replaced It

Google Ads used to be a precision instrument. You chose exact keywords. You wrote every ad. You set manual bids. You decided which search queries triggered your spend and which ones got added to negative lists. That model is effectively dead.

The shift happened gradually but accelerated between 2024 and 2026. Google retired Broad Match Modifier back in 2021. Exact match expanded to include "same meaning" and "same intent" matching — your exact match keyword [personal loan] can now trigger on queries like "borrow money online" or "quick cash advance" without your explicit approval. The matching logic lives inside Google's AI. You can see some of the matched queries in the search terms report, but a growing percentage is hidden behind "other search terms" — queries Google matched to your campaigns but will not reveal.

Independent estimates from PPC agencies suggest 20 to 30 percent of clicks now come from queries that advertisers cannot see in their reports. Whether this invisible traffic converts profitably depends entirely on how well your conversion tracking and bidding signals guide the algorithm. Without good data flowing back, the AI optimizes for click volume, not business outcomes.

The result: the competitive advantage in Google Ads shifted from keyword selection skills to data infrastructure. The advertiser who feeds Google the best conversion signals wins, regardless of whether their keyword list is better than the competitor's. This is a fundamentally different game than what paid search was five years ago.

Broad Match as Default: What That Means in Practice

In July 2024, broad match became the default match type for new Search campaigns when using Smart Bidding. Not a recommendation. The default. New campaigns start with broad match unless the advertiser actively changes it.

Google's published data says broad match paired with Smart Bidding delivers 10% more conversions than manual approaches. That number comes from Google — which has an obvious interest in advertisers using broad match, since it increases the number of queries their ads serve on (and therefore the total ad spend).

The counterpoint from agencies managing large accounts: broad match opens the floodgates to irrelevant query matches. Multiple PPC management firms report up to 15% budget waste from queries that look semantically related to Google's AI but have zero commercial intent for the specific advertiser. "Personal loan calculator" is semantically close to "personal loan application" — but the user intent is completely different. One is research, the other is ready to apply.

Brand controls help. Brand inclusions limit broad match traffic to queries containing specific brand names. Brand exclusions prevent your ads from appearing on competitor brand queries (or vice versa). These rolled out across all match types and Dynamic Search Ads during 2024. They don't solve the relevance problem entirely, but they stop the worst cases of brand confusion.

The practical approach we settled on: broad match for campaigns with strong offline conversion data feeding back into Smart Bidding. Phrase and exact match for campaigns where the conversion signal is weaker (form fills without downstream quality data). Broad match works when the AI knows what a good conversion looks like. Without that signal, broad match just means broad spending.

AI Max: Keywordless Search Campaigns

AI Max launched in May 2025 as a suite of AI features for Search campaigns. By April 2026, it became available to all advertisers globally. It represents Google's clearest statement about the future of Search advertising: a future that does not revolve around keywords.

Three components define AI Max. Search term matching uses broad match and what Google calls "keywordless" technology — the system analyzes your landing pages, your existing ad copy, and your conversion data to decide which queries are relevant, whether or not you listed those queries as keywords. Asset optimization dynamically generates headlines and descriptions using Gemini, Google's language model. Your ads become fluid: the text changes per query, per user, per device. Text guidelines, added to all advertisers in April 2026, let you set guardrails around tone, brand language, and specific phrases to include or avoid. This last feature was Google's response to widespread complaints about AI-generated copy that violated brand voice.

Google's performance data: campaigns using AI Max see an average 14% lift in conversions at similar CPA or ROAS. For accounts that previously relied exclusively on exact and phrase match keywords, the lift reportedly reaches 27%. Those are Google's numbers. They count all conversions equally.

That is the catch. AI Max treats a qualified lead and a junk form fill as the same event unless you tell it otherwise. The 14% "lift" can include low-quality conversions that inflate the number without improving business outcomes. Without offline conversion tracking or a lead-quality signal feeding back into the bidding system, AI Max optimizes for quantity. Which is exactly what Google's conversion count rewards.

Smart Bidding Exploration and Flexible ROAS

Google described Smart Bidding Exploration as their biggest bidding update in over a decade. The core idea: flexible ROAS targets that allow the AI to test less obvious query categories — ones that historically did not convert but might at higher scale or with different creative.

The numbers from Google: 18% increase in unique search query categories with conversions, and 19% increase in total conversions for advertisers using the feature. Those are significant lifts if they hold up across diverse accounts.

The "exploration" framing matters. Exploration means spending money on unproven queries. The AI allocates part of your budget to queries where it has no conversion history, betting that at sufficient volume, patterns will emerge. Sometimes that finds profitable pockets of demand that manual keyword research would never have identified. Sometimes it finds that people searching "what is a personal loan" are not the same people who want one.

Early adopters report that Exploration works best when the account has a large conversion dataset (hundreds per month, not dozens) and when the conversion event is clearly defined. For accounts with thin conversion data or fuzzy conversion definitions (like "time on site" proxies), the exploration phase burns through budget before the AI learns anything useful.

Our take: Smart Bidding Exploration is a tool for scaling proven campaigns, not for launching new ones. Get the campaign profitable first with tighter controls, then turn on Exploration to find adjacent query categories the AI can exploit. Starting with Exploration on day one is expensive experimentation.

Performance Max at 45% of All Conversions

Performance Max went from optional to dominant in under three years. As of early 2026, PMax campaigns account for 45% of all Google Ads conversions across the platform. That was closer to 15% in 2023. Google pushed PMax aggressively — as a campaign type, as a recommendation inside the platform, and through account representatives who routinely suggest migrating existing Search campaigns into PMax.

PMax runs ads across every Google surface: Search, Display, YouTube, Gmail, Discover, Maps. You provide creative assets, a budget, and a conversion goal. The AI handles everything else. The average CPC for PMax is $1.40, roughly a third of Search campaign costs, which partly explains the adoption numbers — advertisers looking at cost-per-conversion tables see PMax winning, even though the conversion quality from Display and Discover placements is often lower than from Search.

The improvements since launch addressed real complaints. Campaign-level negative keywords finally rolled out to all advertisers — a feature the PPC community had demanded since PMax launched. Brand exclusions for different formats let retailers prevent branded inventory from being claimed by PMax. Search themes (up to 25 per asset group) give advertisers directional input on which queries to pursue. Improved reporting now shows channel-level breakdowns: how much spend went to Search vs YouTube vs Display.

What PMax still does not offer: full search term visibility, individual bid control per keyword, the ability to exclude specific placements beyond broad categories, or transparency into why the algorithm chose one creative combination over another. For advertisers who need precise control over brand safety and message consistency, PMax remains a black box with strong aggregate numbers and opaque individual decisions.

Demand Gen: 3 Billion Users Across Video and Display

Demand Gen campaigns, Google's replacement for Discovery campaigns, reach over 3 billion monthly users across YouTube, YouTube Shorts, Discover, and Gmail. This is the upper-funnel play: showing video and image ads to users who are not actively searching for your product but match the behavioral and demographic profile of your existing customers.

Product feeds within Demand Gen drive a reported 20% increase in conversions when using target ROAS bidding. Target CPC bidding was added for cross-platform performance optimization. New customer acquisition goals show an 11.5% improvement in new customer ratio for accounts that implement them — meaning Demand Gen can be told to prioritize reaching people who have never interacted with you before, rather than remarketing to existing visitors.

AI-generated video assets, optimized for YouTube reach and viewer experience, were added in late 2025. The quality is surprisingly usable for direct-response advertisers — not brand-campaign quality, but good enough for performance ads that need volume across Shorts and in-feed placements.

For media buying operations that rely on Search for bottom-funnel conversions, Demand Gen serves a different purpose: it fills the top of the funnel with awareness and consideration traffic that Search campaigns can then capture. The two campaign types work in sequence, not in competition. Running Search alone means you only capture demand that already exists. Demand Gen can create it — at a higher cost per conversion, but with incremental volume that Search alone would never reach.

CPC Inflation by Industry: The 2026 Benchmarks

87% of industries saw cost-per-click increases in 2025. The direction is nearly universal. Costs went up.

Legal leads the inflation: $6.75 to $9.21 average CPC with spikes above $15 for high-intent queries, up 8% year over year. Finance and insurance follow at $4.85 to $7.40, up 6%. B2B and SaaS sit at $2.30 to $5.26, up 9% — the fastest growth rate among major verticals. Healthcare holds at $3.20 average, up 7%. E-commerce remains the most affordable at $0.82 to $1.80 average CPC, but competitive categories within retail are creeping upward.

Cross-campaign averages: Display at $0.65 CPC. Shopping at $0.95. PMax at $1.40. Search at $2.10 to $5.26 depending on who measures and how. The cost hierarchy reflects the intent hierarchy — Search captures active demand and costs accordingly, Display builds awareness cheaply, PMax blends everything at a middle price point.

Geographic variation adds a critical dimension. The same keyword can cost 3 to 4 times more depending on location. A "personal loan" click in the US versus the same click in Colombia is a different economic equation entirely. For operations like ours that buy traffic in LATAM markets, the CPC gap between Latin America and North America or Western Europe remains substantial — but it is narrowing year over year. Markets that offered arbitrage opportunities three years ago are becoming conventionally priced as more advertisers discover them. The window of low competition in emerging markets does not stay open forever.

What Separates Accounts That Perform from Those That Bleed

Accept the automation, but feed it better data. That is the one-sentence version of what works in Google Ads right now. Fighting the platform's direction — insisting on manual bids, exact match exclusively, no PMax — puts you at a structural disadvantage because Google's systems are increasingly built around their AI tools. Opting out means opting out of the features receiving the most engineering investment.

Offline conversion tracking is the single highest-leverage improvement most accounts can make. When AI Max or PMax optimizes toward a conversion event, the quality of that event determines whether the AI learns to find valuable prospects or cheap clicks. Feeding actual sale data, lead qualification scores, or customer lifetime value back into the bidding system steers the algorithm toward what matters to your business — not just what generates form fills. Advertisers running offline conversion imports consistently report lower CPAs on qualified leads, even when their raw dashboard conversion volume appears lower.

First-party data is the targeting advantage that cannot be commoditized. Customer Match lists, remarketing audiences from your own properties, and CRM conversion data are the primary signals that differentiate your targeting from a competitor bidding on the same keywords with the same budget. Two advertisers with identical keywords and identical budgets will get materially different results if one feeds Google rich first-party audience data and the other relies on default targeting.

Creative volume wins in an AI-generated ad environment. AI Max and PMax test creative combinations at a speed no human team can match. But the system can only test what you supply. Accounts providing 15 headlines, 4 descriptions, multiple images, and video assets give the AI more raw material — and the performance gap between asset-rich and asset-poor campaigns widens as the platform's testing capabilities improve.

The pattern among accounts that are performing well: they stopped trying to control every variable and focused on controlling the inputs that matter most. Better conversion data. Stronger first-party audiences. More creative assets. Clear brand guidelines in AI Max. Tight negative keyword lists in PMax. They gave the AI better raw materials rather than fighting to do the AI's job manually. Those who fully surrendered control (default PMax settings, no offline data) or fully resisted automation (exact match only, manual bids) both underperform the middle path. Google Ads in 2026 rewards advertisers who collaborate with the algorithm — providing direction, data, and guardrails — rather than trying to replace it or ignore it.