Beyond keywords: Mastering AI-driven campaigns

The days of building campaigns around long lists of keywords are fading. Today, AI-powered Google campaigns and features like Performance Max (PMax) and AI Max are changing the rules.

These keywordless campaigns lean on automation, audience signals, and machine learning to find new opportunities, often faster and at greater scale than humans can.

At SMX Next, three PPC pros — Nikki Kuhlman, VP of search at Jumpfly; Brad Geddes, founder of Adalysis; and Christine Zirnheld, director of lead gen at Cypress North — explained where PMax and AI Max fit into your broader campaign strategy, where humans still make the difference, and how to strike the right balance between automation and control.

AI Max for Search: Best practices and what not to do

AI Max for Search is not a new campaign type. It’s a one-click opt-in setting within existing Search campaigns.

Without requiring you to switch to broad match, it expands your keywords — similar to broad match or Dynamic Search Ads — using your landing pages and other site assets. It then personalizes the ad copy and landing page the searcher sees.

The evolution from traditional setup

In the old setup, you might have used a keyword like “skincare for dry sensitive skin” that sent users to a moisturizer page with generic ad copy because you couldn’t capture every variation. With Google’s current matching, a specific ad group no longer guarantees that keyword will trigger that ad group.

AI Max for Search addresses this by generating ad copy based on the search query, making it more relevant and directing users to a landing page that better matches their needs.

Success with blog content

One area where AI Max for Search is seeing success beyond the norm is blog content. While DSA campaigns traditionally excluded blogs, AI Max for Search can now serve blogs as landing pages—and they’re converting. The key is that these blogs guide readers to specific products, not just general content.

The generated headlines are compelling and longer than what traditional RSAs allow, creating a more engaging user experience.

Best Practices for AI Max for Search

Do:

• Use it on existing campaigns with history and data, not brand new campaigns
• Test it as a 50/50 experiment instead of an outright change
• Use it on brand campaigns with brand inclusion capabilities
• Apply it to campaigns not hitting budget that could use more volume
• Review landing pages and utilize URL exclusions (individual or rule-based)
• Use landing page inclusions at the ad group level
• Review search queries regularly and add negative search terms
• Enable both text customization and final URL expansion for maximum value
• Turn off AI Max at the ad group level when specific ad groups drive poor traffic

Don’t:

• Use it on brand new campaigns without data
• Change all campaigns at once without testing
• Use it on brand campaigns without name recognition or brand inclusion ability
• Apply it to budget-constrained campaigns
• Turn off both URL expansion and text customization — if you’re not using both features, stick with broad match and smart bidding
• Assume it works universally — test on individual campaigns

Your action plan

Week 1: Pick a search campaign to test (brand with brand inclusion, with budget capability, needing more volume). Review landing page URLs and add inclusions or exclusions.

Week 2: Review search queries and add negatives.

Week 3: Continue optimization and turn off AI Max at the ad group level as needed.

Experiment checklist:

• Ensure enough volume for a 50/50 experiment
• Give the experiment 6 weeks to 2 months
• Set up a custom experiment if you need to enable brand inclusion or update settings
• For one-click experiments, change confidence level to medium and turn off auto-apply

Match type performance: What the data shows

A comprehensive study analyzing over 16,000 campaigns revealed surprising insights about match type performance across different bidding strategies.

Match type basics

• Exact match: Should match only when the search term has the same intent as your keyword. Misspellings and word order haven’t mattered for years — focus on user intent.
• Phrase match: The search intent should match your keyword, but could have additional information around it, whether modifiers, phone numbers, or websites.
• Broad match: Shows for anything related to the search intent. The key difference is that broad match uses additional signals that exact and phrase don’t, such as content on the landing page, other keywords in the ad group, and, most powerfully, previous search history for that user.

Performance by bidding strategy

Max Conversion strategies (Max Conversions, Max Conversion Value):

Most campaigns using max bid strategies have under 30 conversions per month, giving machines limited data to work with. The findings:

• Exact match has the best click-through rates and conversion rates
• Broad match had the worst conversion rates but the best return on ad spend
• Broad match also had lower CPA than phrase match
• Phrase match performed worst overall

Recommendation: Start with exact match, then skip phrase match entirely and layer in broad match if you have more budget to spend.

Target Bid Strategies (Target CPA, Target ROAS):

Most campaigns using these strategies have over 30 conversions per month, with many at 50 or 100+, giving machines substantially more data. The findings:

• Exact match is again the best match type
• Phrase match comes second
• Broad match is third
• Phrase match performs better with more data

Recommendation: Start with exact match, layer in phrase match with more budget, then add broad match if additional budget is available.

The phrase match puzzle

Why does phrase match perform poorly with limited data but better with more data?

Broad match uses additional signals, particularly previous search queries, to determine bids. When conversion data is limited (under 30 conversions monthly), broad match’s ability to leverage previous search history makes it much stronger than phrase match.

However, with sufficient data (50–100+ conversions), Google can properly match phrase match keywords using machine-learning pattern matching.

Brand vs. non-brand considerations

When you combine brand and non-brand data, exact match becomes even more powerful, delivering significantly higher click-through rates, higher conversion rates, lower CPAs, and much higher return on ad spend. That’s why segmenting keywords by brand and non-brand is crucial when determining your match type strategy.

Ecommerce exception

For ecommerce companies, broad match (and sometimes phrase match) can produce higher average order values than exact match. When someone searches for a specific product, and you carry that exact item, conversion rates are high, but they’re usually buying a single product with a lower checkout value.

When shoppers haven’t decided on a product, they tend to match broader keywords and build larger carts — resulting in lower conversion rates but higher order values.

Performance Max for lead generation

There’s a common misconception that Performance Max only works for ecommerce and is too difficult for lead generation. That couldn’t be further from the truth.

The critical success factor

The biggest mistake you can make—one you should avoid entirely—is optimizing campaigns for form submissions alone. If you treat every form submission as your campaign goal, you’ll end up with spammy submissions and frustrated sales teams.

The solution: integrate your Google Ads account with your CRM and import bottom-of-funnel leads—sales-qualified leads (SQLs), marketing-qualified leads (MQLs), opportunities, or even customers if the sales cycle is short.

When you tell Google Ads what you actually want and set it as your campaign goal, Performance Max can cast a wide net while still bringing in qualified prospects.

Available controls for regulated industries

Performance Max has significantly more controls now than at launch, making it viable for highly regulated industries:

• Brand exclusions: Exclude all brand traffic from Performance Max campaigns
• Campaign-level negative keywords: Exclude unwanted search terms directly
• Search term reports: See what’s triggering your ads and exclude accordingly
• Channel reporting: View spending and performance across different networks
• Page feeds: Control where you send traffic on your site
• Final URL expansion toggle: Turn it off completely if needed
• Text enhancement controls: Optional feature that can be disabled entirely
• Text guidelines: Specify words to avoid (e.g., “discount” or “directory”)

Device control: The secret weapon for B2B

One of the most underutilized levers for B2B and regulated industries is device control, introduced at the beginning of 2025. You can turn off any device from your Performance Max campaign.

A B2B SaaS example demonstrates the impact: Before device segmentation in January, the account had 224 SQLs from desktop at an acceptable CPA, but 33 from mobile at $319 CPA (above goal). After creating separate mobile campaigns with more aggressive target CPAs, they achieved 190 desktop SQLs and 37 mobile SQLs in a shorter month, with mobile CPA dropping to $204 and overall Performance Max CPA declining from $238 to $204.

Real Performance Max results for B2B SaaS

Despite lower conversion rates from Performance Max compared to search campaigns (due to broader reach), the results speak for themselves. In September 2025, one B2B SaaS account achieved:

• Search Campaigns: 150 SQLs at $237 CPA
• Performance Max: 204 SQLs at $220 CPA

Performance Max cast a wider net with cheaper CPCs, bringing in not just more leads but more sales-qualified leads at a lower cost.

How they did it:

• Optimized for SQLs, not form submissions
• Set lower target CPAs in Performance Max than search (to control spend while casting wider net)
• Created separate campaigns for off-hours to control weekend spending
• Turned off final URL expansion and text enhancements (client preference)
• Implemented separate mobile and tablet campaigns with aggressive target CPAs

AI Max for Search in lead generation

AI Max for Search brings the power of Performance Max to the search network, where bottom-of-funnel intent is strongest. This is especially valuable for lead generation accounts that spend on other networks in Performance Max but don’t generate leads from them.

Early results: Higher ed financial services

A higher education financial client (loan products) showed promising early results:

Approved applications (primary KPI):

• Standard Search: 86 approved applications at $660 CPA
• AI Max: 70 approved applications at $579 CPA

AI Max brought in qualified leads cheaper despite the highly competitive keyword environment.

Down-funnel performance

Beyond the initial conversion action (soft credit check), AI Max showed superior performance throughout the funnel:

• 42% of AI Max form submissions resulted in soft pulls vs. 36% for standard search
• 9.9% of AI Max form submissions resulted in bookings vs. 5.58% for standard search

AI Max isn’t just bringing more qualified prospects at the top—lead quality remains higher throughout the entire funnel.

How they did it:

• Optimized for approved applications, not form submissions
• Set lower target CPAs in AI Max than standard search
• Used high-performing bottom-of-funnel keywords with broad match types
• Kept final URL expansion and text enhancements disabled (still worked well without them)

Win with AI without losing control

PPC success requires embracing AI-driven campaigns while maintaining strategic human oversight. Whether you use AI Max for Search, Performance Max for lead generation, or adjust match types based on bidding methods and data volume, the key is understanding how these tools work and applying best practices aligned with your business goals.

The data is clear: exact match remains powerful across scenarios, but phrase and broad match perform differently depending on bidding strategy and data volume. For lead generation, the game changer is optimizing for true bottom-of-funnel conversions rather than form submissions, combined with strategic device controls and proper campaign segmentation.

The future of PPC depends on knowing when — and how — to apply automation and control for maximum impact.