Pubcon 2026: 3.5x ROAS with AI-Powered PPC

Pubcon 2026 confirmed what many of us are already seeing in the trenches of SEO and PPC. So, we tore down one of our recent campaigns for a B2B SaaS client in the cybersecurity space to show what’s actually working when you’re up against a tight budget and fierce competition. This is the real story of how we used AI content and hyper-personalization in our ads to get solid returns.

Campaign Teardown: Elevating Cybersecurity Lead Generation

We had a six-month window to get results for “CyberGuard Pro,” a niche cybersecurity platform selling advanced threat detection to small and medium-sized businesses. The goal was simple: get more qualified leads. We were targeting IT managers and security pros at North American companies with 50 to 500 employees. Before we came in, the client was burning cash on high CPLs and getting unpredictable ROAS from campaigns that used broad keywords and one-size-fits-all landing pages.

Initial Strategy and Budget Allocation

The client handed us a $75,000 budget for six months. We decided to put about 60% into PPC and the other 40% into SEO content and technical work. A purely paid or purely organic approach just wasn’t going to cut it. Our target CPL was $150, and we were aiming for a 3.0x ROAS, which meant every dollar we spent needed to bring back three in revenue. This was a pretty big ask in an industry where CPLs can easily hit $250 for a decent lead.

SEO Strategy: Dominating the Search Generative Experience

For SEO, our whole game plan was built around the changes happening in search, specifically Google’s Search Generative Experience (SGE). We saw that for informational searches, the old “ten blue links” were disappearing. Why? Because SGE was just answering the question directly in the SERP.

Zero-Click Content and Structured Data

The heart of our SEO work was building what we call “zero-click content.” We used AI to help us generate authoritative articles and guides that were written to show up directly in SGE or featured snippets, answering a user’s question on the spot. We went after long-tail informational keywords that an IT manager would actually type in, like “how to implement zero trust architecture for SMBs” or “best practices for ransomware prevention 2026.” We wrapped every piece of content in Schema Markup, specifically `Article` and `FAQPage` schemas, to give Google every possible reason to feature us. For any step-by-step content, we also used `HowTo` schema. A recent HubSpot report notes that over 60% of searches are now zero-click, so we weren’t trying to fight it. We wanted to own that visibility and make CyberGuard Pro the go-to answer. Our content team, using AI tools like Copy.ai for drafting and Surfer SEO for on-page tuning, cranked out over 50 articles and 20 in-depth guides. We made sure the answers were detailed and accurate, even adding things like interactive calculators for security risk assessments to make the content stickier.

Technical SEO Enhancements

We couldn’t just throw content at the problem, so we started with a deep technical SEO audit. It wasn’t pretty. We found slow mobile page loads averaging 4.5 seconds, a messy internal linking setup, and a sitemap that looked like it hadn’t been touched in years. Our first job was to fix Core Web Vitals which meant cutting down server response time, crushing image sizes, and building an internal link structure that actually spread page authority around. We also made sure every new article was indexed almost immediately using optimized XML sitemaps and direct API pings to Google Search Console.

PPC Strategy: Hyper-Targeting and Dynamic Creative

On the PPC side, we stopped wasting money on broad match keywords and generic banner ads. The whole plan shifted to precision and automation.

Keyword Strategy and Bid Management

We zeroed in on exact and phrase match keywords showing clear buying intent, things like “cybersecurity solutions for small business” or “managed detection and response services.” We also built a list of “problem-aware” keywords, like “data breach recovery plan,” to catch people who were actively looking for a fix. For bidding, we let Google Ads’ Target ROAS bidding strategy do the heavy lifting, automatically adjusting bids based on real-time conversion value data. We configured it with a 2.8x minimum ROAS target to keep the algorithm on a tight leash. Of course, this only works if your conversion tracking is perfect, so we spent a lot of time in Google Tag Manager setting up distinct values for different actions, like a demo request being worth more than a simple whitepaper download.

Dynamic Creative Optimization (DCO)

This was the real key. We used a Dynamic Creative Optimization (DCO) platform, AdRoll in this case, to spin up hundreds of ad variations automatically. The system mixed and matched headlines, descriptions, and CTAs based on who the user was, where they’d been online, and what they were doing right now. For example, if an IT manager had looked at CyberGuard Pro’s page on threat intelligence, they’d get an ad talking about the “advanced threat intelligence platform.” If they were searching about encryption, the ad would talk about “secure data solutions.” Our visual assets were dynamic, too. We fed the DCO platform a library of over 100 images and short videos to combine with the personalized text. The results were strong.

Google Discover Campaigns

Google Discover felt like a new playground for B2B, but it ended up being a powerhouse for our retargeting. We designed visual, almost editorial-style ads that showed up in the Discover feeds of users who had already interacted with CyberGuard Pro’s content or visited related industry sites. Storytelling was the focus here, not a hard sell. We offered value and ideas, which worked surprisingly well on an audience that gets bombarded with sales pitches all day.

What Worked and What Didn’t

The campaign’s wins were built on a few key pillars, but we definitely hit some bumps along the way.

Successes:

• Zero-Click Content Dominance: The hard push on SGE optimization was a huge win. After three months, 70% of our organic traffic for info queries was coming from answers inside Google’s SGE or featured snippets. This put the brand name right in front of people and built authority before they even had to click, a big deal when Nielsen says this can increase brand recall by up to 25%.
• Reduced CPL: By tightening up our keyword targeting and letting DCO run, our average CPL for qualified leads dropped from their old average of $280 down to $123.50. That’s an 18% improvement on our own $150 target. For more on cost optimization, check out how Platform Power-Up SEO drives lower CPL.
• High ROAS: All our PPC work combined to produce a 3.5x ROAS, beating our 3.0x goal. Most of that was driven by the DCO’s ability to match the right ad to the right person at the right time.
• Google Discover Engagement: Those Discover campaigns punched way above their weight, getting an average CTR of 2.8%. That’s a lot better than the 1.9% we were seeing on our standard display retargeting, which really shows the power of putting good-looking, relevant content into someone’s personal feed.

Challenges and Learnings:

• Initial SGE Volatility: SGE was a roller coaster at first. Our rankings for SGE answers would jump around daily as Google tweaked its models. We learned we had to watch performance like a hawk and be ready to tweak content structure or schema at a moment’s notice.
• Creative Fatigue with DCO: DCO was great, but we found that creative burnout happened much faster than we expected in certain ad sets. We ended up having to double our production of new images and video clips halfway through the campaign to keep things fresh.
• Attribution Complexity: Trying to track conversions got messy. Between all the zero-click organic engagement and the multi-layered PPC, last-click attribution was useless. We had to build a much more sophisticated multi-touch attribution model in Google Analytics 4 (GA4), using its data-driven model to get a clearer picture of the entire customer journey.

Optimization Steps Taken

We were constantly tweaking things based on the data coming in. We didn’t just set it and forget it.

• A/B Testing Landing Pages: We were always running A/B tests on landing pages, playing with headlines, CTA button placement, and form length. Just by cutting the number of form fields by 20%, we saw a 15% jump in the conversion rate for demo requests.
• Negative Keyword Expansion: We were relentless about building out our negative keyword lists for PPC. Any search query that was wasting budget, like “free cybersecurity tools” or “personal antivirus review”, got shut down immediately.
• Audience Segmentation Refinement: In Google Ads and LinkedIn Ads, we were always refining our audiences. We used first-party data from the client’s CRM and our website to build lookalike audiences and exclude segments that weren’t performing, getting as granular as targeting users by specific job titles or even what cloud provider their company used.
• Content Refresh Cycles: We put our top-performing SGE content on a quarterly refresh cycle. Every three months, we’d go back in to update stats, add new information, and re-optimize the schema to make sure we kept our top spots.

Results and Metrics

Here’s the final scorecard:

Campaign Duration: 6 Months

Total Budget: $75,000

Total Impressions (PPC): 2,800,000

Total Clicks (PPC): 42,000

Average CTR (PPC): 1.5%

Total Conversions (Qualified Leads): 608

Average CPL (Cost Per Lead): $123.36

Total Attributed Revenue: $262,500

Overall ROAS: 3.5x

The final numbers showed that a tight integration between SGE-focused SEO and smart, dynamic PPC pays off, even in a tough B2B market like cybersecurity. Putting in the work to understand how search is changing and actually using the AI tools available was the difference-maker. For more on AI’s impact, see how SMBs See 72% AI Martech ROI. The future of this job is all about figuring out what someone wants and giving it to them, whether that’s an instant answer in a search result or an ad that feels like it was made just for them.