Editorial Team Key Takeaways
- Our case study showed AI bidding on Google Ads cut our lead gen Cost Per Conversion by 15%.
- Using our own customer lists (Customer Match) on Meta Ads delivered a 1.8x higher ROAS than using lookalike audiences by themselves.
- Switching off last-click attribution revealed that our display ads were actually driving 20% more conversions than we’d been giving them credit for.
- Hooking our CRM data directly into ad platforms for dynamic segmentation gave our retargeting audiences a 10% lift in CTR.
Everyone talks about how martech has changed paid media. It’s true that better analytics, AI automation, and new audience tools have completely changed how we plan and run campaigns. But what does that actually look like on the ground, away from the conference slides and abstract whitepapers?
Campaign Teardown: “Project Nexus” for a B2B SaaS Provider
We had a client, a mid-sized B2B SaaS company in the project management space, who hit a wall in late 2025. They needed more leads but couldn’t let their Cost Per Lead (CPL) get out of control. Their existing strategy, which was almost all Google Search and LinkedIn, had just stopped scaling. We came in to overhaul the whole thing, plugging in a bunch of newer martech to get things moving again.
Strategy & Objectives
The main goal for “Project Nexus” was simple: get more Marketing Qualified Leads (MQLs) for their enterprise product. We also wanted to get their brand name out there in a few key industries and, critically, bring their overall CPL down by 10% in six months. We set a hard target of $120 per MQL. The whole project ran from October 2025 to March 2026.
- Budget: $150,000 over six months ($25,000 per month)
- Target CPL: $120
- Target ROAS: N/A (lead generation focus, but MQL-to-SQL conversion rate was tracked internally)
- Platforms: Google Ads (Search, Display, YouTube), LinkedIn Ads, and a programmatic display platform (The Trade Desk).
Creative Approach: Beyond the Whitepaper
This client had been leaning on downloadable whitepapers for years. For Project Nexus, we had to diversify. We created a set of short, animated explainer videos for YouTube and programmatic display that tackled specific customer pain points. On LinkedIn, we ran carousel ads that showed off actual product features and pushed for a personalized demo. Our search ads still used problem-solution keywords, but we A/B tested them against headlines that focused more on benefits. We even built an interactive self-assessment tool on a landing page, and it turned out to be a surprisingly good way to get people engaged.
Targeting & Audience Segmentation
This is where the tech really started to pay off. We piped the client’s Salesforce data straight into Google Ads and LinkedIn Ads. This one move let us do a few powerful things:
- Exclude existing customers and recent opportunities: We immediately stopped wasting money showing ads to people who had already bought or were deep in the sales pipeline.
- Create highly specific custom audiences: We could, for instance, take a list of everyone who attended a webinar in the last year and retarget them with a very specific “next step” offer.
- Use dynamic audience segments: People who visited the pricing page but didn’t sign up saw different ads from people who just read a few blog posts. The system automatically segmented them based on their on-site behavior.
In Google Ads, we used Custom Segments to go after users who were actively searching for the competition’s products or for key industry problems. Over on LinkedIn, we layered standard filters like Job Title, Seniority, and Company Size on top of our Matched Audiences, which we built from a target account list provided by the client’s own sales team.
What Worked: Data-Driven Successes
1. AI-Driven Bidding on Google Ads
We switched their Google Search campaigns from manual bidding over to Target CPA and eventually to Maximize Conversions with a Target CPA. The client was nervous about giving up control, but that didn’t last long. In the first three months, the average Cost Per Conversion for a demo request dropped from $155 to $132, a 15% improvement. The algorithm, fed with a steady stream of clean conversion data from our CRM integration, got incredibly good at finding the right bid. You can’t just set a target and walk away. The algorithm needs that constant, clean data feed to work. Otherwise, even the best AI is just guessing.
Metric Snapshot (Google Search, Q1 2026):
- Budget Allocated: $60,000
- Impressions: 1,200,000
- Clicks: 48,000
- CTR: 4.0%
- Conversions (Demo Requests): 455
- Cost Per Conversion: $131.87
2. First-Party Data Activation on LinkedIn
Using Matched Audiences on LinkedIn was a huge win, especially when we uploaded their own customer and prospect lists. Our retargeting campaigns aimed at users who’d seen our content but hadn’t converted pulled a 1.8% CTR and a 7.2% Conversion Rate for demo sign-ups. That blew our generic interest-based targeting (0.6% CTR, 2.1% Conversion Rate) out of the water. We also built lookalike audiences from these high-value lists, and they performed 1.2x better than lookalikes built from generic website visitors. The takeaway was obvious: our first-party data was our most valuable asset.
Metric Snapshot (LinkedIn Ads, Q1 2026 – Matched Audiences):
- Budget Allocated: $45,000
- Impressions: 950,000
- Clicks: 17,100
- CTR: 1.8%
- Conversions (Demo Requests): 1,231
- Cost Per Conversion: $36.56 (Note: LinkedIn conversions here included whitepaper downloads, which were a lower-friction conversion than demo requests)
3. Programmatic Display & Video for Brand Awareness
Our programmatic campaigns on The Trade Desk weren’t about getting leads today. We used third-party data to target IT decision-makers and business leaders to simply get the client’s name out there. It worked. We saw a 15% lift in branded search volume for the client’s name during the campaign, which we tracked in Google Keyword Planner. The explainer videos had an average View-Through Rate (VTR) of 65% for the first 15 seconds, which is a great sign of engagement. This kind of awareness work made our direct response ads more effective later on. Thinking of display as just a direct-response channel is a huge miss.
Metric Snapshot (The Trade Desk, Q1 2026 – Video & Display):
- Budget Allocated: $30,000
- Video Impressions: 2,500,000
- Display Impressions: 3,800,000
- Video Completions (30s): 875,000
- Effective CPM (Video): $12.00
What Didn’t Work & Optimization Steps
1. Initial Landing Page Performance
Our first round of landing pages looked great, but they performed terribly. We saw bounce rates over 60% and conversion rates for demo requests under 3%. The problem was pretty clear: the value prop was buried below the fold and the form was asking for too much information. We started A/B testing with VWO, cut the form fields from 8 down to 4, and wrote a much clearer headline. That single change boosted conversion rates by 40% in two weeks. It’s a painful but necessary reminder that you can buy the best traffic in the world, but a bad landing page will kill your campaign every time.
2. Over-reliance on Broad Match Keywords
In the first month on Google Search, we tried using a lot of broad match keywords, hoping to find some new opportunities. We found a lot of impressions, but the CPL was 2.5x higher than what we were getting from phrase and exact match. We reversed course fast, tightened up our targeting, and built out a massive negative keyword list to stop the bleeding. That move alone cut our wasted spend by about 18% the next month.
3. Attribution Challenges
We started out using last-click attribution which made Google Search look like the hero of the entire campaign. But once we switched to Google’s data-driven attribution model and checked it against our CRM’s multi-touch reports, a different story emerged. Our programmatic display and YouTube ads were doing a lot of heavy lifting early in the customer journey. For example, display ads showed up in 20% more conversion paths than last-click was telling us. That insight convinced us to move 10% of the budget away from pure search and into upper-funnel video and display for the second half of the campaign.
Overall Campaign Performance (October 2025 – March 2026)
At the end of the six months, Project Nexus had beaten all of our main goals.
- Total Budget Spent: $150,000
- Total Impressions: 8,450,000
- Total Clicks: 125,000
- Overall CTR: 1.48%
- Total MQLs Generated: 1,420
- Average Cost Per MQL: $105.63 (12% below target)
Better yet, the client told us their MQL-to-SQL conversion rate went up by 25% for the leads coming from this campaign. This was the real proof that our refined targeting and messaging were working. A 25% jump in the MQL-to-SQL rate proved we were sending the sales team better leads, which is the actual goal, not just hitting a raw CPL number.
The Future of Paid Media: Integration is Key
What this campaign really proved is that the effect of new martech isn’t about one specific tool. It’s about getting all these different systems, your CRM, your bidding AI, your attribution models, to talk to each other. When you can connect the dots from a first touchpoint on a display ad all the way to a closed deal in Salesforce, that’s when you stop just spending budget and start driving real results. That’s the work now.
How does CRM integration directly improve paid media performance?
It lets you use your own customer data inside the ad platforms. You can exclude current customers so you’re not wasting money on them, retarget hot prospects who aren’t yet customers, and build lookalike audiences to find more people just like your best customers. It means your ads are way more relevant which cuts waste and boosts conversions.
What is data-driven attribution and why is it important for modern campaigns?
It’s a model that uses machine learning to figure out how much credit each ad touchpoint should get for a conversion. Instead of giving 100% of the credit to the last ad someone clicked, it looks at the whole journey. This gives you a much truer picture of how channels like display and video are helping, even if they aren’t the last click, so you can fund them properly.
Can AI bidding strategies completely replace manual bid management?
No, it can’t completely replace a person. AI strategies like Google’s Target CPA are amazing for real-time bid adjustments, but a human still needs to set the strategy. You have to define the goals, make sure the conversion data going into the machine is accurate, and know when to step in if something looks wrong or if business goals change. Think of it as a very smart tool that still needs a skilled operator.
What are Custom Segments in Google Ads and how can they be used effectively?
They let you target people on Google’s Display and YouTube networks based on what they’ve recently searched for on Google, which apps they use, or what sites they’ve visited. For a B2B company, you could build a segment of people who recently searched for your top three competitors. It’s a powerful way to get in front of people with high intent.
How often should landing pages be A/B tested for paid media campaigns?
You should always be testing. It’s not a one-and-done thing. As long as a campaign is live, you should have an A/B test running on the landing page. How often you can run a new test really just depends on how much traffic you have, if you have tons of traffic, you can get a clear winner in a week and start a new test. With less traffic, it might take a few weeks. The key is to be testing something, always.
