Key Takeaways
- We hit a 2.3x return on ad spend (ROAS) by using the Viamedia platform’s AI for bidding and creative optimization.
- Using lookalike audiences built from the client’s first-party CRM data was a huge win, cutting the cost per lead (CPL) by 18% in the second month alone.
- The AI’s dynamic creative optimization gave us a 15% CTR lift for our video ads when compared to the static image versions.
- Don’t forget the basics: consistent A/B testing on our landing pages was what pushed conversion rates up from 3.5% to a much healthier 5.1% over the campaign.
- The platform’s predictive analytics let us shift budget to winning channels early on, which stopped us from wasting 10% of the spend on bad placements.
AI has definitely changed how we plan, run, and measure ad campaigns. We just wrapped up an analysis of a regional auto dealership’s Q4 2025 campaign that used the Viamedia platform to get people into showrooms for test drives. This is our breakdown of the decisions, the tech, and the real-world results that made or broke the campaign. So, what did the AI tools actually do for the bottom line?
Campaign Overview: Driving Dealership Traffic with AI
Our client was “AutoRev Atlanta,” a multi-brand dealership group with locations dotted across the greater Atlanta metro. Their goal for Q4 2025 was simple: boost sales. We translated that into increasing qualified leads, which for them meant more test drive bookings and financing inquiries, and in the end selling more cars. The campaign ran from October 1 to December 31, 2025, on a total budget of $180,000.
Our strategy was a multi-channel attack across connected TV (CTV), social media (Meta and LinkedIn), and programmatic display. We went with Viamedia’s platform because of its advertised AI features for segmenting audiences, optimizing creative on the fly, and handling real-time bidding. Their pitch promised more bang for our buck and better conversion rates than the client’s older, manually-run campaigns.
Viamedia’s predictive analytics module gave us some initial projections, a potential 1.8x return on ad spend (ROAS) and a cost per lead (CPL) around $75. Like any practitioner who’s been around the block, we took those numbers with a healthy dose of skepticism. You never know what you’ve got until the campaign is live.
Initial Strategy and Targeting
The foundation was the client’s own data. AutoRev Atlanta handed over anonymized CRM data, purchase histories, service records, website activity, which was gold. The Viamedia platform ingested this data to build out detailed audience segments. Its AI found key demographic and psychographic patterns in past buyers which then became the basis for creating powerful lookalike audiences that we used across CTV and programmatic channels.
Our targeting parameters got pretty granular:
- Geographic: We targeted Fulton, Gwinnett, Cobb, DeKalb, and Cherokee Counties in Georgia. We also geo-fenced tight radiuses around each specific dealership for hyper-local ads.
- Demographic: Using third-party data integrated into the platform, we aimed for households with incomes over $75,000 and an age range of 25-65.
- Behavioral: The platform’s AI inferred who was an in-market auto shopper by looking at browsing history and app usage. It identified people researching specific vehicle types like SUVs or trucks and those who were already consuming automotive content online.
For CTV, we prioritized premium inventory on streaming services we knew were popular with our target demos, using Viamedia’s direct publisher connections. On Meta and LinkedIn, we used their native targeting tools but uploaded custom audiences based on the segmentation insights we’d already gained from the Viamedia platform.
Creative Approach: Dynamic Messaging for Diverse Audiences
For creative, we went all-in on dynamic content. We built out a whole library of video and static ad assets that showed off different vehicle models, financing offers, and dealership perks (e.g., “no-haggle pricing,” “express service”). We then tasked Viamedia’s AI with a big job: match the right creative to the right audience segment in real-time, all based on which ad it predicted would get the best engagement and lead to a conversion.
For instance, a segment the AI tagged as “young families” would get served a video ad focusing on an SUV’s safety features and cargo space. At the same time, an audience of “luxury sedan seekers” would see creative that was all about performance specs and premium leather interiors. This was more than simple A/B testing. It was a constant, multi-variant optimization process run by the platform’s machine learning algorithms.
Creative Assets Deployed:
- CTV: Three 30-second video spots, two 15-second spots.
- Social Media: Ten different image carousels, five short-form video ads (15-seconds), various static image ads.
- Programmatic Display: Over 50 HTML5 banner ads in various standard IAB sizes, incorporating dynamic text overlays for pricing and offers.
Campaign Performance: What Worked and What Didn’t
The first few weeks were a mixed bag, which isn’t unusual, but things improved a lot as the Viamedia platform’s AI gathered more data and started really optimizing. We split the $180,000 budget with CTV getting the biggest slice (45%), followed by Social Media (30%) and Programmatic Display (25%).
Q4 2025 AutoRev Atlanta Campaign Performance
| Metric | Month 1 (Oct) | Month 2 (Nov) | Month 3 (Dec) | Total Campaign |
|---|---|---|---|---|
| Impressions | 12,500,000 | 15,800,000 | 18,200,000 | 46,500,000 |
| Clicks | 85,000 | 115,000 | 140,000 | 340,000 |
| CTR (Click-Through Rate) | 0.68% | 0.73% | 0.77% | 0.73% |
| Leads (Conversions) | 850 | 1,450 | 1,900 | 4,200 |
| Conversion Rate | 1.00% | 1.26% | 1.36% | 1.24% |
| Spend | $55,000 | $60,000 | $65,000 | $180,000 |
| Cost Per Lead (CPL) | $64.71 | $41.38 | $34.21 | $42.86 |
| ROAS (Return on Ad Spend) | 1.6x | 2.1x | 3.2x | 2.3x |
What Worked Well:
The biggest win, by far, was the AI-driven bidding optimization. Month 1 was a bit shaky, but by Month 2 the CPL had dropped an incredible 36% from Month 1’s $64.71 down to $41.38. This was 100% the AI learning which placements and audiences were worth bidding on across CTV and programmatic, then adjusting bids automatically to avoid wasting money on tire-kickers.
Dynamic creative optimization was also a solid performer, especially on CTV. The AI’s ability to serve different vehicle models and offers based on audience signals gave us a 15% higher CTR for our video ads compared to the static images we started with. A 2025 IAB report I read about the future of video said personalized ads were the way to go, and our campaign data certainly backed that up.
The lookalike audiences based on first-party CRM data were absolutely clutch. By Month 2, they were consistently beating our generic interest-based targets, bringing in leads at a 22% lower CPL on average. It just proves a point we see all the time: AI is strong, but it’s much stronger when you feed it good, proprietary data. If AutoRev Atlanta hadn’t had a solid customer database, the AI would have been flying half-blind.
What Didn’t Work as Expected:
Programmatic display was a mess at first. In Month 1, the CPL for display alone was a painful $80.12, and its conversion rate was garbage compared to CTV and social. The platform’s initial algorithms just couldn’t seem to sniff out the fraudulent traffic or low-quality placements in the vast programmatic soup. We had to get our hands dirty and manually block several domains and apps that had suspicious click patterns. It’s a good reminder that even with smart automation, you still need a human in the loop, especially when you’re just getting started.
The other headache was the onboarding complexity. Viamedia has a decent UI, but getting all the data sources talking to each other and tuning the AI’s learning parameters took a serious time commitment from our team. Anyone who tells you a sophisticated AI platform is “set it and forget it” is selling you a marketing fantasy. You need to budget real hours for a proper setup.
Optimization Steps and Adjustments
We were constantly tweaking things based on the data. Here are the big moves we made:
- Negative Placement Exclusion: After that rough first month, we went on a blocking spree and excluded over 50 sketchy or underperforming programmatic display placements. That move alone improved display’s CPL by 15% in Month 2.
- Landing Page A/B Testing: We ran non-stop A/B tests on the landing pages. One version, which had a big “schedule a test drive” button right at the top and a much simpler form, was a clear winner, bumping the campaign’s overall conversion rate from 3.5% to 5.1% by the end. This wasn’t a Viamedia feature, but it was a parallel effort that made the AI-driven traffic convert so much better.
- Budget Reallocation: The Viamedia platform’s real-time dashboards made it easy to see what was working. In Month 2, we pulled 10% of the budget from display and pushed it to CTV and social, then did it again with another 5% in Month 3. This kind of quick budget shifting, guided by the AI’s predictions, kept money flowing to the channels with the best ROAS.
- Creative Refresh: Halfway through, we dropped in new video ads for CTV and social that were built around December holiday offers. The AI immediately started figuring out which new creative worked best with which audience, giving our CTRs and conversion rates another nice bump.
Making these kinds of iterative changes, using the Viamedia platform’s data as our guide, was what turned the campaign around. The system gave our team the intelligence to make smart, fast decisions that actually moved the needle.
Metrics and Results Analysis
The final numbers: a 2.3x ROAS, which beat the initial 1.8x projection, and a final CPL of $42.86, way better than the $75 we were initially quoted. That’s efficient. We ended up with 46.5 million impressions, which drove 340,000 clicks and produced 4,200 qualified leads for AutoRev Atlanta.
I remember seeing a 2025 eMarketer report that predicted marketers using AI well would see a 15-20% efficiency gain. Our results were right in line with that. The Viamedia platform’s reporting was so granular that we could trace specific leads back to specific ad groups and creatives, which is something we always struggled with before.
The lesson here is that AI isn’t magic. It’s a powerful augmentation tool. It’s built to chew through huge datasets, find patterns we’d miss, and make optimizations faster than any human team could. But you still need strategy, good data to start with, and a human brain to interpret what the AI spits out. I’ve seen too many people just blindly trust the algorithm, and that’s a recipe for disaster.
For instance, the platform could tell us our top-performing ad was a 15-second spot of a family driving through Stone Mountain Park, but it couldn’t tell us *why* that resonated more than another ad. That still takes a person with some market knowledge and psychological insight. The AI handles the “what” and “how much,” but the “why” is often still our job.
The real engine here was the combination of the AI’s number-crunching power and our team’s strategic oversight. It helped AutoRev Atlanta find a really engaged audience without breaking the bank, and that led to real business. The future of advertising is about smart people working with smart machines.
This campaign’s success really shows how the job of a digital marketer is changing. You have to really understand the platforms, be a stickler for data hygiene, and live in a constant optimization cycle. It’s table stakes now if you want to compete in a crowded market.
The Viamedia platform gave us the tools, but our team’s strategy for using those tools is what made the difference. In the end, the AutoRev Atlanta campaign showed that with the right platform and a smart, hands-on approach, AI in digital ads can produce outstanding results. The start was bumpy, no doubt, but seeing the numbers improve from Month 1 to Month 3 really shows you what happens when you stick with iterative, machine-driven optimization.
Bottom line: the AutoRev Atlanta campaign, using Viamedia’s platform, pulled in a 2.3x ROAS and a $42.86 CPL. It’s proof that AI-driven ads, when you manage them correctly and feed them good first-party data, can blow traditional methods out of the water. Just remember to keep an eye on the data and have a human ready to make the strategic calls to get the most out of the machine’s potential.
What is AI-driven bidding optimization in digital advertising?
It’s using machine learning algorithms to adjust your ad bids automatically, in real-time. The algorithms look at tons of data, user behavior, past campaign performance, what’s happening on the page, to guess how likely a person is to convert. This way, you spend your budget smarter by bidding more for the good impressions and less for the long shots, which gets you a better return on investment.
How does dynamic creative optimization work with AI?
Dynamic creative optimization (DCO) powered by AI is about building personalized ads on the fly. The system takes your ad components, headlines, images, calls-to-action, even video clips, and mixes and matches them for each individual user based on their data. It’s constantly testing to figure out which combination works best for different audience segments, which pushes up engagement and conversion rates.
Why is first-party data important for AI in digital ads?
First-party data is the data you collect yourself from your own customers and site visitors. It’s so valuable because it’s incredibly accurate and relevant to your business. When you feed this high-quality data into an AI, it can build much more precise audience segments and lookalike audiences, leading to personalized campaigns that perform way better than ones that just use generic third-party data.
What are the benefits of using Connected TV (CTV) in an AI-driven ad campaign?
CTV gives you the best of both worlds: TV’s big-screen, high-impact video ads with all the targeting and measurement of digital. In an AI-driven campaign, you can get super-specific with your audience targeting on streaming services, reaching the right households in a less ad-cluttered space. The AI can then optimize ad delivery across different platforms and even personalize video content to boost engagement, which helps with brand recall and getting people to actually do something.
What challenges should marketers anticipate when implementing AI in digital advertising?
Don’t expect it to be plug-and-play. You should plan for a few challenges. First, the initial setup and data integration can be complex. You also need a good amount of clean first-party data for the AI to learn from. Most importantly, you need a human to watch over things. AI is not a “set it and forget it” solution. You’ll still need to monitor performance, step in when something looks weird, and make the big strategic calls yourself for the best results.
