MTN’s 2025 CTV ROAS: A 1.8x Success Story

Proving that your Connected TV (CTV) spend is actually driving sales is a constant battle. MTN, a major telco, went right at this problem in its Q3 2025 promotional push for a new fiber internet service. Their playbook for CTV attribution modeling is worth a look for any brand that’s trying to measure the whole customer journey, not just the final click.

The Campaign: Launching Fiber in Competitive Markets

MTN’s goal was simple: sell its new high-speed fiber internet in Atlanta, Dallas, and Denver. Anyone who’s worked in telco knows these are brutal markets, packed with incumbents and low-ball offers. They put a $1.2 million media budget into a 10-week campaign, running from July 15 to September 23, 2025, and split the money across CTV, programmatic display, and paid search.

Strategy & Objectives

The plan was a classic full-funnel approach. They used CTV to get on people’s radar and build some interest, then hit those same audiences with display ads for retargeting, and finally used paid search to grab people who were ready to buy. The specific goals were:

• Hit a 1.5x Return on Ad Spend (ROAS) or better on the CTV portion.
• Bring in at least 15,000 qualified leads, which they defined as someone completing a service availability check on their site.
• Keep the Cost Per Lead (CPL) under the $80 mark.

Creative Approach: Speed, Reliability, and Local Relevance

For creative, MTN produced a 30-second spot and a 15-second version. Both ads hammered home the speed and reliability of fiber over old-school broadband. They did something smart, too, they localized the ads with shots of landmarks in Atlanta’s Midtown, Dallas’s Uptown, and Denver’s LoDo, and even used voiceovers with subtle regional accents to connect better. Every ad had a clear call to action: a vanity URL (like MTNfiber.com/Atlanta) and a QR code that sent people to the right market’s landing page, giving them at least some direct tracking from the TV screen.

Targeting: Precision at the Household Level

Their targeting was surgical. They started with lookalikes from their own first-party CRM data of high-value customers, layered on third-party segments like high-income households and cord-cutters, and got super specific with geo-targeting. Instead of just hitting zip codes, they used their DSP, The Trade Desk, to do household IP matching inside specific Nielsen Designated Market Areas (DMAs). That meant they could serve ads right to the smart TVs of eligible homes and cut down on waste.

Attribution Model: Beyond Last-Touch

Last-click attribution is basically useless for a top-of-funnel play like CTV, and MTN knew it. They built their own weighted model to give credit where it was due, based on the simple fact that a CTV ad starts the conversation, it doesn’t usually end it. Here’s how they broke down the credit for a conversion:

• CTV Impression (First Exposure): 40% credit
• Programmatic Display Click (Mid-Funnel): 20% credit
• Paid Search Click (Last-Touch): 30% credit
• Direct Visit (Post-Exposure): 10% credit

They wired this all together in their customer data platform (CDP), Segment, which pulled in data from their DSPs, ad servers, and CRM. Critically, they set the lookback window for CTV impressions to 30 days, because signing up for a new internet provider isn’t an impulse buy you make in an afternoon.

Campaign Performance: Metrics and Analysis

The results gave them some clear wins and a few lessons learned on how CTV actually works when you’ve got a proper attribution setup in place.

Overall Campaign Performance (10 Weeks)

• Total Budget: $1,200,000
• Total Impressions: 28,500,000 (CTV: 18M, Display: 9M, Search: 1.5M)
• Total Conversions (Qualified Leads): 16,800
• Overall CPL: $71.43
• Overall ROAS: 2.1x (based on estimated lifetime value of a fiber subscriber)

CTV Specific Performance

The CTV-specific slice of the campaign involved a $750,000 budget.

What Worked Well

1. Multi-Touch Attribution Validation: The custom model absolutely proved its worth. Of the 16,800 total leads, a whopping 12,400 (73.8%) had seen at least one CTV ad in their 30-day path to conversion. That’s the kind of hard data you can take to finance to justify the budget, and it’s something a last-click model would have completely missed.
2. Creative Length and CTA: The 30-second spots with QR codes were the clear winners. Looking at the data in Google Analytics 4, they saw that people who scanned a QR code converted 22% more often than people who just saw the ad and typed in the vanity URL later. Making it easy for people to act *right now* on their TV matters.
3. Audience Sequencing: The sequencing hypothesis paid off. People who saw a CTV ad and then got hit with a display retargeting ad converted at a 15% higher rate than people who only saw the CTV ads. It confirms the old-school funnel logic: CTV created the interest, and display nurtured it.
4. Geo-Targeting Accuracy: Using household IP matching in their target DMAs meant they cut wasted ad spend by about 18% compared to their previous campaigns that just used zip codes. This directly improved their effective reach in the areas they could actually service.

What Didn’t Work as Expected

1. Direct Site Traffic from CTV: People just didn’t type in the vanity URLs. The traffic from people seeing the ad and then working through directly to the site was way lower than they’d hoped. This just goes to show that CTV isn’t a direct response channel like search. The QR codes definitely helped, but they didn’t make up the entire difference.
2. Short Attribution Windows: They almost made a big mistake here. Initially, the team thought about using a 14-day attribution window for CTV, but early data showed a huge chunk of conversions happening after day 10. Pushing the window out to 30 days captured an extra 8% of conversions which proved the point that people take their time deciding on a new internet provider.
3. Creative Fatigue with Static Ads: One of the 15-second spots started strong but then its view-through rate dropped by 10% after about five weeks. They chalked it up to creative fatigue. For longer campaigns, you’ve got to have more ads in rotation or plan for a mid-flight creative refresh.

Optimization Steps Taken

Because they were watching the data constantly, the MTN team was able to make some smart adjustments on the fly:

1. Increased QR Code Prominence: They made the QR code bigger and kept it on screen for 3 more seconds in the 30-second ads. This simple change led to a 7% bump in direct scans.
2. Dynamic Creative Optimization (DCO) for Display: Their display retargeting got smarter. Ads were updated with dynamic content, for instance, if you visited the pricing page, the next display ad you saw would feature a specific package deal.
3. Bid Adjustments by DMA: Atlanta was converting better and at a lower CPL than the other markets. So, halfway through the campaign, they shifted 15% of the budget from Denver over to Atlanta to double down on what was working.
4. Expanded Audience Segments: Mid-campaign, they tested a new “smart home device owners” third-party data segment. The theory was that these people would be more interested in fast fiber internet, and they were right. The segment converted 5% better than the general streaming enthusiasts, so they shifted more budget there.

Conclusion

MTN’s Q3 2025 campaign is a great real-world example of how to do CTV attribution modeling correctly. It’s not about last-touch. If you build a weighted, multi-touch model, get your targeting right, and actually watch the data to optimize creative and audiences, you can prove CTV’s value to the business and get a positive return on investment.