So, you’re running Connected TV (CTV) ads for your app, but you can’t actually tell if they’re working. Basic last-touch attribution is useless here, since it completely misses the winding path a user takes from their TV screen to their phone. To get a real grip on your CTV campaign performance in 2026, you need to stitch together your data sources and get a lot smarter about tracking how people behave across different devices.
Key Takeaways
- Get your data straight by connecting your CTV advertising platform, mobile measurement partner (MMP), and CRM. They have to talk to each other.
- Set your MMP’s view-through attribution window to at least 7 days for CTV impressions. People don’t install apps on their TV remote.
- Make sure your MMP sends conversion data back to your CTV ad platform using server-to-server (S2S) postbacks to close the data loop.
- Slice up your CTV audience data inside your ad platform by demographic, device type, and engagement to sharpen your targeting strategies.
- Constantly check for data gaps between your ad platform and MMP, because you’ll need to adjust bid strategies using a single set of unified performance numbers.
Step 1: Establishing a Unified Data Foundation
You’re flying blind if your CTV advertising platform, your mobile measurement partner (MMP), and your internal CRM system aren’t talking to each other. It’s that simple. Without connecting these pieces, you’re looking at fragmented reports that make it impossible to know if that expensive Roku campaign is actually driving installs or just burning cash.
1.1. Integrating Your Mobile Measurement Partner (MMP)
Your MMP, whether it’s AppsFlyer or Adjust, is the heart of your app tracking. First, make sure its SDK is correctly implemented and you’re tracking all the important in-app events like registrations, purchases, and subscriptions. For CTV, you then have to tweak specific settings to account for impression-based attribution.
- Access MMP Dashboard: Log into your MMP. In AppsFlyer, this is under Configuration > Integrated Partners. In Adjust, you’ll find it under Partner Ad Networks.
- Select CTV Partner: Find the CTV platform you’re spending money on, like Roku Advertising, Amazon Ads, or Samsung Ads.
- Configure Postbacks: In the partner settings, you have to turn on Install Postbacks and In-App Event Postbacks so conversion data gets sent back to the CTV platform where you manage your bids. Critically, make sure the postback type is set to Server-to-Server (S2S), which is far more reliable and secure than other methods, especially as privacy rules get tighter.
- Define Attribution Windows: Go to the attribution settings for that CTV partner and lengthen the View-Through Attribution (VTA) window. A 24-hour window works for some mobile formats, but it’s totally wrong for CTV. I recommend starting with a VTA window of at least 7 days, sometimes even 14, because users see an ad on their big screen, get distracted, and then might not actually install the app on their phone for several hours or even days.
Pro Tip: Always pass a device ID parameter in your postbacks if the CTV platform sends it (Roku’s RIDA and Amazon’s ADID are common examples). This gives your MMP a deterministic signal to work with, strengthening your attribution even when probabilistic methods are also being used.
1.2. Connecting Your CTV Advertising Platform
Getting conversion data sent *from* your MMP is one thing. You also have to ensure your CTV campaigns are sending the right tracking data *to* your MMP in the first place. The postback you just set up won’t work if the initial ad impression isn’t properly tagged.
- Campaign Tracking Parameters: When you’re building a campaign in the Roku Ad Manager or a similar platform (look under a tab like Campaigns > New Campaign > Tracking), you must append your MMP’s tracking parameters to your creative URLs. These are usually macros that dynamically insert the campaign ID, creative ID, and placement ID so you can analyze performance later.
- Impression Tracking Pixels: For view-through attribution to work at all, you need to implement your MMP’s impression tracking pixel. Your MMP’s integration guide for each CTV partner will have this pixel URL, and you’ll usually paste it into a field named “Third-Party Impression Tracker” when setting up your ad group.
Common Mistake: Forgetting to test the postbacks. After you’ve done all this, run a tiny test campaign. Then check both your MMP and CTV platform dashboards to see if the installs and events are showing up and matching. Finding a discrepancy here early will save you from making terrible optimization decisions based on bad data down the road.
Step 2: Implementing Advanced Attribution Models
Last-touch attribution is a liar when it comes to CTV. You have to move past this simplistic model to get a real sense of CTV’s value, which means properly accounting for multi-touch journeys where the TV ad is just one step.
2.1. Using Probabilistic Matching
You’re almost never going to get a clean, deterministic match between a user’s CTV device and their mobile phone because they operate in different ecosystems with no shared identifier. This is where probabilistic matching comes in. Your MMP uses a cluster of signals, IP address, device model, operating system, and the timestamps of the ad view and the app install, to make an educated guess that the two events were caused by the same household.
- Review MMP Settings: Inside your MMP dashboard, go find the Attribution Settings > Probabilistic Matching section. Confirm that it’s turned on and check if there are any sensitivity controls you can adjust.
- Cross-Device Graph Integration: To make those probabilistic guesses even more accurate, some MMPs let you plug into third-party cross-device graphs from companies like Tapad or LiveRamp. If your MMP supports this (check under Advanced Settings > Cross-Device Providers), enabling it can give you a much wider identity graph to match against.
Expected Outcome: When this is set up right, you’ll suddenly see a much more realistic number of installs attributed to your CTV campaigns. This gives you the confidence to actually allocate a real budget to CTV instead of treating it like a risky experiment.
2.2. Incorporating Incrementality Testing
Attribution tells you which campaigns get credit for a conversion. Incrementality testing tells you if your ads are actually causing conversions that wouldn’t have happened anyway. This is how you answer the CFO’s question about the true value of your CTV spend. The process involves creating a holdout group that doesn’t see your ads.
- Define Test Groups: Use the audience tools in your CTV ad platform to create two segments that are statistically identical in terms of demographics and behavior. This is often under a menu like Audiences > Custom Audiences > Create Segment. One will be your test group (sees the ads), and the other will be the control group (sees nothing).
- Campaign Setup: Launch your CTV campaigns targeting only the test group. You must also explicitly set the control group as an exclusion audience to keep the test clean. Then you just let it run and track installs and other key events for both groups for a set period, usually 4 to 6 weeks.
- Analyze Results: Once the test is over, you compare the conversion rate of the test group to the control group. The percentage difference is your incremental lift. As a 2024 Nielsen report pointed out, this kind of measurement is “the gold standard for understanding true campaign value.”
Pro Tip: Don’t run these tests too often or for too short a time. You need a big enough sample size and enough time for the delayed effects of CTV ads to show up in the data. For most established campaigns, running a quarterly incrementality test is a good cadence.
Step 3: Using First-Party Data for CTV Audiences
Stop treating CTV as just a brand-awareness play. Your own CRM data is your most powerful weapon for turning CTV into a performance channel and closing the loop between what users do on TV and in your app.
3.1. Uploading Customer Segments
You should be creating custom audiences from your own CRM data, think lists of churned users you want to win back, your highest-value purchasers, or people who haven’t finished your onboarding flow.
- Export CRM Data: Pull a list of user IDs from your CRM. This could be email addresses or phone numbers. For privacy, you must hash this data using a standard like SHA256 before you upload it anywhere.
- Upload to CTV Platform: Go to the audience manager in your CTV platform (e.g., in Roku Ad Manager, it’s Audiences > Custom Audiences > Upload Audience List) and upload your hashed list. The platform will then match these against its own user base to find those people on their TVs.
- Create Lookalike Audiences: This is where the real growth comes from. After your high-value customer list is uploaded and processed, you can ask the platform to create a lookalike audience, which finds a huge new pool of users who behave just like your best ones.
So many marketers fixate on pure acquisition and forget this step. Re-engaging your own users on CTV, especially for subscription apps trying to reduce churn, can be incredibly cheap and effective. Don’t leave this money on the table.
3.2. Suppressing Existing Users
This is basic campaign hygiene for app install campaigns. Why would you ever pay to show an “Install Now” ad to someone who already has your app on their phone?
- Create Suppression List: Just as you uploaded a list of customers to target, you should also upload a hashed list of all your current active users.
- Apply to Campaigns: When you set up a new app install campaign, navigate to the Targeting section and add this “Existing Users” list as an exclusion audience. This simple action ensures your acquisition budget is spent only on acquiring new users.
Expected Outcome: Your campaign efficiency will shoot up and your Cost Per Install (CPI) for new users will drop, simply because you stopped wasting impressions on people who have already converted.
Step 4: Advanced Reporting and Visualization
All this data is useless if it’s living in separate spreadsheets and platform dashboards. You have to pull everything from your MMP and CTV platform into one place to actually see what’s going on.
4.1. Building a Cross-Platform Dashboard
Use a BI tool like Looker Studio or Power BI to create a single dashboard that pulls everything together. Even a well-built spreadsheet is better than nothing.
- Data Connectors: Use your BI tool’s built-in connectors to pull data directly from your MMP’s API and your CTV platform’s reporting API. All the major players have APIs for this.
- Key Metrics: Your dashboard should combine top-of-funnel metrics from the CTV platform (impressions, clicks) with bottom-of-funnel outcomes from your MMP (attributed installs, CPI) and business metrics (Cost Per Activation, ROAS).
- Visualization: Use charts to track how metrics like attributed installs and CPI trend over time, especially as you test different VTA windows. You need to be able to quickly see performance broken down by creative, audience segment, and CTV platform.
Common Mistake: Never, ever trust the conversion numbers inside the CTV platform’s dashboard as your source of truth. They use their own attribution logic and will often over-report. Your MMP is your single source of truth for app installs, period. Your dashboard should reflect this hierarchy.
4.2. Analyzing User Journey Paths
Now that your data is all in one place, you can finally start answering the interesting questions about how CTV fits into the bigger picture. This is where you move beyond simple attribution and start analyzing the entire user journey.
- Path-to-Conversion Reports: Dig into your MMP’s Path-to-Conversion or Multi-Touch Attribution reports. You’re looking for common patterns. How often does a CTV impression show up as the first touchpoint, followed by a branded search click and then an install? This tells you the role CTV is playing.
- Cohort Analysis: Use your MMP’s cohort tools (like in AppsFlyer > Cohorts) to compare users who were acquired through CTV campaigns against users from other channels. Do the CTV users stick around longer? Do they have a higher Lifetime Value (LTV)? This is how you prove the quality of the users you’re acquiring.
By actually analyzing the full user journey, you can finally understand CTV’s true role in driving your app growth. Integrating data, using smarter attribution models, activating your first-party data, and building unified reports are the steps that let you confidently scale your CTV campaigns instead of just guessing. It’s how you stop wasting ad spend and start maximizing your actual return.
What is the optimal view-through attribution window for CTV campaigns?
For CTV, you should start with a 7 to 14-day view-through attribution window. It’s totally different from mobile, where conversions happen fast. A user might see your ad on a Roku during a show, get distracted, and not actually grab their phone to search for and install your app until the next day. A short, 24-hour window misses that conversion entirely.
Why is server-to-server (S2S) postback important for CTV attribution?
S2S postbacks are all about data accuracy and privacy. Instead of relying on something happening in a user’s browser, the conversion data is sent directly from your MMP’s server to the CTV ad platform’s server. This direct line reduces data loss from ad blockers or network errors, cuts down on fraud, and keeps you compliant with privacy rules.
How can I measure the incremental impact of my CTV app install campaigns?
You measure incrementality by running a classic holdout test. You create two identical audience groups in your CTV platform. You show your ads to the “test” group and explicitly exclude the “control” group. After running the campaign for 4-6 weeks, the difference in the install rate between the two groups is your true incremental lift, the value your ads actually created.
Can I use my existing CRM data to improve CTV targeting?
Yes, and you absolutely should. You can upload hashed data from your CRM (like email lists) to your CTV platform. This lets you create custom audiences to re-engage past users, build powerful lookalike audiences to find new users who are just like your best ones, and suppress your current users from install campaigns so you stop wasting money on them.
What are the primary challenges in CTV app attribution compared to mobile-only attribution?
The biggest headache with CTV attribution is that you’re always working across devices without a shared identifier to connect them. A TV and a phone don’t have a common ID. This means you have to deal with a messy, cross-device user journey, much longer conversion delays, and a heavy reliance on probabilistic matching just to connect the dots and figure out what’s working.