Editorial Team Marketers are struggling to prove their campaigns work because customer journeys are shattered across multiple devices. You can’t accurately attribute conversions or understand your true return on investment when you have this fragmented view. It’s a simple, frustrating problem: a customer sees your ad on a smart TV but buys the product later on their phone, and you have no way to connect the two and prove your TV budget is actually working.
Key Takeaways
- You need a dedicated cross-platform measurement solution to stitch together user data from CTV, mobile, and web.
- Prioritize deterministic matching, using authenticated user IDs is always the most accurate way to map a user’s journey.
- Roll out your measurement in phases. Start by tracking a few core KPIs before you try to build out more complex attribution models.
- Constantly audit your data collection points to keep them consistent across platforms. Otherwise, you’ll end up chasing discrepancies and bad data.
Your standard mobile attribution tools are great for app-to-app journeys, but they break down the second the user path crosses multiple screens. Think about this really common scenario: someone sees an ad for a new streaming service on their Connected TV (CTV) at night. A few minutes later, during a commercial break, they grab their phone, search for the service, and download the app. How do you give credit to that CTV ad for the mobile install? For years, that connection was a total mystery for most marketing teams, which led directly to misallocated budgets and a poor understanding of what was actually driving results. I’ve personally watched teams pour money into CTV campaigns with nothing more than a gut feeling that it was working, because they had zero concrete data to prove it.
One of the first stabs at solving this was probabilistic matching, which was basically a series of educated guesses. It relied on signals like IP addresses, device types, operating systems, and timestamps to infer that it was the same user on different devices. This was better than nothing, but it was deeply flawed. With the rise of shared household IP addresses and stronger privacy controls, probabilistic matching became less and less reliable. We saw cases where a single family’s CTV ad view was incorrectly attributed to three different mobile devices, which massively inflated the campaign’s reach and completely skewed the conversion rates. It felt like trying to solve a jigsaw puzzle where half the pieces were missing and the other half came from a different box.
Another huge mistake was relying too heavily on last-touch attribution. If a user saw your CTV ad, then a social media ad, and finally clicked a search ad before installing your app, that search ad got 100% of the credit. This model completely ignores the CTV ad that kicked off the entire journey. The consequence is pretty clear: if you only credit the final touchpoint, you end up cutting budgets for top-of-funnel channels that are actually creating all the initial demand. A lot of companies, including some I’ve worked with, found themselves gutting budgets for channels that looked like they weren’t performing, when in reality they were essential parts of the conversion path, just not the final one.
You solve this attribution headache by putting a solid cross-platform user journey tracking system in place. This requires an advanced measurement platform that can unify data from all your touchpoints, from CTV and mobile right through to web. The key is to move away from the guesswork of probabilistic matching and get as deterministic as possible. That usually means relying on authenticated user IDs, where people log in to your service across different devices. When a user logs into a streaming app on their smart TV and then uses that same account on their phone, the system can definitively tie those two actions to a single person. That kind of certainty is the only way to get accurate attribution you can actually build a strategy on.
Look at the example of MTN, a huge telecommunications company in Africa and the Middle East. They had the classic problem: they were spending a ton of money on CTV advertising but had no idea if it was driving mobile app installs or engagement. Their marketing was spread across tons of channels, but the link between someone seeing an ad on their TV and then downloading the MTN app was a mystery. This made it really hard to justify their CTV budget or optimize their media mix. And with CTV ad spend continuing to grow so fast, as a 2024 eMarketer report confirms, flying blind in that channel is just not an option anymore.
MTN decided to partner with AppsFlyer, a mobile attribution and analytics platform, to get this sorted out. Their strategy was to use AppsFlyer’s CTV measurement solution, which is built specifically to track impressions on CTVs and connect them to later actions on mobile, like app installs and in-app events. The whole process broke down into a few key steps.
First, MTN’s developers integrated the AppsFlyer SDK into their mobile apps. The SDK is a bit of code that lets the app report installs, purchases, and other user actions back to AppsFlyer’s servers. At the same time, their marketing team worked with their CTV ad partners to make sure ad impression and click data from TVs was also being sent to AppsFlyer. This usually requires setting up server-to-server integrations or passing universal IDs between the platforms.
The real trick here is making the connection between an anonymous CTV impression and a known mobile user. AppsFlyer does this by prioritizing deterministic matching whenever possible. For example, if an MTN customer logs into an MTN service on their TV and then uses that same login on their mobile app, AppsFlyer has a confirmed match. For cases where there’s no login, the platform uses a much more sophisticated form of probabilistic modeling that goes way beyond basic IP matching, analyzing device graphs and behavioral patterns to make a highly-educated guess about who is who.
One of the smartest things they did was use AppsFlyer’s Incrementality Testing framework. Instead of just attributing conversions after the fact, MTN could run controlled experiments to see the true lift their CTV campaigns were generating. This means creating a control group of users who aren’t shown the CTV ads and comparing their behavior to the group that is. This gives you a much clearer answer about causation, not just correlation. I always tell my clients that incrementality testing is more work to set up, but it’s the only way to get an honest answer to the question, “Is this channel actually driving new business that wouldn’t have happened otherwise?”
The results for MTN were huge. By using AppsFlyer’s solution, they finally got a clear line of sight into their CTV to mobile user journeys. They could see exactly which CTV campaigns were driving the most app installs and what users did inside the app afterward. This specific data let them start optimizing their CTV ad creative, targeting, and bidding with confidence. For instance, they discovered that one ad creative shown on a specific CTV platform led to a much higher rate of mobile app downloads. That’s a level of insight they simply didn’t have before, and it led to much smarter budget decisions.
Best of all, MTN could finally calculate a real return on ad spend (ROAS) for their CTV campaigns. The guesswork was gone. This allowed them to pull budget from underperforming CTV placements and put it into the ones that were proving to be more efficient and have a bigger impact on their mobile business. Being able to connect a TV ad view directly to a user subscribing on their phone created a complete picture of the customer journey, which enabled a more effective, cohesive marketing strategy. They could now answer questions that justify an entire budget, like, “How many users who saw our ad on a Samsung Smart TV installed our app within 24 hours and then subscribed to a premium service?” This is the kind of detailed user journey analytics that turns a bunch of separate campaigns into a single, data-driven story.
Switching from fragmented, channel-by-channel reports to a unified cross-platform view isn’t just a convenience for your analytics team. It’s a survival tactic in today’s multi-device world. Brands that don’t connect these dots are not only wasting money but also fundamentally misunderstanding how their customers behave. This integrated approach allows for better personalization, smarter campaign optimization, and in the end, a much stronger position against competitors.
Having a strong cross-platform measurement solution is no longer optional. It’s fundamental for any brand that wants to understand its real customer acquisition costs and optimize marketing spend in a complex digital world. The ability to follow a user from their first ad exposure on a Connected TV to their final conversion on a mobile device provides the hard data needed to make decisions that actually improve the bottom line.
What is cross-platform user journey tracking?
It’s about following and analyzing a customer’s interactions across all their devices and channels, like their CTV, mobile app, and web browser, to see the complete path they take on their way to becoming a customer.
Why is it important to track CTV to mobile conversions?
Because consumers often discover products on their smart TVs and then use their phones to complete the action, like installing an app or making a purchase. If you don’t track that link, you can’t accurately measure how valuable your CTV campaigns really are.
What are the main challenges in cross-device attribution?
The biggest headaches are trying to link anonymous device IDs back to a single person, working through privacy restrictions that limit data sharing, and the pure technical complexity of stitching together data from dozens of different platforms and ad networks.
What is the difference between deterministic and probabilistic matching?
Deterministic matching is a sure thing. It uses a known identifier like a logged-in user ID to confidently link devices to one person. Probabilistic matching makes an educated guess by using shared signals like IP addresses, device types, and behavioral patterns to infer a connection, which is less certain but still useful.
How does incrementality testing improve attribution accuracy?
Incrementality testing proves cause-and-effect by using control groups to measure the true impact of a marketing campaign. It helps you figure out if a channel is actually driving new conversions that wouldn’t have happened on their own, giving you a much more accurate understanding of your ROAS than simple attribution models.
