Editorial Team If you’re not tracking how users jump between their phones, tablets, and desktops, you’re basically flying blind. It’s the only way to get a true picture of your marketing’s impact. For app marketers, it’s essential. We just wrapped a campaign for “FocusFlow,” a productivity app, where our goal was to boost premium sign-ups, and the whole thing hinged on stitching together those fragmented user journeys. This breakdown shows exactly how connecting that data produced real results.
Key Takeaways
- Using a device graph for better attribution and targeting can lift your return on ad spend (ROAS) by over 15%.
- A/B testing creative for different devices works. We saw click-through rates (CTR) on mobile and desktop improve by an average of 10%.
- When you build unified customer profiles from cross-device data, you can spot high-value segments and lower your cost per conversion.
- Retargeting people who start something on one device but leave can bring back up to 20% of conversions you would have otherwise lost.
Our goal was pretty simple: get 20% more premium subscriptions for the “FocusFlow” productivity app within three months. The real challenge, as always, was figuring out which ads actually worked when users see them on a Mac, download on an iPad, and finally subscribe on an iPhone. Without a way to connect those dots, our marketing spend was a mess of inefficient, siloed channels.
We had a $150,000 budget for the 90-day campaign, which ran from January 1 to March 31, 2026. We were watching our Cost Per Lead (CPL), Return On Ad Spend (ROAS), Click-Through Rate (CTR), impressions, conversions, and Cost Per Conversion (CPC). Going in, we were expecting a CPL around $10 and hoped to hit a 150% ROAS by the end of the 90 days.
Strategy: Unifying the User Journey with Device Graphs
The entire strategy was built around a solid device graph solution. This tech let us connect all the random user interactions on different devices into a single, coherent user profile. We piped this data into our existing stack, which included our mobile measurement partner AppsFlyer and our customer data platform Segment. The point was to get away from last-click attribution and see how every single touchpoint contributed to a sale.
We used a mix of two device graph types. Deterministic matching is the high-confidence method. It uses logged-in data like an email address to definitively link a user’s laptop and their phone when they log into FocusFlow on both. Then there’s probabilistic matching, which uses algorithms to make educated guesses, connecting devices based on things like shared IP addresses, Wi-Fi networks, and browser settings. While it’s not as precise, probabilistic matching gave us visibility into the much larger pool of anonymous users. Using both gave us accuracy where we could get it and scale where we couldn’t.
We weren’t just guessing this would work. A 2023 eMarketer report found that companies using this stuff saw an 18% average jump in campaign effectiveness, which made us feel pretty good about the investment.
Creative Approach: Contextual Messaging Across Devices
Our creative had to be smart about what device it was on. We didn’t just run the same ad everywhere. For desktop ads, where people are usually in a research mindset, we ran longer-form copy and detailed screenshots that showed off FocusFlow’s deep-focus features and integrations with other software. It’s a different user context.
Tablet ads were all about portability and the visual experience, so we focused on the app’s drag-and-drop UI and collaboration tools. The copy was shorter with a really clear “download now” CTA. For mobile ads, it was all about speed and convenience. We used short video clips showing how fast you could manage tasks or check notifications on the go, demonstrating a clear use case for that specific device.
We also layered on a dynamic creative optimization (DCO) strategy through AdRoll’s platform. This system automatically mixed and matched our headlines, images, and CTAs based on user behavior and device, so we were always personalizing the ads in real-time. This meant a user would see the right message at the right time, wherever they were.
Targeting: Segmenting by Cross-Device Behavior
The device graph data fed directly into our targeting. Instead of just targeting “mobile users,” we could build audiences based on what they did across all their devices. For instance:
- Desktop Initiators: People who checked out the FocusFlow site on a desktop but didn’t download the app. We hit them with app download ads on their tablets and phones.
- Tablet Explorers: Users who downloaded the app on a tablet but didn’t buy the premium subscription. We’d retarget them on their smartphone with an ad about premium features, maybe with a small discount to nudge them over the line.
- Cross-Device Engagers: The high-intent group. They’d used the app on at least two devices but still hadn’t paid. They got hyper-personalized ads with testimonials and a very simple path to subscribe.
We built these segments as custom audiences in Google Ads and Meta Ads Manager by uploading anonymized user IDs from our device graph. That kind of segmentation is so much more effective than old-school, single-device targeting.
Our biggest win was finally getting attribution accuracy. The device graph showed that about 35% of our premium subscriptions came from a multi-device journey, a huge blind spot in our old last-click models. It turned out our desktop ads, which looked like poor performers because they had few direct conversions, were actually critical for building initial awareness that led to mobile conversions later.
The retargeting campaigns informed by this data were incredibly successful. When we retargeted users on their smartphones after they’d started a trial on a tablet, their conversion rate was 22% higher than users in our generic retargeting pool. The Cost Per Conversion for these specific campaigns dropped by 18%, from an average of $25 to $20.50. That’s efficiency.
Overall campaign metrics:
- Total Impressions: 15,000,000
- Average CTR: 1.8% (up from a baseline of 1.2%)
- Total Conversions (Premium Subscriptions): 4,800
- Average CPL: $8.50 (for trial sign-ups)
- Average Cost Per Conversion: $31.25
- ROAS: 185%
Hitting a ROAS of 185% blew past our 150% goal, bringing in a net revenue gain of $127,500 from the ad spend. This was about spending smarter, not just spending less.
What Didn’t Work: Over-reliance on Probabilistic Matching
Our system wasn’t perfect. We found that probabilistic matching really struggled with quick, single-session engagements. If someone saw an ad on their laptop for 20 seconds and then popped up on their phone a minute later, the model often couldn’t confidently connect the two sessions. This meant some true cross-device journeys, especially for brand new users with no history, fell through the cracks and weren’t attributed correctly.
Also, our first batch of tablet ads completely bombed. They were just lazy, scaled-up versions of our mobile ads and they looked awful. They didn’t use the larger screen space properly and felt totally out of place. It was a painful reminder that you have to create bespoke creative for each device, not just responsive layouts.
Optimization Steps Taken: Refining Algorithms and Creative Iteration
Mid-campaign, we had to tune the probabilistic matching algorithms to get better at those shorter sessions. We started feeding them more data points, like device models and OS versions, and changed the weighting to prioritize behavioral patterns over some of the less reliable static signals. That work resulted in a 5% lift in successfully linked sessions for users with very little interaction history.
In response to the terrible tablet ad performance, we launched a series of rapid A/B tests. We tried more interactive formats and longer videos, and one variant that included a mini-tutorial embedded in the ad was a huge winner. This new approach boosted CTR on tablets by 15% and dropped the Cost Per Install (CPI) by 10%. A one-size-fits-all creative approach is a huge drag on performance in multi-device marketing.
We also adjusted our frequency capping. Instead of capping impressions per device, we started capping them per user across all their devices, using the device graph. This stopped us from annoying people with the same ad everywhere and cut our wasted impressions by an estimated 7%.
Conclusion: The Imperative of Cross-Device Insight
The FocusFlow campaign proved that investing in cross-device tracking isn’t about collecting data for data’s sake. It’s about turning that data into actions that make your marketing better. When you can see the complete user journey, you can stop guessing, optimize every ad and landing page, and get a much higher return on your ad spend. This kind of thinking is going to be standard for any serious app growth hacking in 2026, where you need every advantage. Getting attribution right also strengthens the case for your AI marketing budget by proving you can deliver conversion lift. For any app trying to break through the noise during big retail moments, mapping these journeys is key to building a winning strategy for Prime Day App Store Optimization or other big campaigns.
What is a device graph in the context of cross-device tracking?
Think of a device graph as a map that links a single person’s various devices (phone, laptop, tablet) together. It creates this map by using a mix of two techniques: deterministic matching (which uses hard evidence like a user login) and probabilistic matching (which makes educated guesses based on IP addresses, browser types, and behavior).
How does cross-device tracking improve ad attribution?
It lets you see the whole story instead of just the last chapter. Standard attribution often only credits the very last ad a user clicked, but cross-device tracking shows you every single touchpoint that led to a conversion, like seeing an ad on a desktop and later converting on mobile. This lets you properly value your awareness campaigns and allocate budget more effectively.
What are the primary benefits of using cross-device tracking for app marketing?
The main benefits are getting your attribution right, running much smarter retargeting campaigns across devices, and creating more personal user experiences. You also cut down on wasted ad spend with better frequency capping and get a full picture of the customer journey, which all leads to a better ROAS.
What are the challenges associated with implementing cross-device tracking?
The big ones are privacy (you need solid anonymization and clear user consent), the technical headache of integrating all your data sources, and keeping the device graph’s algorithms accurate. Another tricky part is figuring out who is who in households where multiple people share a tablet or computer.
Can cross-device tracking be effective without logged-in user data?
Yes, it can still be very effective. While using logged-in data (deterministic matching) is the most accurate way, you can still get great insights from probabilistic matching. This method makes intelligent inferences about who owns which device based on behavior and other signals, giving you a good view even of your anonymous users, just with a slightly lower confidence level.
