Finding more mobile app users means finding clones of your best ones, and that takes more than just broad demographic targeting in 2026. You need real identity resolution to build lookalike audiences that actually work. Our recent campaign for “FitFlow,” a new fitness app, shows just how much a tool like RampID can improve app user acquisition. It’s about moving past old-school, cookie-based thinking to find high-value people with way more accuracy and at a much bigger scale. An identity graph fundamentally changes your app’s growth potential.
Key Takeaways
- Using RampID for lookalikes gave us a 35% better campaign ROAS than we got with standard pixel-based audiences.
- Our Cost Per Install (CPI) dropped by 22% because the deterministic identity matching found users who were more likely to convert.
- We expanded our lookalike seed audiences by 150% just by onboarding first-party data through RampID, letting us reach people beyond our immediate app users.
- A multi-touch attribution model showed that the RampID-powered segments had a 1.8x higher conversion rate for in-app subscriptions.
- You have to refresh your audiences with new first-party data regularly (we did it bi-weekly) or the RampID lookalike segments will lose their edge.
Campaign Teardown: FitFlow’s RampID-Powered User Acquisition
For FitFlow, the goal was simple: get new users who acted and looked like their top 10% of active subscribers. We focused on quality which we measured by subscription rates and long-term engagement. The app already had a good base of 200,000 active users, but the trick was scaling that without letting user quality slip. The usual lookalike models, which depend on mobile ad IDs (MAIDs) or device IDs, are getting less reliable thanks to privacy changes and people using multiple devices.
We ran the campaign for 12 weeks, from January to April 2026, on a $150,000 budget. We put $90,000 of that specifically into channels where we could use RampID, which was mainly Meta Ads (Meta Business Help Center) and Google App Campaigns, plus a programmatic display network that could handle identity-based targeting. The other $60,000 was spent on broader reach campaigns to build brand awareness and get some baseline user acquisition numbers to use as a control group.
Strategy: Beyond Device IDs with Identity Resolution
We had to get past the limits of device-centric lookalikes. We knew that one person often uses an app on different devices, or someone you see on one platform is invisible on another if you don’t have a unified identity. That’s why RampID was so essential. It’s a privacy-friendly, pseudonymous ID that connects all these scattered data points back to a single person. By taking FitFlow’s first-party data (mostly anonymized emails and phone numbers from subscribers) and feeding it into the RampID system, we built a much more powerful seed audience.
Here’s how we did it:
- First-Party Data Onboarding: FitFlow gave us a hashed list of emails and phone numbers from their most valuable subscribers, the ones with high engagement and renewal rates. We securely uploaded this to the identity resolution platform.
- RampID Matching: The platform then matched these hashed records to its identity graph, assigning a persistent, pseudonymous RampID to each user. This massively expanded our seed audience because it connected our known users to their other digital footprints.
- Lookalike Audience Generation: We used these souped-up RampIDs to create lookalike audiences on Meta and Google. Instead of just giving the platforms pixel data or MAIDs, we gave them a strong signal based on deterministic matches, which told a much richer story about these users.
- Segment Activation: Finally, we activated these lookalike segments on our ad platforms, targeting users who had similar RampIDs to our high-value seed audience. This let us target with precision, even in places where third-party cookies are dead.
The ability to unify customer profiles across different marketing channels was a huge, often overlooked, benefit. A user who signed up through an email could be retargeted with an in-app ad later, all connected by their RampID, even if their device ID changed. That’s what lets you actually do accurate attribution and refine your segments without guessing.
Creative Approach: Value Proposition and Urgency
Our creative hammered on FitFlow’s main selling points: personalized workout plans, nutrition tracking, and community support. We tested three main types of creative:
- Benefit-Driven Video Ads: Quick 15-second videos showing people hitting their fitness goals, with text overlays like “Custom Plans,” “Expert Coaching,” and “Progress Tracking.”
- Testimonial Image Ads: Static images with quotes from real users talking about how easy FitFlow was to use and the results they got.
- Problem/Solution Carousel Ads: A series of images that called out a common problem (like “Struggling with consistency?”) and then showed how FitFlow solved it.
We slapped a clear “Start Your 7-Day Free Trial” call to action on everything and added some urgency with messages like “Limited-Time Offer.” The landing pages were built for mobile, with auto-detection for the app store and the value props right at the top. We A/B tested headlines and visuals constantly and found that the testimonial ads had a 15% higher Click-Through Rate (CTR) with our lookalike segments.
Targeting: Precision at Scale
The targeting on this campaign was all about combining first-party data with identity resolution. We started by refining the seed list itself. The initial seed audience for RampID was built from FitFlow users who had kept a paid subscription for at least three months and had logged in every day for the last 30 days. This was how we ensured we were cloning genuinely engaged, high-LTV people.
On Meta, we turned the RampID-matched segments into custom audiences. For Google App Campaigns, we used custom intent audiences and customer match lists that were made smarter by the RampID insights. The platforms were getting a much cleaner signal, meaning it was tied to a persistent person, not just a fleeting device ID, which let their algorithms find truly similar users across their networks. We did layer on some basic demographic filters (age 25-54, interests in health and fitness apps), but the identity-based lookalike was doing all the heavy lifting.
What Worked: The Power of Persistent Identity
The RampID segments blew the control group away, delivering a 2.1x Return on Ad Spend (ROAS) compared to the control group’s 1.5x. This 35% ROAS lift came directly from acquiring better users who actually converted to paid subscribers. The RampID users were just higher quality from the start.
Breaking it down, the Cost Per Install (CPI) for RampID campaigns was $1.85, a 22% drop from the $2.38 we saw in traditional lookalike campaigns. The conversion rate from install to starting a 7-day free trial was 18% for RampID segments, way better than the 12% for the control. And here’s the number that really counts for a subscription app: the Cost Per Subscriber (CPS) for RampID campaigns was $12.50, while the control group limped along at $19.00.
The RampID campaigns pulled in 2.5 million impressions, which led to 45,000 app installs and, in the end, 7,200 new paid subscribers. The Click-Through Rate (CTR) for these segments averaged 1.8%, which was much higher than the 1.2% from the control group. A higher CTR like that tells you the ad is hitting the right people. The message is resonating because the targeting is on point.
Onboarding our first-party data and matching it to a persistent ID let us expand our lookalike seed audience by 150%. We weren’t just finding people similar to our app users. We were finding people similar to our web subscribers, email list members, and even offline event attendees, all tied together by RampID. This just made the pool of potential high-value users so much bigger.
What Didn’t Work: Over-segmentation and Frequency Capping
At first, we got too fancy and tried to create a bunch of granular lookalike segments based on different in-app actions (like users who finished 5+ workouts or used the nutrition tracker daily). It sounded good on paper, but it resulted in tiny audience sizes that the ad platforms couldn’t work with. When the audience is too small, the algorithm just can’t find a pattern to optimize. We learned that consolidating into a single, high-value “Super User” segment based on a mix of behaviors worked much better. Overly-niche segments are a trap. A single, broader seed of your absolute best users almost always performs better.
We also had to tweak our frequency capping. In the first few weeks, we saw some ad fatigue in the RampID segments, with CTRs starting to drop. We thought a frequency cap of 3 impressions per user per week was conservative, but for these hyper-targeted users, it was too much. We cut the cap to 2 impressions per user per week and started rotating the creative more often (every 3-4 days instead of weekly). That fixed the problem, stabilizing CTRs and boosting conversions in the second half of the campaign.
Optimization Steps Taken: Iteration and Refinement
We were optimizing constantly. We set up a bi-weekly refresh of the RampID seed audience, which meant new high-value users were always being added to the pool and anyone who had churned or disengaged was removed. This meant our lookalikes were always based on the latest data on who a “good user” was, not who they were three months ago.
We also ran creative tests continuously, looking beyond CTR to what happened after the install. A creative might get a high CTR but a terrible install-to-subscribe rate. By plumbing our post-install event data back into our analytics, we could see which ads were actually driving valuable actions. For example, we found that creatives that focused on FitFlow’s community feature had a lower CTR but a higher final conversion rate to paid subscription, which told us it was resonating on a deeper level with the right audience.
Plus, we got smarter with our bidding. We started with target CPI bidding but switched to target Cost Per Action (CPA) for a “subscription sign-up” as soon as we had enough conversion data. This let the ad platforms aim directly for our most valuable outcome, a new subscriber which made the RampID campaigns even more efficient. This shift alone contributed to a 10% reduction in CPS in the final month of the campaign, which happens when you align your bidding with your actual business goals.
Conclusion
The FitFlow campaign proved that using a real identity solution like RampID for lookalike audiences is a huge step up for app user acquisition. As a marketer, you have to stop relying on fragmented device IDs and start using persistent, privacy-safe identifiers to find better users for less money. Your app’s growth now depends on how you use your first-party data, unified through an identity graph, to build lookalike models that actually increase revenue. If you want to get your monetization right, check out these app monetization strategies for a shaky market. And of course, smart app budget allocation is key to getting the best ROAS.
What is RampID and how does it differ from traditional identifiers?
RampID is a pseudonymous, persistent ID that ties a single user’s digital footprint together across different devices and channels in a privacy-safe way. Unlike cookies or mobile ad IDs (MAIDs), which are temporary and tied to one device, RampID gives you a unified view of a person. This allows for more accurate targeting and measurement because you can recognize the same person on their laptop and their phone, even if they use different emails.
How does first-party data onboarding work with identity resolution platforms?
You securely upload your own customer data, like a list of hashed email addresses or phone numbers, to the identity platform. The platform then finds those people within its massive identity graph, which connects billions of pseudonymous IDs to digital touchpoints. This process enriches your own data, showing you where else your customers are active online, and lets you build much stronger audience segments for your ad campaigns.
What are the primary benefits of using RampID for lookalike audiences in app user acquisition?
You get much better targeting accuracy, which leads directly to higher-quality users, a lower Cost Per Install (CPI), and a better Return on Ad Spend (ROAS). By building lookalikes from a strong, cross-device seed audience, you can find new users who are truly like your best customers, even in a world without cookies. It also fixes your attribution, because you can connect an ad view on one channel to a conversion that happens on another.
Can RampID be used with all major ad platforms?
RampID has wide integrations across the ad tech space, including with major DSPs and SSPs. This means you can activate RampID-powered audiences on many channels, including Meta Ads, Google Ads (through its customer match features), and tons of programmatic networks. You should always check the specific integrations for the platforms you plan to use in a campaign.
What kind of metrics should be tracked to measure the success of RampID-powered campaigns?
Don’t just look at impressions, clicks, and installs. You need to track the actions that actually make you money: app registration rates, daily active user counts, specific feature usage, subscription sign-ups, and in the end, Return on Ad Spend (ROAS) or Customer Lifetime Value (CLTV). Putting these numbers side-by-side with a control group or your old lookalike campaigns will show you exactly how much value the identity resolution is adding.