Activating Users with Right Message and Right Experience via Deeplink
+10%
Install-to-Reg Lift, 3 Months
11
In-App Deeplinks Shipped
3 mo
Time to Sustained Lift
0.08%
IVR First Attempt, Killed
Install-to-registration rate was flat. We tried recovery first (IVR calls, 0.08% conversion, killed). The real bottleneck was that ads promised a feature and users landed on a generic home screen. Shipped 11 in-app deeplinks, tied paid creative to specific features, and used deferred deeplinks so post-install users land on the promised page. Install-to-reg lifted 10+% over 3 months.
Context
Shikho was acquiring installs at $35K monthly paid spend, but the install-to-registration rate had stayed flat for two quarters. The funnel from registration-initiation to registration-completion was also stuck. Acquisition was working; activation was not. The gap was costing the equivalent of a full month's free user growth, every month.
Challenge
The first instinct was wrong. We assumed the bottleneck was post-install drop-off and tried to recover users who didn't register. That diagnosis sent us toward the wrong fix.
Approach
Phase 1: failed first attempt
IVR call recovery
Called recently uninstalled users and offered to walk them through registration. The funnel collapsed at every step. Documented kill at the next monthly review. The number told us the diagnosis was wrong, and that was the unlock.
IVR funnel collapse
11,934 sampled to 10 registered, 4 steps
0.08% conversion was unambiguous enough that no follow-up cycle could rescue the experiment. The whole story is the magnitude collapse from top to bottom.
View data as table (4 rows)
| Step | Count | Step conversion |
|---|---|---|
| 1: Sampled (outreach) | 11,934 | baseline |
| 2: Answered IVR | 5,563 | 46.6% |
| 3: Pressed 1 (opted in) | 410 | 3.4% of answered |
| 4: Registered (outcome) | 10 | 0.08% of sampled |
Phase 2: diagnosis pivot
The bottleneck was experience, not recovery
Users were not leaving because they had decided no. They were leaving because the ad promised a specific feature (a mentor list, a free animated lesson, a routine page) and the post-install app dropped them on a generic home screen with no path back to what motivated the install. The promise broke at the threshold of the app. Recovery cannot fix a broken promise.
Phase 3: solution, three parts
Right message, right experience, deferred attribution
- 1. Worked with Product and Tech to ship in-app deeplinks for 11 destination pages: Routine, Homework, Course Details, Free Animated Lesson, My Course, Exclusive Guideline, Most Important Class, Mentor List, Mentor Details, Shorts, App Launch / Home.
- 2. Rebuilt paid creative around specific features. Each ad set carried a feature-specific message ("Find your HSC mentor", "Watch the most important class for HSC27"). The promise on the ad and the page in the app now matched.
- 3. Used deferred deeplinks in app campaigns. User taps ad → app store → installs → opens app for the first time → deferred deeplink fires → lands directly on the promised page. The contextual promise survives the install boundary.
The Flow, Before and After
Results
Install-to-registration rate lifted 10+% sustained over the first 3 months. The lift compounded: every paid install now had a higher probability of registering, which fed every downstream metric (paid users, retention, revenue) without additional spend. On the $35K monthly paid-acquisition baseline, the lift translates to several thousand additional paid registrations per month at no extra spend.
Install-to-registration rate, monthly
Pre-deeplink oscillation followed by sustained shift after Feb 2026 rollout
The inflection at February 2026 marks the deeplink rollout. The post-rollout band sits ~5pp higher and holds for 3 sustained months. The lift is structural, not a single-month spike.
View data as table (10 rows)
| Month | Install-to-reg % |
|---|---|
| Jul 25 | 47.9% |
| Aug 25 | 54.1% |
| Sep 25 | 52.2% |
| Oct 25 | 56.7% |
| Nov 25 | 52.4% |
| Dec 25 | 50.0% |
| Jan 26 | 47.6% |
| Feb 26 | 55.5% |
| Mar 26 | 58.2% |
| Apr 26 | 57.8% |
The campaigns that drove the largest individual lifts were the ones with the strongest feature-to-creative match. Generic "download the app" creative did not benefit. Feature-specific creative did, and the deferred deeplink was the link that made the lift possible.
Learnings
Right message and right experience are one unit
The promise survives only if the landing matches it
Feature-specific creative without a feature-specific landing is just better-written acquisition copy. The lift came from pairing the two.
Deeplinks are a Product investment, not a marketing one
Cross-team execution mattered more than the channel
The 11 deeplinks needed Product and Engineering work. Marketing owned the diagnosis and the creative; Product and Tech owned the destinations. Neither side could have shipped this alone.
Tools
Meta and Google paid, in-app deeplinks, deferred-deeplink attribution
Team
Lead, cross-functional with Product and Engineering
Timeframe
3-month rollout, sustained lift
Role
Lead, Performance and Growth Marketing