Retention messages save 17% · BoldVoice paywall test failed

Apple's native retention messages are live with a 17% median save rate, though in one test a discount earned less per customer than no offer.

BoldVoice found that a hard paywall lifted trial starts 15–20%, but refunds wiped out the gain.

Platform

Monetization

▶ YouTube · 50 min RevenueCat (YouTube) Apple's retention messages: 17% median save, but discounts can backfire

Key takeaway: Apple's native retention messages have a median save rate of 17%, but in Dipsea's test a discount earned less per customer than no offer at all.

• 95% of annual subscribers who cancel never return within a year, making the system-level cancellation screen the last cheap chance to keep them.

• Analysis of 22 million+ messages shows a median save rate of 17%, though the top 10% of apps reach 42% or more.

• Only 1.1% of users actually accept a promotional discount; most saved users simply choose to keep their current subscription after being reminded of the value.

• Audio app Dipsea found that while a 50% discount increased saves, the group shown no offer generated significantly more revenue per customer ($3.16 vs $1.24).

• The standard App Store Connect setup excludes free trial users; you must use the API to target users before their first payment.

• Apple counts a save the moment a user taps a button; RevenueCat only counts users still subscribed 24 hours later.

• Technical constraints are strict: titles are capped at 66 characters, body text at 144 characters, and images must be 3840 x 2160 PNGs.

Why it matters: This is a native, system-level way to fight churn at the exact moment of cancellation without needing to build custom in-app UI.

Worth it: Yes, because it provides the first concrete benchmarks for a new platform feature that directly impacts LTV and churn.

Watch on YouTube →

themobiletakeaway.com

▶ YouTube · 26 min SubHub by Adapty How Flowa lifted a journaling app's ARPPU 50%+ with its web funnel

Key takeaway: Optimizing a web-to-app funnel for an iOS journaling app through creative testing, geographic expansion, and personalized plan defaults increased ARPPU by over 50% while significantly reducing acquisition costs.

• Static images beat video and UGC; creatives alone cut CAC about 50%, and top ads hit $17–$25 per purchase vs $55.53 initially.

• Expanding beyond the US to Tier-1 markets like Canada, the UK, and Singapore yielded a CAC 60% lower than US benchmarks.

• Changing the first screen of the web quiz from a simple gender question to a "Habit Promise" caption ("Build a habit that changes your life") lifted screen-one conversion from 35% to 43.7%.

• Defaulting users who previously used note-taking apps to an annual plan resulted in 70% of those users choosing the longer subscription, boosting ARPPU by 30% to 40%.

• An upsell pop-up triggered when users selected a 1-month or 3-month plan offering 35% off the annual plan increased ARPPU by an additional 20%.

• The team found that "productivity seekers" converted at only 3.2%, while "emotional" users looking for consistency converted at 11.4%.

• Funnel copy targeting users aged 45+ resulted in an ARPPU of $19.01, whereas copy attracting 25-44 year olds dropped ARPPU to $14.56.

Why it matters: For founders using web-to-app funnels, small changes in copy and plan defaults can drastically shift unit economics and define which age demographics actually complete a purchase.

Worth it: Yes, because it provides specific, reproducible tactics for increasing subscription revenue and lowering CAC outside of the competitive US market.

Watch on YouTube →

themobiletakeaway.com

▶ YouTube · 18 min App Masters Steve P. Young's voice-coding workflow with Typeless

Key takeaway: Using AI-driven dictation tools like Typeless can increase input speed from 45 to 220 words per minute, allowing founders to dictate complex code prompts and marketing copy while the AI automatically strips filler words and stutters.

• Typeless integrates across macOS, iOS, and Android to convert messy speech into structured AI prompts, Slack messages, and emails in over 100 languages.

• Steve P. Young demonstrates "voice coding" by dictating a complex request to Claude, which built a functional Meta Ads Library video downloader despite verbal errors and restarts in the original prompt.

• The tool filters out "ums," "uhs," and verbal stumbles, formatting the output into professional bullet points or technical instructions ready for LLMs.

• For ASO localization, dictating a request for the Dutch market generated translated metadata, specific keyword candidates like "senioren," and localized screenshot concepts.

• Marketing tactics include using AI to generate "apology hooks" for ad scripts and creating curiosity on LinkedIn by pixelating app names in Sensor Tower screenshots to drive comments.

• The efficiency gain is estimated at one full day per week by replacing traditional typing with high-speed voice input that works everywhere you write.

Why it matters: For small teams, the bottleneck is often the time spent writing documentation, prompts, and marketing copy; voice-to-AI workflows bypass the friction of typing and editing.

Worth it: Yes, because it provides a concrete tool and specific prompting tactics to speed up both development and distribution tasks.

Watch on YouTube →

themobiletakeaway.com