Behind that smile shines hope. Here’s why:
1996: I was born with the most beautiful name a mother can give her child: Magloire, which means “My Glory”. As a child, I dreamed big at 7, I wanted to be an inventor, constantly dissecting electrical devices to understand how they worked. By 11, my ambition shifted to becoming a mechatronics engineer.
2010: I realized freedom required money. I told my mother I wanted to study at the best engineering university in Paris. Her reply: "We can't afford it. We are poor...Maybe with a loan." That moment changed my perspective on life.
2011-2014: I became fascinated by finance after watching Wall Street: Money Never Sleeps. I wanted to manage a family office, to be financially free while holding onto my principles. But when it was time for university, I didn't want a lifetime of debt, so I chose economics in Rennes (FR). A week later, I dropped out. It wasn’t for me – they didn’t teach how to earn money, just theory.
2014-2019: I decided to find my own path. Without a degree, I climbed the corporate ladder, taking jobs from housekeeping to banking. I had two principles: always keep learning and change jobs often to pick up new skills. I worked 10 different jobs, all while searching for the right business idea.
2019: I discovered a book about startups, Then @_TheFamily ,@_Koudetat and realized this was my world. I launched a venture capital fundraising company, believing I just needed to connect the right startup with the right investor. Simple, right? Wrong! I quickly learned that in VC, it’s about who you know, not just what you can do.
2020: My life changed again. A month before COVID hit, my mother passed away. My brother and I moved back to Rennes to care for our younger sister till she graduated high school. On my mother’s deathbed, I made a promise: "I would proudly carry the name she gave me and make her proud."
2021: I shifted from VC to data science – the best decision of my life.
Thanks to @lewagonlyon , I earned a Master’s in Data Science, and by 2024, I had mastered SQL, Python, GCP, ETL, LLMs, RAGs and more. AI has no secrets for me.
2024: 10 years after high school, I’m still smiling, still hopeful, and still chasing freedom.
2025- Today, I’m an AI engineer, building projects working toward my goal of 10K MRR.
I’m giving it 120% and know I’ll reach my goal in 2 years. 𝗜𝗳 𝘆𝗼𝘂 𝘄𝗮𝗻𝘁 𝘁𝗼 𝗳𝗼𝗹𝗹𝗼𝘄 𝗺𝘆 𝗮𝗱𝘃𝗲𝗻𝘁𝘂𝗿𝗲 𝗮𝗻𝗱 𝗹𝗲𝗮𝗿𝗻 𝗺𝗼𝗿𝗲 𝗮𝗯𝗼𝘂𝘁 𝗔𝗜 𝗳𝗼𝗹𝗹𝗼𝘄 𝗺𝗲.
Oh, I almost forgot ! never give up, keep smiling, be kind and believe in yourself.
Magloire ☀️🌻
Magloire Ndabagera | AI engineer making Apps retweeted
Day 416: New motion design format. Views yes, conversions no. 🎬
Testing a motion design format today.
What I see so far:
✅ It gathers views
❌ It doesn’t convert (yet)
I’ve been here before. Day 413: 119K views, almost no conversions, because the audience wasn’t the one that spends on apps.
So I’m not judging this format on views alone. I’m giving it a short window and watching two things:
Does it break out and go viral?
Does the audience that watches it look like my ICP?
If it does neither, I move on.
If it does one, I adapt the angle.
Views are the easy part.
A format only earns its place if it also brings the right people.
Magloire Ndabagera | AI engineer making Apps retweeted
Day 412: $1,467 MRR. The curve is moving again. 📈
Where I stand:
💰 MRR: $1,467
📅 Active subscriptions: 70
⏳ Active trials: 43
💵 Revenue (28 days): $2,140
👥 Active customers: 2,768
Day 400: $1,006
Day 410: $1,223
Day 411: $1,269
Day 412: $1,467
+$198 in 2 days.
What caught my eye:
Trials went down (60 → 43).
Subscriptions went up (60 → 70).
That’s trials turning into paying subscribers, which is exactly what I want to see.
I run a weekly subscription app, so every week of renewals counts. The next number I’m watching is week-2 and week-4 retention, because that decides how much budget I can safely add.
$8,533 left to $10K MRR.
Magloire Ndabagera | AI engineer making Apps retweeted
Day 416: Giving Pinterest a second chance. 📌
Back on Day 275 I dropped Pinterest. Under 10 views per pin, so I killed it.
Why try again now:
Pinterest is a search engine more than a social feed. People arrive with intent, which is closer to how ASO keywords work.
A pin can keep getting traffic for months, unlike a reel that dies in 48 hours.
Back then I had no persona and no clear angle. Now I do.
How I’m testing it:
Content built for my persona, not recycled reels
Keyword-driven titles and descriptions
A few weeks of data before any verdict
One failed test on day 275 wasn’t a final answer. It was a test with a weak setup.
Day 416: New motion design format. Views yes, conversions no. 🎬
Testing a motion design format today.
What I see so far:
✅ It gathers views
❌ It doesn’t convert (yet)
I’ve been here before. Day 413: 119K views, almost no conversions, because the audience wasn’t the one that spends on apps.
So I’m not judging this format on views alone. I’m giving it a short window and watching two things:
Does it break out and go viral?
Does the audience that watches it look like my ICP?
If it does neither, I move on.
If it does one, I adapt the angle.
Views are the easy part.
A format only earns its place if it also brings the right people.
14 people worth following if ur problem is keeping subscribers, not getting installs.
each one posts specifics: real mrr, real cancel flows, real tests.
1. @jakecastilloooo. co-founder and coo of cal ai, which was acquired by myfitnesspal. his "operator brain" post maps the magic moment, basic vs power users and retention curves as one system.
2. @consumerxai. recently posted a retention operator role for a nurses' subscription app at $45K mrr with 1,100 paying members and 67% trial to paid, and named the first 30 days as the leak. that's the right diagnosis.
3. @freddyfvng. builds at adapty and hosts the app leaders podcast. his churn checklist starts with setting up retention messaging on the cancel screen.
4. @david_attisaas. devrel at adapty. paywall breakdowns and product rundowns, including apple's retention messaging api.
5. @dnlpearce. wrote one of the clearest summaries of apple's retention messages: live nov 4, up to ten per app, one per subscription.
6. @hermanbjorheim. step-by-step setup for retention messages inside app store connect. short and usable.
7. @ShamanthRao. ceo of rocketship hq and host of the mobile user acquisition show. his point that meta shows cost per trial but not cost per paying user is a retention point dressed as a ua one.
8. @philgcarter. growth advisor for subscription apps. writes about how ai costs change subscription pricing, now that serving a subscriber isn't free.
9. @a_shakh. ceo and co-founder of funnelfox, web onboarding and web payments. runs expert tables on pricing, retention and churn.
10. @Cyrus_Kiani. ceo of touchzen media, 75+ apps shipped. posts trial benchmark breakdowns, like how many day-0 cancels happen on 3-day vs 30-day trials.
11. @AnshulSoni2010. app co-founder who has built apps past $30K mrr. turns long subscription benchmark reports into a few things to actually do.
12. @NdabageraM. posts his numbers almost every day: mrr, active subs, active trials, 28-day revenue. watching trials and subs move side by side teaches more than most courses.
13. @chetanbvaghela. building a couples app in public and asking the real question: paywall before people see value, or after.
14. @regalstreak. co-founder of refix. his list of ai agent uses starts with grouping every cancellation reason and routing each group to a different win-back flow.
the common thread: they show u the cancel screen, not the vibes.
Magloire Ndabagera | AI engineer making Apps retweeted
Day 415: Growing slowly. 🐢
No big spike today.
No new milestone.
Just a curve that keeps going up a little at a time.
300 days at $0.
Now a dashboard that moves every week.
The sprint was exciting.
The slow phase is where the real business gets built:
Fresh creatives for Meta
Retention on the weekly plan
Small improvements that add up
Day 415: Slow growth is still growth.
Day 416: Giving Pinterest a second chance. 📌
Back on Day 275 I dropped Pinterest. Under 10 views per pin, so I killed it.
Why try again now:
Pinterest is a search engine more than a social feed. People arrive with intent, which is closer to how ASO keywords work.
A pin can keep getting traffic for months, unlike a reel that dies in 48 hours.
Back then I had no persona and no clear angle. Now I do.
How I’m testing it:
Content built for my persona, not recycled reels
Keyword-driven titles and descriptions
A few weeks of data before any verdict
One failed test on day 275 wasn’t a final answer. It was a test with a weak setup.
Magloire Ndabagera | AI engineer making Apps retweeted
Day 414: $1,500 MRR crossed. 🎯
Where I stand:
💰 MRR: $1,510
📅 Active subscriptions: 72
⏳ Active trials: 38
💵 Revenue (28 days): $2,235
👥 Active customers: 2,748
The path:
Day 393: $221
Day 400: $1,006
Day 410: $1,223
Day 412: $1,467
Day 414: $1,510
For 300 days this dashboard read $0.
Now it reads $1,510 a month.
The curve is slower than the sprint, and that’s fine. The sprint came from Meta Ads and ASO finally lining up. This phase is about keeping subscribers and feeding the ads fresh creatives.
I run a weekly subscription app, so every renewal week matters. The next lever is retention plus new creatives, not just more budget.
$8,490 left to $10K MRR.
Day 415: Growing slowly. 🐢
No big spike today.
No new milestone.
Just a curve that keeps going up a little at a time.
300 days at $0.
Now a dashboard that moves every week.
The sprint was exciting.
The slow phase is where the real business gets built:
Fresh creatives for Meta
Retention on the weekly plan
Small improvements that add up
Day 415: Slow growth is still growth.
Magloire Ndabagera | AI engineer making Apps retweeted
Day 413: 119K views. Almost no conversions. 📉
The numbers:
👀 119,051 views
👤 84,467 viewers
🎬 100% Reels
🌍 99.5% from non-followers
Sounds like a win. It wasn’t.
The views came from countries where people don’t spend on apps, so they didn’t convert.
How I made them:
Claude + motion design to create short animations.
What it proved:
✅ The format can reach a huge audience
✅ AI + motion design is a fast production pipeline
❌ Reach alone doesn’t pay the bills
Views ≠ conversions.
Where the views come from matters as much as how many there are.
Next move: keep the format, change the angle. Content built for my ICP in high-spending markets, not for whoever the algorithm hands me.
Views are free. The right audience isn’t.
Day 414: $1,500 MRR crossed. 🎯
Where I stand:
💰 MRR: $1,510
📅 Active subscriptions: 72
⏳ Active trials: 38
💵 Revenue (28 days): $2,235
👥 Active customers: 2,748
The path:
Day 393: $221
Day 400: $1,006
Day 410: $1,223
Day 412: $1,467
Day 414: $1,510
For 300 days this dashboard read $0.
Now it reads $1,510 a month.
The curve is slower than the sprint, and that’s fine. The sprint came from Meta Ads and ASO finally lining up. This phase is about keeping subscribers and feeding the ads fresh creatives.
I run a weekly subscription app, so every renewal week matters. The next lever is retention plus new creatives, not just more budget.
$8,490 left to $10K MRR.
Day 413: 119K views. Almost no conversions. 📉
The numbers:
👀 119,051 views
👤 84,467 viewers
🎬 100% Reels
🌍 99.5% from non-followers
Sounds like a win. It wasn’t.
The views came from countries where people don’t spend on apps, so they didn’t convert.
How I made them:
Claude + motion design to create short animations.
What it proved:
✅ The format can reach a huge audience
✅ AI + motion design is a fast production pipeline
❌ Reach alone doesn’t pay the bills
Views ≠ conversions.
Where the views come from matters as much as how many there are.
Next move: keep the format, change the angle. Content built for my ICP in high-spending markets, not for whoever the algorithm hands me.
Views are free. The right audience isn’t.
Day 412: $1,467 MRR. The curve is moving again. 📈
Where I stand:
💰 MRR: $1,467
📅 Active subscriptions: 70
⏳ Active trials: 43
💵 Revenue (28 days): $2,140
👥 Active customers: 2,768
Day 400: $1,006
Day 410: $1,223
Day 411: $1,269
Day 412: $1,467
+$198 in 2 days.
What caught my eye:
Trials went down (60 → 43).
Subscriptions went up (60 → 70).
That’s trials turning into paying subscribers, which is exactly what I want to see.
I run a weekly subscription app, so every week of renewals counts. The next number I’m watching is week-2 and week-4 retention, because that decides how much budget I can safely add.
$8,533 left to $10K MRR.
Day 411: Breaking $1,200 MRR is harder than reaching it. 🧱
Where I stand:
💰 MRR: $1,269
📅 Active subscriptions: 60
⏳ Active trials: 60
💵 Revenue (28 days): $2,117
Day 400: $1,006
Day 410: $1,223
Day 411: $1,269
The sprint is over. Growth now takes a decision.
I run a weekly subscription app, so churn shows up fast. Two levers:
Increase budget
More installs, more trials. But it only works if subscribers keep renewing past the first weeks.
Work on retention
With weekly plans, every renewal week counts. Keeping a subscriber 2 extra weeks is like getting a free install.
My plan: check trial-to-paid and week-2 / week-4 renewal first.
Strong? Scale budget gradually.
Leaky? Fix retention before spending more.
8
5
131
Day 410: $1,223 MRR. The slow part of the curve. 📊
Where I stand today:
💰 MRR: $1,223
📅 Active subscriptions: 58
⏳ Active trials: 57
💵 Revenue (last 28 days): $1,959
👥 Active customers: 2,662
Day 400: $1,006 MRR
Day 410: $1,223 MRR
+$217 in 10 days.
Honest take:
The vertical sprint is over.
The curve is flatter now: steady growth instead of explosive growth.
That’s normal. After a sprint comes the phase where:
Ads fatigue and need fresh creatives
Churn starts to matter as much as new trials
Growth comes from the pipeline, not from one winning ad
$100 MRR took 300 days.
$1,000 took 11 days.
$1,200 took 10 more.
The next stretch isn’t about luck. It’s about creative volume and keeping subscribers.
3
4
78
Day 409: Same AI video, three platforms, three different outcomes. 🤖
What I’ve noticed across Instagram, TikTok and YouTube:
📷 Instagram: AI content performs like any other content. Reach depends on the video, not how it was made.
🎵 TikTok: Doesn’t push AI content, and sometimes flags it. Two of my accounts got shadowbanned over it.
▶️ YouTube: Doesn’t seem to care for reach.
What I’m doing with that:
Human-made videos for TikTok
AI content goes to Instagram and YouTube
Label synthetic content where the platform asks for it
One video doesn’t fit every platform.
Each one has its own rules.
1
41
Day 408: I paused one of my best ads. 🛑
It was still converting.
It was also starting to fatigue.
Keeping it meant riding it down.
Pausing it means testing what’s next while the account is still healthy.
It’s hard. A winner feels like income, and turning it off feels like losing money.
But the rule is simple:
Never let one ad carry the whole account.
What I’m doing:
Pause the fatigued winner (not delete)
Log its hook, angle and persona
Test new creatives in its place
Bring the concept back later with a fresh hook
A winning ad is a lead. The pipeline is the business.
3
3
54
Day 407: Third TikTok account. Ads are tired. 🔄
Two things today:
Warming up a third TikTok account 😅
Lesson from the first two: don’t give the algorithm a reason to flag you. Label AI content, no cross-posted duplicates, boring warmup first.
App #1 ads are fatiguing.
Good problem to have. It means they ran long enough to wear out.
The fix isn’t more budget. It’s more creatives:
Pause the fatigued ads
Add fresh angles per persona
Let each new ad hit 1K-2K impressions before judging
Winning ads don’t last forever.
A creative pipeline does.
Day 407: Fatigue isn’t failure. It’s a signal to feed the mac
1
4
69
Day 406: One shadowban, one plateau. 🧱
App #2:
TikTok shadowban again, caused by one AI-generated video.
Appeal sent. From now on: AI content gets labeled, and human-made videos stay in the mix.
App #1:
MRR is flattening around $1,100.
My instinct says: spend more.
My discipline says: check first.
Before raising budget, I’m looking at:
Frequency and CPI: are my ads fatigued?
Trial to paid rate: is the funnel still healthy?
Churn: is old MRR leaking out?
If the numbers are healthy, I scale budget gradually.
If the ads are tired, I feed Meta new creatives instead.
Plateaus aren’t a wall. They’re a diagnostic.
Day 405: Two apps, two missions 🎯
App #2:
Testing new creatives. One flop doesn’t make a verdict, so I’m feeding Meta fresh angles and letting the data talk.
App #1:
Pushing to break €1,500 MRR.
Where I stand: $1,111 MRR.
What’s driving the push:
Meta Ads running on the winning structure
ASO ranking holding in the top 10
More creatives cycling in so no single ad carries everything
The rule I keep repeating to myself:
Test creatives in batches, don’t touch the campaign every hour, judge on equal impressions.
2
6
109







