Let me start by saying how much I hate these kinds of guides. A random guy lucks into a few dollars and immediately starts telling everyone on the internet that if they follow his advice, they are going to make it big. It never works that way.
However, I get at least one or two DMs every day asking me for more details on how I grew @vaultaire_app, so instead of repeating myself constantly, I am writing it all down.
I decided to capture my whole journey instead of sharing the final state. It's quite long, so you might have to dig through it a bit.
The numbers
Everybody loves numbers, so here are mine.
Vaultaire launched on the App Store on March 13, 2026. The first version took roughly three weeks to build with the help of coding agents. The website went live slightly earlier, on February 27.
Vaultaire generated $66 in gross sales during its first partial month.
In July:
The website received 55,000 unique visitors, including 16,000 visitors using iOS devices
The App Store listing received 342,000 impressions
The app generated 17,100 downloads
1,287 users purchased a plan, representing a 7.53% download-to-paid conversion rate
Gross App Store sales reached $26,384
Those paying users consisted of:
545 annual purchases at $9.99
47 monthly purchases at $1.99
695 lifetime purchases at $29.99
Lifetime purchases accounted for approximately 54% of paying customers and 79% of July gross sales.
My current monthly costs are:
Cloudflare: $5
Codex: $200
Claude: $200
Apple Developer Program: $8.25
Apple Ads: approximately $700
Apple’s App Store commission: 15% through the Small Business Program
That leaves approximately $21,300 before income taxes, refunds, currency adjustments, and miscellaneous expenses.
The idea
I have never published an iOS app before. But for many years I had an idea for what is now the Vaultaire app. It's an application that allows a user to securely store private information on their iPhone without announcing it to everyone.
When I started I had a fully formed idea, which made the development quite easy. And I built it in roughly 3 weeks by mostly prompting a coding agent. That was surprisingly the easy part. Then I wanted to publish the app and I couldn't figure out what to do. There are so many rules that it felt more like joining a religion than publishing a piece of software. But with the help of my trusty LLM agent we figured it out. And after the first rejection, the app was finally published.
Even before the app was ready, I knew distribution would be the much harder part. I have built two fairly successful B2C services in the past and learned a valuable lesson: not the best product, but the best distribution wins. So I started with what I knew, the website.
I used to know a lot about SEO. I have built half a dozen companies and maybe two dozen projects over the last 25 years. But things had changed since I last worked on it. Then I discovered new terms: GEO, AEO, AIO, ASO. My head was spinning.
Here are the basic definitions:
SEO (Search Engine Optimization): Optimizing web pages to rank high in traditional search engine result pages (like Google or Bing) to drive organic clicks.
AEO (Answer Engine Optimization): Structuring content so direct questions are immediately answered, targeting featured snippets, voice search, and knowledge panels.
GEO (Generative Engine Optimization): Tailoring a brand and its website data to be cited, sourced, and recommended inside generative AI tools and conversational models like ChatGPT, Perplexity, or Google AI Overviews.
AIO (Artificial Intelligence Optimization / AI Overviews): Ensuring large language models (LLMs) accurately understand, trust, and present your brand data—though it is also frequently used to describe optimizing directly for Google AI Overviews or using AI to build content.
ASO (App Store Optimization): The process of improving mobile app visibility and conversion rates in app stores like Apple's App Store or Google Play (similar to SEO for websites).
This was so overwhelming that I had no clue where to start, so I went back to the playbook I knew:
Start with basic website & standard app metadata
Measure everything
Build on it as I go
I also never really "built in public" before. I had never seen the appeal for it. But building alone can be lonely. And it's fun to tap into the community. While it invites a bunch of jealous people and copycats, I have met some wonderful people so far. Not only that, but many gave me some great pointers and helped me with their advice. This is my poor attempt to pay it forward.
The design
Once the first version of the app was ready, I started working on the visual identity. I originally wanted to call the app just Vault, but I couldn't get a domain and the App Store is crowded with this term. So I looked for other options and came up with Vaultaire (ok, let's be honest, Claude came up with it by using InstantDomainSearch's MCP).
Once I had the domain, I needed a logo. If I am calling the app Vaultaire, why not make a logo with Voltaire in it? I used Nano Banana for it (do people still use it these days?).
I actually liked the red background one and used it for a while. But just couldn't really get into it. Had some ideas about animating it and doing some other stuff, but eventually decided against it. This is a serious utility app. One that is supposed to protect users' most prized data.
Played with many ideas
but finally settled on the current one
The website
While I was dealing with App Store review, I started building the website. With a logo in hand, I started looking at a design tool to build the site. I tried Codex, Claude, and Lovable, but they were all too generic. Then I found Variant which generates endless designs straight as HTML output.
After scrolling for maybe 10 minutes, it actually one-shotted what you see today.
I wired it into 11ty and connected it to Cloudflare so it's automatically built when new changes are produced.
Next up was tracking. Because you can't optimize without actual data. I am so old that I used to use Urchin before it got acquired by Google and renamed to Google Analytics. But that changed a lot. One now needs a degree to figure this one out. ChatGPT recommended Umami and I really liked it, but I worried it might be too simple. I also used Statcounter a very long time ago, but also feels pretty obsolete today. So I settled on Posthog. I truly dislike their gimmicky website, but the software is solid, especially with their MCP connected to an agent.
I then added the domain to Google Search Console, Bing Webmaster Tools and Yandex Webmaster to get some ideas about traffic. I also added it to Ahrefs and Semrush, but never found that much value in them (also never paid for them).
As the app was getting more finished and done, I decided to publish the site. Then I ran it through PageSpeed Insights and optimized everything it returned. Then I took a memory snapshot in Chrome:
Went to Chrome DevTools
In the Memory tab, took a memory snapshot and downloaded it
Gave it to my coding agent
Now I had a pretty decent baseline for the website and I actually made it public even though the app actually wasn't in the App Store. But I wanted to see if I can get some initial traffic going.
Then I focused on GEO. Used the GEO-first skill to audit and improve the site. I ran it in loops until there wasn't much to do. It complained ceaselessly about the app not being in the App Store. Drove me bananas. But overall it was helpful I think, especially around llms.txt and a few other small areas.
Next on the list was new content. I saw someone tweeting about AnswerThePublic where you can look up what people are searching for in ChatGPT and Gemini. I tried it. Wasn't that impressed but it gave me the initial seed. I started using Google's Keyword Planner but it was a lot of manual work to look up keywords, download, feed it to the agent, etc.
As usual, I decided to solve it programmatically:
It's much more efficient and allows me to manipulate all Google-related services through an agent, including the Search Console and Keyword Planner.
The last thing I have added was looking into "People Also Ask" on Reddit. Of course, this is when Reddit decided it wouldn't allow scraping anymore, and I was forced to find a way to reach the site from an agent without being blocked.
Once again, I decided to solve this programmatically:
These tools function as seeds for new content that is published daily on the site in the form of a guide or scenario. They are directly related to the queries users ask on Reddit or an LLM agent.
This mostly solved content, but I still couldn't get the site indexed much in any of the search engines, no matter what I did.
26 clicks in two months. Not good. Then I read something about domain rating. There is a lot of contradictory information. It's essentially some kind of internal number of Ahrefs that is not really relevant to Google. You know what is? Backlinks.
Submitting your site to lots of directories is tedious as hell. I used two or three services that do it for you for a few hundred dollars. At the same time I saw a suggestion to publish the same articles after a delay on Wordpress.com, Blogger.com and Medium.com. First, there wasn't much change but then all of a sudden the domain rating jumped
and traffic increased with it from Google to the site
For getting my website indexed and growing its organic search traffic, backlinks made a larger difference than anything else I tried. I cannot prove that the backlinks alone caused the increase, but the jump in Domain Rating and Google traffic followed shortly afterward.
The app
When I started working on the app, I was planning to make it free. Only changed my mind when it was almost done. The goal was to keep the free version very generous and only charge for additional features, like iCloud backup and sharing.
Technical setup
As a career engineer, I obsess over things that others might not care about, strong encryption, small binary size and the lowest attack surface, especially with network dependencies.
So my focus was on making Vaultaire incredibly small (just 7MB). The app itself has no third-party network dependencies and communicates only with Apple services. And of course it uses modern authenticated encryption.
These choices have consequences. The app has no recurring backend or infrastructure costs, and I never have to worry about servers, databases, or uptime. But I also have very little visibility when something breaks inside Apple’s ecosystem. Because Vaultaire is privacy-focused and hides protected content from screenshots and screen recordings, users often cannot easily show me what went wrong. And because every vault is cryptographically authenticated, a corrupted or missing file fails its integrity checks and becomes inaccessible. That protects users from silently altered data, but it also means I cannot reconstruct or recover a damaged vault.
Monitoring
As I said above, there is almost no monitoring available. I ask users to allow error collection but Apple App Review recently classified my optional error collection as tracking and required me to use ATT. Users of a privacy-focused app are not particularly keen on seeing that.
So I had to put it into app's setting where essentially nobody turns it on.
I have aggregate acquisition and revenue attribution through App Store Connect and campaign links, but almost no visibility into what users do inside the app. I can't easily test features or onboarding design. I have to rely on my judgement and what other developers have shared.
Pricing
That brought up a new question for me: pricing.
I read a few posts on pricing and settled on a simple structure: monthly, annual and lifetime plans, priced at $1.99, $9.99 and $29.99 respectively. This seemed like a good balance of options for the users. Most guides talk about not overwhelming users and pushing them towards a singular preferred option. In the case of a utility app it's the lifetime option over subscriptions. If people are like me, they hate paying subscriptions for these kinds of apps. It turned out to be right. Across all purchases since launch, lifetime is responsible for more than 60% of transactions.
I have now replaced the plans with weekly, annual and lifetime, priced at $1.99, $19.99 and $39.99 respectively. Strong demand gave me confidence that the original pricing was too low. The app has also gained substantial new functionality, including multi-device support and an upcoming native Mac app.
For the paywall I settled on a very simple design showing what the pro plan has. I got a few tips on how to improve it, but honestly didn't get to it. Or rather, the cycle to get the app reviewed is so long, that I am having a hard time motivating myself to play with it too much.
Around 7% of trial and subscription starts are eventually canceled, although I have not yet separated trial abandonment from cancellations by paying subscribers. Because of that, I do not yet have a reliable churn figure.
For the trial I read that longer trials convert better, so I settled on 7 days on a soft paywall. I want users to experience the full app but also don't want them to be blocked from using it. This is a utility app, not one of the maxxing ones. Also my operating cost is close to nonexistent.
Every app category behaves differently, so I would not blindly apply these pricing decisions outside utility apps.
Distribution
More challenging than the whole process of developing the app was producing all the associated design and metadata information. I genuinely struggled to figure out how to produce reasonable screenshots for the app.
Every post I saw emphasized how much the screenshots make a difference. So I bit the bullet and found this handy template for Figma. I made the first series by hand. Took me a full day.
I also kept seeing posts about ASO. Never heard of it before and the more I dug into it, the less it made sense. People recommended Astro which essentially exposes keywords from Apple's ad service. First I was confused what I was seeing or what to even do. But it has an MCP so the LLM agent could tap into and figure it out on its own.
ASO is now responsible for around 60% of all my downloads and I am glad I didn't skip it
One massive help with the app is asccli which allows agents to manipulate the store without me having to touch it. It's quite a godsend to be honest, because the whole interface was quite confusing the first time I used it.
Localization
Another thing I kept seeing constantly was to localize your app in other languages. The US is an extremely competitive market. But there are many others that are as lucrative but much easier to break through. Normally translations are challenging. I speak only a couple of languages. But with the modern LLMs, this is a breeze.
I started with the app. Each time I created a translation, I created one for the website as well. Every page, every guide, everything. Not only translating them to their languages, but also adjusting links to be more relevant to the users. So, for example, while the US version of a guide might link to an FTC resource, the EU version will point to the relevant European authority.
This made a huge difference in traffic. In fact the first uptick wasn't in the US, it was in Japan. Japan now accounts for nearly half of all App Store traffic and generates thousands of dollars in sales each month.
The more challenging part was screenshots. Figma's MCP is quite limited. While it works fine for most things, it's far from perfect. However my friend @tomkrcha and his team made an agentic design tool called Pencil
which works quite well for this. I use the MCP to control it instead of the models built into it.
Expansion & new platforms
I kept looking for new ways to expand the sources of traffic. I have never really played any social media status games, so my knowledge is limited.
The first idea I had was to turn the guides into visual guides and post them in various places: Linkedin, X and Pinterest. I asked Codex to generate them based on the published guides using ChatGPT Image 2. I tweaked a few things and then started posting. Linkedin and X produced almost nothing for me. But Pinterest keeps on giving and growing. Constantly.
311 clicks to the site resulting in 244 downloads, 51 paid conversions, generating $1,122 so far. All without paid distribution.
Apple Ads
My account included a $100 testing credit for their ads system so I decided to try it. I have actually never run any ads of any kind so this was quite interesting for me. I tried to manage it by hand and honestly wanted to throw my laptop out the window.
So once again, I solved it programmatically
With this update your agent can set up and manage all ad campaigns without you having to worry about it. In my case it takes all the information from Astro so the targeting seems to be quite precise. The results speak for themselves
Recently I started looking for new sources of traffic again and found this tweet quite handy
My first foray into it is a set of free browser-based tools. I picked those based on Google's Keyword Planner (through gum) and only built the ones that don't require a backend because I honestly don't want to manage one.
Based on Posthog, the tools are being slowly discovered by users and used. I am curious what it will look like in a few months
Future plans
The next one I really want to try is UGC. I am starting slowly, unrelated to Vaultaire, mostly just to learn the ins and outs of it. So far I haven't really figured it out, but it has only been a few days.
I am learning the idiosyncrasies of the platforms, what content works, how to best produce it and promote it, etc. The hardest part is to find an emotional trigger around which I can build the narrative. For a utility app like Vaultaire, this is not an easy feat. People use it because it solves their problem but they are not emotional about it. This is not one of the maxxing apps that helps with weight or quitting addiction, etc.
I experimented with Reddit, but that went quite poorly for me. I gave up on it after this. Honestly not super keen on investing in a platform where rules are so arbitrary that two mods can't agree on them.
The one that is quite out of left field but feels that it could be interesting is ads on adult sites. Vaultaire is about protecting private content, including adult content, so maybe it converts.
Parting words
I hope you found this guide helpful. I tried to share everything I have learned so far as clearly as I could. I would appreciate any feedback or suggestions for improving either my process or this guide.
And, of course, I would appreciate any likes and reposts.