
In this video we bundle a nodejs lambda function using lambda-build (powered by esbuild).

<p>
  <iframe width="560" height="315" src="https://www.youtube.com/embed/FmnFqjBk0to" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
</p>

### ⚡️ Links:
- <a target="_blank" href="https://github.com/alexkrkn/lambda-build">lambda-build</a>
- <a target="_blank" href="https://lambdabuddy.com">LambdaBuddy</a>

Use esbuild to create a bundle of your lambda function

Bundle a nodejs lambda function


In iOS 14.5 Apple made changes to the way attribution works in iOS, due to privacy concerns.

The good news: There is a new way to track campaign attribution of mobile campaigns on major ad networks (such as Facebook, Instagram, TikTok), by using **SKAdNetwork** and one of the measurement partners (such as branch.io).

The bad news: It's a mess to try to figure out what's going on with attribution on iOS, because now there are several ways to measure campaign results, each one giving only partial results. This means that you need to be aware of the differences between the methods when setting up campaigns. In most cases, it's probably better to focus on modern attribution and avoid the confusion altogether. Also, google search is full of tutorials and guides about attribution that are outdated and only add to the confusion.

This post is my attempt to briefly explain the different attribution methods, old and new, so that if you're new to iOS attribution tracking, and just want to get it up and running, you can understand what's important and what you can ignore.

## SKAdNetwork - ✅ the new method (works for Facebook, Tiktok, Youtube etc)

Also referred to as: "SKAN".

Apple introduced <a href="https://developer.apple.com/documentation/storekit/skadnetwork/" rel="nofollow" target="_blank">SKAdNetwork</a> as a new alternative to measure the success of ad campaigns, while maintaining user's privacy. This is the recommended way to measure campaigns. As time passes, I assume, Apple will add more limitations to alternative methods and push SKAdNetwork to become the the only way allowed. When using SKAdNetwork the app **does not** need to ask the user for permission to track their attribution to a campaign, because the data is anonymized anyway. It will always work, regardless of the user's tracking authorization status.

To use SKAdNetwork, you'll also need an authorized MMP (Mobile Measurement Partner) such as branch.io, appsflyer, etc. You will install their SDK and integrate it in your app. This MMP will interact with the advertiser's APIs and Apple's APIs on your behalf and show you the final results in their dashboard.

It might be confusing that you need an additional service to track your ads, but due to how SKAdNetwork works, it'll be much easier to use one. These MMPs are vetted and authorized by both Apple and the advertisers. I recommend branch.io because you can just create an account and get started right away, the iOS SDK docs are clear and thorough, and they have a generous free plan so you can become familiar with it and make sure that it works for you, before committing.

One caveat is that if you advertise **only** on Meta (Facebook, Instagram and WhatsApp) then you might not need an MMP because Meta has all this functionality built in. They already have the SDK and the analytics dashboard of their own that integrates with SKAdNetwork. However, TikTok doesn't have this option and they require you to choose one of the authorized MMPs in order to track attributions at all. Also, by using an MMP you will be able to see all the campaigns across all advertisers in a single place.

A few important points to understand about how SKAdNetwork works:
- You can see for each campaign how many conversions they generated, but in aggregate. You cannot associate a specific user to a specific campaign, like you could before iOS 14.5 using IDFA (described below). Apple maintains crowd anonymity so you cannot identify any specific user.
- There is no real time information, the attribution results are purposely delayed by Apple so you cannot identify specific users. There's also a minimum conversion threshold required to show any results, for the same reason.
- The attribution window is limited in time. If 24 hours pass with no new attribution data for a given user, then no more attributions will be measured for this user at all.

## AdServices - ✅ works for Apple Search Ads

SKAdNetwork is meant to be used with any of the big advertisers like Facebook, Tiktok, Youtube etc. If you're using Apple search ads, you can use the much simpler <a href="https://developer.apple.com/documentation/ad_services" target="_blank" rel="nofollow">AdServices</a> which you can implement with a simple callback and without using an MMP. But since you need to use an MMP anyway, you might as well send your Apple Search Ads attributions there too, because they support it.

AdServices works only with Apple Search Ads, and allows you to attribute the campaign and keyword to a specific user. You might be wondering, how come specific attribution is allowed when coming from Apple Search ads but not when coming from the other advertisers? I would like to understand as well.

## IDFA - 🚫 the old method

Also referred to: "App Tracking Transparency", "ATT".

Apple used to provide a persistent id (called IDFA - "identifier for advertisers") to identify all users. It allowed you to track the user and how they responded to ads, across campaigns and apps. It was a simple and powerful solution that was easy to implement and allowed full transparency, but hurt the user's privacy and was abused by advertisers.

In iOS 14.5 Apple made big changes to how IDFA works, due to privacy concerns. If it was fully deprecated, it would probably make everyone's life much easier. However, it was only 'limited' instead. From iOS 14.5 and on, users are presented with an intimidating popup, asking them if they <a href="https://developer.apple.com/documentation/apptrackingtransparency" target="_blank" rel="nofollow">agree to be tracked</a>. If they agree, the IDFA is provided like before, if they refuse, it's not provided and the campaign performance cannot be tracked using the IDFA.

The rate of users that agree to be tracked is obviously very low (according to various reports) and keeps declining. If you have limited marketing resources, or just getting started, you probably want to ignore this option and only use the new attribution method. However, the MMPs still keep this option in their campaign creation and management tools, which complicates campaign management.

By using this attribution method, you get limited data, only from the users who consented to be tracked. If you use it alongside SKAdNetwork, you'll essentially get two data sets that you cannot cross reference, because there's no way to match attributions. It might make sense for companies with large budgets and high error tolerance, who still try to hold on to any tracking information they can get, but if you're just looking for simple attribution data, it might be better to simply ignore everything related to IDFA and focus all of your resources on the new attribution methods.

## Universal Deferred Links - 🚫 the method that "only sometimes" works

Another method that used to work to track attributions is "universal deferred links" but there were changes to how it works as well, so it can no longer be used reliably.

This method used to work like this:
- You would append an id to the url of your app in the app store.
- A user would use this link, go to the app store and download the app.
- When the app is opened, you would get a callback in your code with the id from the url as a parameter, so you can associate it with the user.

Again, this was a simple and efficient way to track attributions, and with this case as well, it no longer reliably works. The callback in your code (step 3 above) will have an empty result when the link is opened in certain apps, which is Apple's way of preventing this attribution method as being used as a workaround to track users. For example, it will work when opened from an email or messages app, but not from Instagram.

You can <a target="blank" rel="nofollow" href="https://help.branch.io/developers-hub/docs/ios-universal-links#appsbrowsers-that-support-universal-links">read more here</a> about when universal links work and when they don't.

It's probably still a good idea to implement universal links, because the setup is easy and you'll get more data, especially for email/sms campaigns.

There's a somewhat popular workaround to make universal links work despite the limitations, using the clipboard. It works like this: the link sends the user to a webpage which copies the link to the clipboard, the user installs the app, the app reads whatever is in the clipboard to be able to make the attribution. The problem is that the user will see a notification that something was read from their clipboard, when the app is just opened which looks suspect. And worse than that, this is clearly a workaround that Apple tries to prevent, so I assume at some point they will find a way to stop it from working or start rejecting apps which read from the clipboard when opened.

## Attribution on iOS in the future

I think the same idea applies to all of the attribution methods, it's clear that Apple is trying to prevent users from being linked to specific clicks, and since Apple has all the power, it's probably just a matter of time until SKAdNetwork (or a future variation of it) will remain as the only way to track attributions.

If you find this post useful and would like to get notified about my next one, make sure to [subscribe to my newsletter](https://newsletter.undefinedapps.com/) and follow me on [𝕏](https://x.com/alexkrkn).

Simplifying the mess around iOS 14.5+ attribution; SKAdNetwork vs IDFA/ATT


<img height="300" src="/posts/storage-cleaner-app.png" alt="storage cleaner apps" />

## A crowded niche

Both the 'phone cleaner' and 'storage cleaner' search queries on the app store give around 550 results. Which means that if you make a new app in this category, your app (which will likely be somewhere at the bottom) won't be seen by any new users. How do you climb up?

The apps that appear on top of this search results, which are also the top grossing apps, have a high rating, many reviews and already proven to Apple's algorithm that they can convert "app page views" to revenue, so the App Store places them first in the results, and they grow even more.

They also look VERY similar to each other:
- Use similar names with the same keywords (phone, cleaner, storage)
- Almost identical icons (a white brush+sparking on a blue background)

### A few examples:

- <a href="https://apps.apple.com/us/app/cleanup-phone-storage-cleaner/id1510944943" target="_blank" rel="nofollow">Cleanup: Phone Storage Cleaner</a>
- <a href="https://apps.apple.com/us/app/smart-cleaner-clean-storage/id1194582243" target="_blank" rel="nofollow">Smart Cleaner: Clean Storage</a>
- <a href="https://apps.apple.com/us/app/phone-cleaner-clean-storage/id1463756032" target="_blank" rel="nofollow">Phone Cleaner: Clean Storage+</a>
- <a href="https://apps.apple.com/us/app/powerful-cleaner-clean-storage/id1618902724" target="_blank" rel="nofollow">Powerful Cleaner-Clean Storage</a>
- <a href="https://apps.apple.com/us/app/super-cleaner-cleanup-master/id1630312428" target="_blank" rel="nofollow">Super Cleaner - Cleanup Master</a>

But there's an app in the same niche, that was able to squeeze itself in between them. It looks different and uses a creative growth strategy with TikTok ads. I like how this app is able to stand out and the creativity behind its marketing.

- <a href="https://apps.apple.com/us/app/photo-cleaner-swipewipe/id1583884012" target="_blank" rel="nofollow">Photo Cleaner: Swipewipe</a>

The app uses a different icon, keywords and UI than the other apps and it stands out. But these differences don't explain the app's results in isolation. They are effective because of another major component, which is the TikTok campaign. I think it's interesting to simultaneously look at how this app is different, and the campaign, to try to understand the app's marketing, which proved to be very effective.

## The campaign

A few examples of the video ads on TikTok that the app uses:

<a href="https://www.tiktok.com/@makingmemorieswithjess/video/7243771071910907182" rel="nofollow" target="_blank">41M views</a> - by an account with 114K followers

<a href="https://www.tiktok.com/@bethmikeska/video/7244976831529487659" rel="nofollow" target="_blank">24.2M views</a> - by an account with 49K followers

<a href="https://www.tiktok.com/@madison_parkinson1/video/7233880717036489989" rel="nofollow" target="_blank">36.6M views</a> - by an account with 55K followers

<a href="https://www.tiktok.com/@tatumthebates/video/7260405028056337707" rel="nofollow" target="_blank">8.9M views</a> - by an account with 88K followers

<a href="https://www.tiktok.com/@neeenzs/video/7262026160345812257" rel="nofollow" target="_blank">2.6M views</a> - by an account with 790K followers

<a href="https://www.tiktok.com/@isapressoto/video/7281045015810362656" rel="nofollow" target="_blank">5.6M views (Portuguese)</a> - by an account with 900K followers

<a href="https://www.tiktok.com/@eularifreitas/video/7277584397333761285" rel="nofollow" target="_blank">1.4M views (Portuguese)</a> - by an account with 765K followers

There are a few things in common for these ads:
- They mostly target a younger female audience.
- The are all made by creators with the range of 50-100K (with a few exceptions), and posted to their personal feeds.
- There's no rigid script, and I assume the creators decide how the video should be.
- These videos are heavily promoted, besides being posted to the creator accounts - and have many more views than the average video of these creators. For example, the first example above, got over 40M views, but the creator gets below 5K views on average for her organic videos.
- All the videos feature the UI of the app in use. The UI is very colorful, has large buttons and is easy to use - it's perfect for a quick TikTok demonstration.
- Some of the videos emphasize how fun and easy it is to use the app - and focus first on going through the pictures, and the photos cleanup part is secondary. For example, the script for the first video, which demos the app in use:
  - what a fun way to go through the pictures
  - swipe right to keep
  - swipe left to delete
  - videos automatically play
  - super easy to delete in bulk
  - saves so much time and storage on your phone

## The App itself

The metadata of the app makes more sense when examined together with the campaign above, which helps explain why this app decided to go with different metadata than the other apps which are already proven.

- **The icon** - you can immediately notice how the color is different from the blue color that ALL the other apps use. If you google "apricot color" in google images, you'll get mostly results of makeup and other products that target a female audience

- **The screenshots** - The app's screenshots on the app store page, promote "easy and fun" cleaning, using swiping. Unlike all the other apps that focus on the benefits of cleaning up storage, this app focuses on ease of use and the process.

- **The keywords** - The title of an app is the most important place to put the keywords you want to compete for. All the other apps in the niche all stuff the title with the most critical keywords, which there is also a lot of competition for. This app's title 'Photo Cleaner: Swipewipe' stands out in the fact that it isn't so concerned with the keywords, and tries to have an original app name instead. But, the app's name 'swipewipe' is still placed last in the title, to emphasize 'photo' and 'cleaner' for the algorithm.

- **The UI** - The UI has large buttons, in all caps and is very easy and simple to use, very focused on the fun photo viewing experience. If I would imagine a UI that is perfect for TikTok demonstrations, even at the cost of usability, it would be it. (A little like the fake food outside of a Japanese restaurant).

## Takeaway

This is another example for how authentic short video ads are proving to be an effective growth strategy for mobile apps. I though it's especially interesting how this app extended the tiktok campaign into the app itself by optimizing the metadata and even the ui to improve the campaign results. What do you think about this campaign? [Let me know by replying to this tweet](https://twitter.com/alexkrkn/status/1712481352485052631).

If you enjoyed this analysis and would like to get notified about my next one, make sure to [subscribe to my newsletter](https://newsletter.undefinedapps.com/) and follow me on [𝕏](https://x.com/alexkrkn).

How an iPhone App Beat a Crowded Niche Using Creativity


In my [previous post](/post/analyze-app-store-reviews-app-storage-cleanup), I analyzed the reviews of an iphone app to see what I can learn about it and its marketing, without even trying it. One of the common things mentioned in the reviews were the social media ads that brought the user to the app in the first place. I thought it was interesting because it helps explain why this specific app was appearing high in the App Store search results while there are many other very similar apps that barely get any usage. In the post below, I'm going to analyze the social media ads of these apps to see what I can learn about them.

## Do social media ads even work?

Besides the obvious benefit of converting some traffic from ads to paying users, there's another benefit for ads, which is that it also helps with the organic placement of the app in the App Store search results. Let's say that a user from the ad didn't end up paying, the app can still get some benefits from acquiring this user:
- *App Store conversion stats* - Even just the fact that the user viewed the app page and then ended up downloading it, signals to Apple that this app converts views to downloads, which affects the algorithm and places the app higher organically.
- *App Store reviews* - It will try to have the user leave a positive review. Reviews have a big influence on organic search results on the App Store, so this will improve organic rankings as well.
- *App Store retention stats* - It will let the user use the app for free with limited functionality. Apple tracks app usage and retention, so increased usage will again influence the algorithm.
- *Ad revenue* - The app lets users unlock paid features without paying, but by watching video ads instead. So even users who don't pay can bring in some revenue for the app.

The app helps Apple by bringing in new traffic to the App Store, and Apple in return places the app higher in the search results. The other apps in this saturated niche, that don't use paid ads, stay lonely at the bottom, even if they are potentially better or cheaper. Most of this applies to any other platform, like the Play Store, Shopify apps, etc since they all have some sort of algorithm to display search results for their users, and they try to optimize it for the most app usage, reviews and revenue.

## The social media ads

I looked at many different ads for the niche of phone cleaning apps and noticed a few patterns that I thought are interesting, which I describe below. But on a side note, I thought it's amusing how much effort goes into educating people about how to clear storage on their phone. People from different walks of life, different ages, and in various languages explain the same iPhone popup of running out of storage, and how to deal with it, by using the app they are promoting.

The main challenge of video ads is to grab the user's attention fast and stop them from scrolling, even for a second, to give the video a chance. That's why in all of the ads, the first screen and 2-3 seconds are always dramatic, interesting and fast. Only after a few interesting seconds, the ads start talking about the app and show why it's useful.

## Two types of ads

There are two types of ads that I found interesting:
- A person presenting - this kind has the benefit of being more relatable, but dependent on the creator and harder to scale.
- Without a presenter - this kind is scalable and makes it possible to create and test many variations, but less personal.

## Type 1: Ads with a person presenting

<img height="300" src="/posts/ad-with-presenter.jpg" alt="tiktok video ad with a presenter" />

<a target="_blank" rel="nofollow" href="https://www.tiktok.com/@kessel_nathan_official/video/7278011304420363563">Example 1</a>

@kessel_nathan_official is a creator with 10.2M followers who makes tiktok videos unrelated to mobile apps.

The video was posted on his channel so there's the benefit of having his large audience exposed to the ad and will more likely to watch it because they are familiar with him. The video got around 1.2M views.

While many other ads use generic screens and an AI voice, in this ad the creator speaks and shows up in the video which makes it more engaging, but I assume also costs much more.

The first two seconds are grabbing attention:
- A human face speaking to the camera, someone who is already proven to be popular on this platform.
- Mentions a couple's relationship FAST: "my girlfriend and I".
- Gets dramatic fast: "so fed up!"
- Uses dramatic zoom effects and is very fast paced.
There's nothing related to any app or storage cleaning at all. Everything is about grabbing attention fast.

The following part starts talking about the problem, but still full of pictures of a couple and a funny video of a cat, and is very fast paced. It describes the problem that gets them "super mad" which is... you guessed it! Running out of iPhone storage 🤣.

Only in the last part it talks about the app which is the solution to their problem, and shows how to use the app to solve the storage issue.

A nice touch is that in the video, the viewers are sent to the app store to search for the app and download it. I think that this video runs both as an ad (with a direct link) and also appears on the creators page (as a regular video, without a link). So by sending the viewers to the app store to search, gets some additional traffic for the advertiser when the video doesn't have a direct link. It will also likely to increase the ranking position of the app for this keyword. Although, it is a little risky because they might choose a different app from the results, given there are so many similar apps. I'm not sure that's the case because I'm not very familiar with TikTok so let me know if I got it wrong.

## More examples - a person presenting

- <a target="_blank" rel="nofollow" href="https://www.tiktok.com/@michaeltowns/video/7177488090523684142">Example 2</a> - 4.6M views and 65K likes, by a creator with 65K followers.
- <a target="_blank" rel="nofollow" href="https://www.tiktok.com/@gertieinar/video/7077144011991010565">Example 3</a> - 3.8M views and 287K likes, by a creator with 2.8M followers.
- <a target="_blank" rel="nofollow" href="https://www.tiktok.com/@aaliyahhh.5280/video/7110651751950273835">Example 4</a> - 1.1M views and 42K likes, by a creator with 92.3K followers.
- <a target="_blank" rel="nofollow" href="https://www.tiktok.com/@emiliavlogs/video/7118155562633760006">Example 5</a> - 220K views 7.5K likes, by a creator with 1.2M followers.

A few things these ads have in common:
- Starts with a human in the frame
- A dramatic claim or problem right away, such as "Buying storage for your iPhone is a scam!!"
- Fast transitions and a lot of things going on the screen at once
- Shows actual screens of the app and the app in use by a real person

## Type 2: Ads without a presenter

<img height="300" src="/posts/ad-no-presenter.jpg" alt="tiktok video ad without a presenter" />

The second type of ads is without a presenter. These ads usually show hands using an app running on a real physical phone. They can also feature a short part of video of a person pointing to a text or doing something, but not speaking, and can be from a stock footage library. The audio is these ads is either a narration or uses the AI text-to-voice option.

There are many benefits to making ads without a presenter:
- you're not dependent on a specific person, so it's easier to make many ad variations
- anyone, anywhere can be the "hands" that show something on the phone - which can significantly reduce the cost and make the process more scalable
- you can use the AI text-to-voice option (which is very popular anyway and some creators even prefer to use it themselves)
- you can experiment with many variations of audio and stock footage

The main disadvantages are that they are less relatable, and there's usually no "plot".

An example of this kind of ads is the following ad that got 40M(!) views and 125K likes, made by a creator with 114K followers, but she doesn't appear in it at all, and uses the AI voice for narration.

<a target="_blank" rel="nofollow" href="https://www.tiktok.com/@makingmemorieswithjess/video/7243771071910907182">Example 1</a>

A few things I found interesting:
- The voice used is the AI text-to-voice option, even though it's made by a creator that has many videos of her speaking in them - which shows that this AI voice can be even better sometimes.
- The video is essentially just hands showing how she uses the app, but all the photos are of real people, not just photos of nature or nice views.
- It starts with the caption and text of "What a fun way to go through your pictures?" which is different from all the other cleaning apps I mentioned before, even though this app still competes for the same traffic of storage cleaners (it's called "Photo Cleaner: Swipewipe").

I like the ui of this app and how it has a creative angle on this whole "phone cleaning" niche. Instead of just making another app that looks exactly like all the other ones, they found a creative way to stand out. The ui of the app seems like it is perfect for a tiktok video: large buttons, colorful, swiping, etc

## More examples - Ads without a presenter

<a target="_blank" rel="nofollow" href="https://www.tiktok.com/@julianalexanderthies/video/7153617447122144518">Example 2</a>

This video got 677K views and 20.1K likes, made by a creator with 1.7M followers. It also uses the AI voice and shows hands using the app on the phone. But this ad starts with a more generic "Clean up your storage on your iPhone" and then goes to showing the app right away. On the other hand, this video is very similar to the other videos on this channel that 1.7M followers trust and follow, so it probably has trust already established and they might even like this straightforward style.

- <a target="_blank" rel="nofollow" href="https://www.tiktok.com/@xyl0vrr/video/7201101528227253531">Example 3</a> - 78K views and 3.2K likes, by a creator with 25K followers.
- <a target="_blank" rel="nofollow" href="https://www.tiktok.com/@peachiesuga/video/7228380125627206919">Example 4</a> - 64K views and 4.1K likes, by a creator with 24K followers.
- <a target="_blank" rel="nofollow" href="https://www.tiktok.com/@betimm_b/video/7223454887823133979">Example 5</a> - 140K views and 2K likes, by a creator with 1.9K followers.

## Example of many variations for the same app

The reason I initially started writing this post is because in my previous post I analyzed the reviews on an iPhone app and many of them mentioned the instagram and social media apps that brought them to the app in the first place. Using Facebook's page transparency features, it's possible to see all of these ads <a target="_blank" rel="nofollow" href=" https://www.facebook.com/ads/library/?active_status=all&ad_type=all&country=ALL&view_all_page_id=112192393947217&search_type=page&media_type=all">on their Facebook page</a>.

It has a huge variety of ads, both with presenters and without, translated to different languages. Many of these ad variations seem like they were fully automated with AI voices and generic stock videos.

In all of these ads, notice the first two seconds and how they try to capture attention. Sometimes it's a phone being washed with water and soap, sometimes it starts with the problem of "no storage" right away.

## Short form videos are effective to promote apps

I was surprised to see how much content and ads there are on tiktok that is related to showcasing or promoting mobile apps and saas products. I'm not using it myself and I always thought it's mostly for dancing videos, but it seems like many app makers are successfully using tiktok for marketing. Have you tried tiktok ads? What do you think about the examples above? [Let me know by replying to this tweet](https://twitter.com/alexkrkn/status/1709185906194702370).

If you enjoyed this analysis and would like to get notified about my next one, make sure to [subscribe to my newsletter](https://newsletter.undefinedapps.com/) and follow me on [𝕏](https://x.com/alexkrkn).

Analyzing social media ads of small iphone apps that bring in millions in monthly revenue


In my previous post I wrote about how to [learn from the reviews of a competing app](/post/analyzing-competitor-reviews-apple-app-store). In this post, I'm going to look at a real app and analyze its reviews to check what we can learn from them. We'll take a look at what the users like, how they use the app, what they don't like, and which features they wish the app had.

## Getting the reviews to analyze

We'll take a look at the US reviews of the app <a href="https://apps.apple.com/us/app/cleanup-phone-storage-cleaner/id1510944943" rel="nofollow" target="_blank">Cleanup: Phone Storage Cleaner</a> - it's a very popular app with around 200K ratings and a 4.7 average. According to <a href="https://appfigures.com/" target="_blank" rel="nofollow">appfigures</a>, this app makes around $2M per month.

You can read the reviews on the App Store by going to this <a href="https://apps.apple.com/us/app/cleanup-phone-storage-cleaner/id1510944943?see-all=reviews" target="_blank" rel="nofollow">link</a>.

I didn't even try the app, and all of the conclusions are solely from reading the reviews. I think you'll be surprised how much it's possible to learn just from user's reviews.

## Trigger to download

Maybe you've seen the banner on your iPhone letting you know that you are running out of storage. I've had it pop up before and it's annoying, but this app turned it into an estimated $2M+ a month business. Let's analyze how.

Many of the reviewers mention that this banner is what triggered their search, and download. A very common reason for running out of space is that when we take photos, instead of just taking one and risking a bad one, we snap a bunch and plan to choose the best one and delete the other ones. But of course, we never get to do it, so we end up with many more photos than we need until there's no storage left.

Luckily, Apple offers a built in way to delete duplicate photos, as documented <a href="https://support.apple.com/guide/iphone/find-and-delete-duplicate-photos-and-videos-iph1978d9c23/ios" target="_blank">here</a>. But not many know about this solution. The "cleanup app", on the other hand, is constantly appearing on Instagram ads (as mentioned by some reviewers) and appears high in results when users search for a way to "cleanup storage" on the App Store. Also, I think the app finds 'similar' photos, whereas the built-in feature works only with identical duplicates.

But before you get upset at the app for monetizing a solution that exists for free, remember that Apple could easily have added a link from the banner to a "storage cleanup" screen with this feature, but instead there's a link to purchase more storage from Apple. In fact, some reviewers mention that this app is a better solution for them than paying Apple for more storage.

## Onboarding / Monetization

Once the potential user clicks on the Instagram ad or finds the app on the App Store and downloads it, they can use it for free but with a limit of 5 photos a day. To delete more than 5 photos, you have two options: either watch ads to delete 25 more photos per day, or pay $5 a week (with a free trial). There's also an option for a one-time lifetime charge that you pay once to unlock the app forever.

Many reviewers mention the option to watch a video ad instead of paying as a huge positive. These users would probably never pay anyway, and would not use the app at all. But instead, the app maker makes money from ad views, and also gets more downloads and positive reviews.

Given that this app is already estimated to generate $2M/month revenue just from in-app payments, the fact that it also probably makes a similar amount from showing ads is amazing.

## 'Duplicate Photos Removal' VS 'Storage Cleaner'

One interesting thing I've noticed is the app's screenshots mention a variety of different features, but the reviewers all mention only the feature of removing duplicate photos.

The app's screenshots list these additional features:
- merge duplicate contacts
- clean your mailbox
- hide private photos & videos behind a passcode
- set charging animations (that's random and kind of unrelated)

But it seems like none of the reviewers cares about these additional features at all. Only the contacts feature was mentioned once, but besides that, the main use of the app is to delete duplicates.

I can think of two reasons why the screenshots target more features than actually used:

1) It's possible that mentioning a variety of different features converts potential users better and makes them download the app, even though these features don't get any significant usage. Maybe, when searching for apps, users prefer to see an app with many ways to clean up storage, and they choose it over another app that only deletes duplicates. However, when they actually download the app, they end up using only this one feature.

Do new users even know that they are looking for a duplicate photos remover? Their problem, as they understand it, is that they ran out of storage and they look for a "storage cleaner" or "phone cleaner" first and foremost.

2) It's also possible that it would be better to target only the removal of duplicate photos and the app maker just didn't check their usage analytics, or reviewers simply don't mention the other features even though they do use them.

I tend to think it's option number 1, because the app is doing so well and it's well optimized in all other aspects. I think it's more likely that the app maker tried different variations and ended up with a set of screenshots with the highest conversion rate to app downloads.

## Time and storage saved

My favorite thing about the app is how it shows users how much storage space and time they saved by using it. You can see this in one of the screenshots as well.

It's possible to calculate the storage space accurately, but the 'time saved' number is just a very rough estimate. Nevertheless, I was surprised to see how many reviewers mentioned in the comments how much storage and time they saved! Some reviews even mention the actual numbers the app gave them.

It's such a small feature and may look insignificant, but according to the reviews, this feature seems to have a big impact on users and the perception of the value the app provides for them.

## Ads cost money

$4.99 per WEEK is a steep price for an app that likely runs on the device and doesn't even have many servers or API costs. There are many reviewers who complain that the app should be free or cheaper.

But consider that this app is one of the most popular cleaning apps on the App Store, and from other reviews, it seems like they probably do spend a lot of money on advertising. There could be a better, cheaper app out there but we're not talking about it because users simply don't know about it. This hypothetical app can afford to be much cheaper because it doesn't spend so much money on ads.

It seems like in this category of cleaning apps, all of the top apps already optimize for ASO and keywords, and all of them have many reviews. So I don't think it's practical to compete for organic traffic for these keywords on the App Store.

So in reality, there can be two options:
- A cheap or even free app - at the bottom of the search results, that no one knows about.
- An expensive app - that spends a lot of money on ads which brings a lot of users that cover the expenses of the ads and generate some profit for the app maker.

## Common issues with the app

There are a few issues with the app that come up repeatedly in reviews. Some complain that it's slow, and sometimes crashes and loses all progress of images chosen for deletion. There are also a few complaints about the functionality; specifically, that the app does not identify all of the photo duplicates.

Other issues that come up including the photos being randomly rearranged after deletion, or it not being clear which album they were moved to.

One interesting and common complaint was that when users go through many photos, sometimes spending hours on it, and then when they try to delete, the app glitches and loses all progress. Would it be better to delete photos in batches? Or is it possible that the app lets the user select many photos before showing them the payment prompt, so that they will be more likely to pay after spending so much time on marking photos to delete? We can't know for sure from just reading the reviews, but it's an interesting question to explore.

There are also many positives about the app that are constantly mentioned. Many users mention that this app is much easier and better than any other app they have tried.

One thing to keep in mind is that the reviews are likely to have more users with negative experience because they are the ones who take the time to write a review. Also, with such a large number of users, there will always be instances of glitches and unhappy users.

## Keywords reviewers use

While reading the reviews, I've noticed a few terms which the reviewers keep mentioning. I would assume that these are also the terms that potential users use to search for a similar app.

Some of the keywords are as follows:
- clean/clear up, delete, organize, get rid of, remove, free up
- similar, duplicate
- photos, pictures, storage, space

Overall, the app does a great job with ASO and uses all the main keywords in the title and subtitle.

## Similar apps

Are you surprised that this kind of app is so popular and makes so much money? Then you might also be surprised to learn that it's only one app out of many in a highly competitive niche, with other apps making similar amounts or more. Check out these apps for comparison: <a target="_blank" rel="nofollow" href="https://apps.apple.com/us/app/smart-cleaner-clean-storage/id1194582243">Smart Cleaner</a>, <a target="_blank" rel="nofollow" href="https://apps.apple.com/us/app/cleaner-guru-cleaning-app/id1476380919">Cleaner Guru</a>, <a target="_blank" rel="nofollow" href="https://apps.apple.com/us/app/phone-cleaner-clean-storage/id1463756032">Phone Cleaner</a> and there are many many more similar apps.

## Conclusion

It's fascinating to me how much information one can get just from reading the reviews, without even trying the app at all. Listening to and interviewing real users is an important part of marketing, and reading reviews from real users is an efficient way to do market research.

Was there something that surprised you in my analysis? Do you use a similar approach - reading reviews - when doing market research? [Let me know in the comments](https://twitter.com/alexkrkn/status/1700876629898113499).

If you enjoyed this analysis and would like to get notified about my next one, make sure to [subscribe to my newsletter](https://newsletter.undefinedapps.com/) and follow me on [𝕏](https://x.com/alexkrkn).

Let's see what we can learn about an app just by reading its reviews

Decoding a $2M/Month App's Marketing Strategy Through Its App Store Reviews


If you're building an app, there will always be other apps that offer similar functionality and compete with you for downloads and users. It could be a direct competitor competing for the same users, or just a similar app with a few overlapping features.

What if there was a way to learn more about the users of these competing apps? Learn what they like about the app and why they use it, what they don't like about it, and what features they wish this app had.

You can learn all that by reading the app's public reviews. Just by reading reviews, you can get a lot of ideas for what to build, what to avoid and how to market your own app.

## Learn what to build

Many times reviewers mention features that they wish the app had. This is, of course, an excellent opportunity to learn about features you can implement yourself. It's possible that the app's creator either doesn't want to or can't build this feature. They may simply have other priorities and are not interested in working on it. This is a good opportunity to think about whether that is something you want to build. If many users request this feature, maybe it could even become a main offering of a new app that targets these users specifically?

Also, look for features that users mention as likable. Do you have these features as well? If you do, maybe these are features that you should emphasize in your marketing. After all, if it's mentioned repeatedly it's also likely that other users also search for them when looking for a new app to download. Features that are specifically mentioned as likable will also provide insight into common use cases of the app and the target audience.

You can also look for what reviewers complain about. Can you build something that fixes their frustration and gives them what they want? Even if you don't come up with a new feature, at least you now know what users don't like and what to avoid.

If the reviews of a popular app mention bugs, unfriendly UI and a slow app, it could be an opportunity for improvement. It likely means that this app is doing a good job with marketing and satisfies a real need users have, but the app itself can be improved upon. You might be surprised how common this is for very popular apps. It just goes to show how marketing is a more dominant factor than the quality of the app, like it or not.

## Marketing material

Positive reviews that mention what the app does well can highlight which features are the most popular. If your own app has one of these features, maybe you can emphasize it more in your marketing? It could be worth trying to put these features on your first screenshots.

Try to look for keywords and phrases that the reviewers use to describe these features. If these features are popular, maybe you can try to rank higher for these keywords or create a paid campaign for them. New users are likely to look for these features and use these keywords when trying to find an app.

Reviews can also mention questions or common confusion. This could be an opportunity to answer these questions with your marketing or try to write content for SEO.

## Analyze your own apps

If you have an app, you should read your users' reviews and analyze them for the same things I've mentioned above. You are probably already notified about each new review that you receive. However, it can be helpful to take a step back. Consider analyzing all your reviews from the past few months as a whole to identify common patterns, frustrations, and complaints.

## Let's analyze the reviews of a real app

In my next post, I'm going to [choose a popular app and analyze its reviews](/post/analyze-app-store-reviews-app-storage-cleanup). Let's see what we can learn from these reviews and try to find all of the points I've mentioned above.

## Conclusion

Analyzing public reviews of a competing app can offer invaluable insights for various aspects of your own app. Do you read other apps' reviews? What are some of the things you learned? What are some other methods you use to analyze competing apps? [Let me know in the comments](https://twitter.com/alexkrkn/status/1697595039088558327).

If you enjoyed this post and would like to get notified about my next one, make sure to [subscribe to my newsletter](https://newsletter.undefinedapps.com/) and follow me on [𝕏](https://x.com/alexkrkn).


An underestimated way of improving your app and your marketing

How Analyzing Reviews of Competing Apps Can Boost Your Marketing


<p>
  <iframe width="560" height="315" src="https://www.youtube.com/embed/pOda6TDBqcY" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" allowfullscreen></iframe>
</p>

It often happens to me that I need to run some cli command and I know exactly what I'm trying to accomplish, but I can't remember how the tool I'm using works and which flags it provides. Especially when using ImageMagick or FFmpeg.

In the past I used to either read the docs to try to find what I need, or google my question. But since ChatGPT came out, I started using it to write my cli commands for me. I tell it what my OS is and what I want to accomplish, and then it returns the exact command I need and also the explanation for each flag.

It's not only much faster but also sometimes it suggests options I didn't even know about, so I get to learn a new way of doing something. It became such a common use case for me that I built a cli tool that does exactly that, but right in the terminal. It's called <a target="_blank" href="https://github.com/alexkrkn/help-cli">help-cli</a> and it's open source and uses your own OpenAI api key.

I built it using nodejs, react and <a target="_blank" href="https://github.com/vadimdemedes/ink">INK</a>. You can install it by running:

```$ npm i -g help-cli```

After you run it for the first time, it'll ask you for your OpenAI api key that you can get from here: <a href="https://platform.openai.com/account/api-keys" target="_blank">platform.openai.com/account/api-keys</a>.

In the video above I show an example how I use it to crop a video, using FFmpeg. I also go over the code, how it makes calls to the OpenAI api and which prompts it uses.

I plan to make another video in which I'll go over how <a target="_blank" href="https://github.com/vadimdemedes/ink">INK</a> works. I also want to modify help-cli to start using the new function calling option of OpenAI.

To get notified about all the new posts, videos and projects, join my newsletter: [newsletter.undefinedapps.com](https://newsletter.undefinedapps.com/)

And don't forget to subscribe on youtube: [youtube.com/@BiteSizeAcademy](https://youtube.com/@BiteSizeAcademy)

A nodejs cli tool that uses ChatGPT to help you with cli commands

Use ChatGPT to come up with FFmpeg CLI commands


In this video we build a nodejs lambda function while running and testing it locally.

<p>
  <iframe width="560" height="315" src="https://www.youtube.com/embed/51EAwBDdgio" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
</p>

The lambda function that gets arguments from query params, uses axios to download the html of a website, uses cherio to parse out the title of the page and then saves the html file to S3 and returns a json response back to the user.

We write local tests for the function using mocha and sinon (to create stubs).

The video also covers:
- creating and deploying lambda functions to AWS
- setting up an S3 bucket
- running the lambda function from a public url
- managing lambda releases
- setting up github actions to deploy to AWS on 'git push'

### ⚡️ Source Code

- <a target="_blank" href="https://github.com/alexkrkn/lambda-url-to-html">Github</a>

Write local tests for the function using mocha and sinon

Test and develop AWS lambda functions locally


In this video we'll see how to build a video editing crop tool using reactjs and react-spring + use-gesture.

<p>
  <iframe width="560" height="315" src="https://www.youtube.com/embed/vDxZLN6FVqY" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
</p>

First we build the UI using Mui, a React UI Library. Then, we add a react component which overlays on top of our video player element and lets the user make a selection by dragging and resizing a selection area.

When the user confirms the selection, we make an API request to crop the video and return the result to the user.

### ⚡️ Links:
- <a target="_blank" href="https://github.com/alexkrkn/react-crop-video">Source code on github</a>
- <a target="_blank" href="https://videoapikit.com">VideoApiKit</a>
- <a target="_blank" href="https://mui.com/">Mui</a>
- <a target="_blank" href="https://www.react-spring.dev/">react-spring</a>
- <a target="_blank" href="https://use-gesture.netlify.app/">use-gesture</a>

Build a video editing crop tool using reactjs and react-spring + use-gesture

Drag and resize in react using react-spring and use-gesture


In this video we create a lambda function (a random number generator) and use versions and aliases to manage the deployment of the lambda function.

### ⚡️ Summary

- The $LATEST version points to the latest version of the code of the function. For example, when you edit the code or upload a zip file with your code, the $LATEST version will immediately point to that version of the code.
- A numeric version of a function points to a snapshot of the function's code at the time when the version was released. This version of the code cannot be changed.
- An alias points to one of the versions of the function
- To manage the release and deployment of a lambda function, we can point an API Gateway endpoint to an alias and then when we release a new version of the function, adjust the alias to point to this new version.

### ⚡️ Links:
- <a target="_blank" href="https://lambdabuddy.com">LambdaBuddy</a>

Create a lambda function and use versions and aliases to manage its deployment

Deploy a lambda function using versioning and aliases + API Gateway


In this video we build a real-time chat app (like slack) that uses websockets backed by an AWS lambda function and Amazon API Gateway. The app sends JSON over the websocket to deliver the chat messages and notifications in real time, over a two-way communication line, to all the users of the chat.

<p>
  <iframe width="560" height="315" src="https://www.youtube.com/embed/BcWD-M2PJ-8" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
</p>

The client is written in react (with hooks) and uses the WebSocket api that is available in all the modern browsers. The backend is serverless and uses API Gateway + AWS Lambda with javascript/nodejs.

### ⚡️ Source Code

- <a target="_blank" href="https://github.com/alexkrkn/lambda-websocket-server">server</a>
- <a target="_blank" href="https://github.com/alexkrkn/lambda-websocket-client">client</a>

### ⚡️ Links:
- <a target="_blank" href="https://developer.mozilla.org/en-US/docs/Web/API/WebSockets_API">The WebSocket API (WebSockets)</a>
- <a target="_blank" href="https://github.com/websockets/wscat">wscat</a>


In this video we build a real-time chat app (like slack) that uses websockets backed by an AWS lambda function and Amazon API Gateway.

Build a chat using Lambda + WebSocket + API Gateway


<p>
  <iframe width="560" height="315" src="https://www.youtube.com/embed/FFUUhYhxh5Q" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
</p>

In this video we look at how to drag and drop in react using the react-use-gesture and react-spring libraries. We take the introduction screen of create-react-app and add dragging to its components. We then add an overlay which can be dragged up and down to expose a background image behind it. We use the hooks: useDrag (from react-use-gesture) and useSpring (from react-spring).

Also check out my previous post about react-spring <a href="/post/animate-mount-unmout-react-spring">How to animate mount and unmount of a react component using react-spring</a>

### ⚡️ Links:
- <a target="_blank" href="https://github.com/alexkrkn/react-drag-and-drop/">source code</a>
- <a target="_blank" href="https://alexkrkn.github.io/react-drag-and-drop/">live demo</a>
- <a target="_blank" href="https://www.react-spring.dev/">react-spring</a>
- <a target="_blank" href="https://use-gesture.netlify.app/">use-gesture</a>

Drag and drop in react using the react-use-gesture and react-spring libraries

Drag and drop a div in react using hooks


<p>
  <iframe width="560" height="315" src="https://www.youtube.com/embed/gc3w_bMtcQE" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
</p>

In this video we add basic auth to an s3 bucket containing a static website.

This way, the browser will display an authentication dialog and require a username and password to get access to the static site. This will protect the files on our s3 bucket and limit access to them.

We use a cloudfront distribution and a lambda edge function that checks the request being made and authenticates the user.


Display an authentication dialog to require a username and password to get access to a static site

Restrict access to a static S3 site using HTTP Basic Auth


<p>
  <iframe width="560" height="315" src="https://www.youtube.com/embed/V3i25clEvSE" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
</p>

In this video we integrate a lambda function with an API Gateway HTTP endpoint and define a proxy route to direct all incoming requests to our function. We then extract the path and the url query string parameters from the event object which is provided as an argument to the lambda function.

Extract the path and the url query string parameters from the event object which is provided as an argument to the lambda function

Pass a url query string or a route parameter to lambda from API Gateway


<p>
  <iframe width="560" height="315" src="https://www.youtube.com/embed/kT6yYSwK1oA" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
</p>

In this video we add animations to the mount and unmount events of a react component using the react-spring animation library and hooks, in a create-react-app project.

### ⚡️ Links:
- <a target="_blank" href="https://www.react-spring.dev/">react-spring</a>

Add animations to the mount and unmount events of a react component using the react-spring animation library and hooks.

Animate mount and unmount of a react component, using react-spring


<p>
  <iframe width="560" height="315" src="https://www.youtube.com/embed/UQiRhKgQ5X0" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
</p>

In this video we deploy a lambda function using Github actions

### ⚡️ Source Code

- <a target="_blank" href="https://github.com/BiteSizeAcademy/github-action-lambda">Github</a>

Deploy a lambda function using GitHub actions

Github Actions + AWS Lambda


<p>
  <iframe width="560" height="315" src="https://www.youtube.com/embed/UGwY2iT4F80" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
</p>

In this video we create a time limited signed url for an image store on S3, using nodejs.

Using S3 signed urls, we can limit access to S3 files only to users who has access to these urls. We can also limit the amount of time in which these users can access the files.

We also see learn how to generate signed urls for uploading images as well.

Generate a time limited presigned S3 URL using the nodejs AWS SDK

S3 presigned URLs with nodejs


<p>
  <iframe width="560" height="315" src="https://www.youtube.com/embed/jOLgUjcTFEI" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
</p>

In this video we connect from a lambda function to an RDS database, first directly and then by using RDS proxy.


Connect from a lambda function to an RDS database, first directly and then by using RDS proxy

Using lambda with a MySQL database


<p>
  <iframe width="560" height="315" src="https://www.youtube.com/embed/J9QKS0NrH7I" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
</p>

In this video we take a look at how secrets, private keys and passwords should be stored in a lambda function. First, we extract hardcoded secrets into environment variables, then we learn how to encrypt these variables using a key from KMS (Key Management Service) and how to decrypt them in the lambda function.

### ⚡️ Source Code

- <a target="_blank" href="https://github.com/BiteSizeAcademy/lambda-secrets">Github</a>

How secrets, private keys and passwords should be stored in a lambda function

Use environment variables with a Lambda function (+KMS)


ImageMagick is a popular open source utility to transform images. It used to be installed on the OS which lambda functions use, but was removed in the latest version.

<p>
  <iframe width="560" height="315" src="https://www.youtube.com/embed/JdJo-_Y1ZIM" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
</p>

In this video we look at how to use ImageMagick in a lambda function using lambda layers. We look at an open source project which creates a layer for us that has the ImageMagick binaries, then we attach this layer to our function and use ImageMagick to resize an image. Finally, we save the resized image to an S3 bucket.

### ⚡️ Links:
- <a target="_blank" href="https://imagemagick.org">ImageMagick</a>
- <a target="_blank" href="https://github.com/serverlesspub/imagemagick-aws-lambda-2#usage">The repository for the layer which we use to install ImageMagick</a>
- <a target="_blank" href="https://github.com/BiteSizeAcademy/lambda-imagemagick-demo">Demo repository for the function we deploy in this video</a>

Create a lambda layer which contains ImageMagick so that you can use it in your lambda function

Use ImageMagick with a Lambda function on AWS


In this video we take a look at how to install, add and use node library dependencies that we install using npm (node package manager)

<p>
  <iframe width="560" height="315" src="https://www.youtube.com/embed/RnFowJ130pc" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
</p>

Install and use node library dependencies on Lambda (using npm)

Install npm modules in AWS Lambda


In this video we look at how to use layers with AWS Lambda functions. (with nodejs)

We look at a demo app which uses two npm modules (axios and image-size) which were installed using npm and are part of the bundle. We then extract these modules into a layer. We attach this layer to our function and import and use the modules we extracted.

<p>
  <iframe width="560" height="315" src="https://www.youtube.com/embed/i12H4cUFudU" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
</p>

It has 2 advantages:
- it reduces the size of the bundle
- it makes it possible to re-use the extracted code across several functions.

We also look at how to separate local files and extract them into a layer and then import them from the function.

If you would like to see an example of using layers, check out [this video](https://youtu.be/JdJo-_Y1ZIM) in which we use a layer to install ImageMagick to resize images within a lambda function.

A demo app which uses two npm modules extracted to modules