Boost Your App's Visibility with ChatGPT: Utilizing AI for ASO — Guide

App Store Optimization
Oct 17, 2023

ChatGPT, an AI chatbot, assists app developers with App Store Optimization (ASO) tasks such as metadata generation, app description writing, proofreading, and translation. AI crafts metadata, app descriptions, and ASO strategies, and handles marketing tasks like translation, competitor data analysis, and keyword list generation. It can also enhance app graphics using neural networks like Midjourney and DALL-E.

Join the open ASO & User Acquisition community on Discord - ASO Busters! Here, engage in insider discussions, share insights, and collaborate with ASO and UA experts. Our channels cover the App Store, Google Play, visual ASO, ASA, UAC, Facebook, and TikTok.

In this article, we've crafted a step-by-step guide to provide a visual demonstration of its functionality in the ASO field. We'll delve into two distinct approaches for the App Store and Google Play, examining them separately.

ChatGPT in ASO: App Store

Artificial intelligence and App Store optimization can produce superior results. As an example, we will use an app in the Fitness category.

Stage 1: Analytics (preliminary research, gathering competitor information, and analyzing weaknesses in the App Store)

This step is important and can improve your ASO results. Let's start by asking ChatGPT to imagine itself as an expert in ASO, using the cheat code "Let's imagine". It is worth adding that for an effective response. We also need to specify the exact name of the app and its corresponding Store ID or link to the App Store.

A few seconds later, we receive a brief analysis from ChatGPT, which provides basic information without specific details. This information is likely to be the most useful for those who are newcomers to the field of ASO.

The next crucial step is to gather information about competitors for further analysis and to identify our application's comparative advantages. To achieve this, we will provide AI with general information about our application, so it can prepare a list of competitors.

We've identified 10 competitors with ChatGPT's help. Generally, all apps that ChatGPT specified are considered competitors, occasionally including certain other ones that the algorithm deems relevant. This occurs because the neural network selects apps based on similar features, as explained by ChatGPT. In practice, this aids in initial competitor research and basic analysis or expanding your current list.

For a more precise answer, we have provided the competitors it suggested as examples. This avoids redundancy and helps pinpoint the niche more accurately.

The response was quick, and we received a list of 10 additional competitors with an app similar to ours.

With this information, we can revisit ChatGPT and request more detailed analytics for our app. This will provide us with analysis and recommendations for each of the app page components, including the Title, Subtitle, Description, Screenshots, Icon, Ratings, and Localization.

Conclusion. This ASO analysis is semi-professional. Nevertheless, it's crucial to acknowledge that neural network analysis may contain errors, necessitating data verification. For instance, ChatGPT erroneously indicated that the application has a good rating.

*Current rating in the USA

Although such errors are relatively uncommon, they can still occur. Therefore, it's essential to exercise caution and thoroughly review ChatGPT's responses.

Stage 2: Collecting a Keyword List

A neural network typically gathers approximately 20 words as an initial keyword list, but it's important to note that this is not a strict limit. We can further expand it by making a request such as "collect semantic core" or "augment semantic core" to ChatGPT. In response, ChatGPT will initially propose around 20 words and then augment the list with an additional 20 words.

For the next query, we had ChatGPT generate a keyword list consisting of 500 words, which presented some challenges as we initially received a limited number of search terms. After analyzing and comparing the keyword list from four iterations, it became apparent that many keywords were repetitive. However, by consolidating them and removing duplicates, ChatGPT compiled a comprehensive keyword list containing a total of 133 unique keywords.

Conclusion. The neural network lacks crucial metrics for keyword list collection, including:

  • SAP (Search Ads Popularity)
  • Relevance of search results for keywords
  • Conversion rate data (if ASA is active)

Please be cautious when using SAP keywords generated by ChatGPT, as they are random and often inaccurate.

Reminder: ChatGPT 3.5's knowledge is limited to data available until September 2021.

Stage 3: Collecting and Optimizing Metadata

With the keyword list in place, we can instruct ChatGPT to generate metadata for our App Store application. However, to prevent overly long headers, we must manually specify character limits for the Title and Subtitle, such as 30 characters, when setting constraints in our query.

As in previous tasks, we can once again request additional Title and Subtitle variations, resulting in a greater number of choices.

Some of the Titles and Subtitles suggested by ChatGPT may not be suitable due to character limits or the partial use of brand names. There is always the option to edit the suggested choices or use them as inspiration or templates.

However, the situation differs with the Keywords field. When asking ChatGPT to provide Keywords to use in the application, it provides a vast list of keywords, some of which are duplicates from the Title and Subtitle fields. Unfortunately, there is no way to eliminate duplicates, and AI consistently includes at least one duplicated keyword.

ChatGPT itself emphasizes that it gives you suggestions from which you can select the keywords that best suit your needs.

Conclusion: When it comes to metadata collection, AI can be your helpful companion. It cannot operate as a stand-alone tool capable of handling tasks independently. Its strength lies in creating templates and inspiring ideas for your headlines.

Stage 4: Writing application descriptions and user feedback

Another useful functionality at your disposal is working with text-based descriptions of the app and compiling user reviews. This tool is highly versatile, as ChatGPT can create descriptions and responses to feedback tailored to your specific requests and requirements.

Let's begin by addressing the application's description. We have the option to request a general description, or we can specify certain app features for AI to highlight in the description.

Functionally, submitting multiple competitor description variants for analysis can enhance the quality of the output and better convey the app's essence.

In this case, we presented it with five descriptions from different competitors, each with unique app functionalities. ChatGPT extracted the most relevant keywords from these descriptions and integrated them into a unified text.

While app descriptions aren't indexed in the App Store, creating relevant and captivating text content that effectively outlines the app's main features is strongly recommended.
The next step is to create reviews. We can either provide a template or specify app features and keywords for ChatGPT to write the reviews.

Attribute
Value
Format
PNG
Color space
Display P3 (wide-gamut color), sRGB (color), or Gray Gamma 2.2 (grayscale).
See Color Management.
Layers
Flattened with no transparency
Resolution
Shape
Square with no rounded corners
Attribute
Value
Format
PNG
Color space
Display P3 (wide-gamut color), sRGB (color), or Gray Gamma 2.2 (grayscale).
See Color Management.
Layers
Flattened with no transparency
Resolution
Shape
Square with no rounded corners
Word
Frequency in service
Qimai(China)*
AppTweak
Mobileaction
Appfollow
课程 (course)
5387 (high)
56 (mid)
5 (low)
34 (mid)
外语 (foreign language)
5699 (high)
35 (mid)
5 (low)
34(mid)
播放 (play)
5496 (high)
35 (mid)
5 (low)
35 (mid)
*Search index in China provided by Qimai and based on multidimensional calculations such as search results, downloads and relatable keywords.
App Store
Google Play
1. Icon impact on user choice
Important, but not more than a video with screenshots, as they take up most of the screen
Icon is displayed with several other icons in search results, and therefore it has the greatest impact on the user's choice
2. Icon localization
1 icon for all locales
Can be localized for each country
3. Screenshots quantity, orientation
From 1 to 10, any orientation
From 2 to 8, any orientation
4. Screenshots the most important are the ones that are visible before you start scrolling
  • Vertical screenshots: 3 are visible
  • Horizontal screenshots: only the first one is visible
  • Vertical screenshots: 4 are visible
  • Horizontal screenshots: only the first one is visible
5. Screenshots size
Large — the content is legible
Much smaller — the captions are quite illegible
6. Video quantity
Up to 3
One video that must be uploaded to YouTube
7. Video orientation
Any orientation
Horizontal orientation prioritized
8. Video autoplay
Plays automatically for 30 seconds without sound
You need to click on it to play
9. Application cover
Displayed on the application and developer pages
There is no cover
10. Graphics update
After release only
Whenever

AI in ASO: Google Play

As an example, we will use another app in the Fitness category. Implementing ChatGPT in Google Play metadata closely mirrors the process in the App Store. The collection and analysis of competitors and the keyword list are essentially the same.

We only need to instruct AI to generate a list of competitors and a list of keywords using the main app's name and ID.

Stage 1: Working with Metadata

Google Play doesn't feature a dedicated keywords field; instead, it relies on the full description to gather keywords from. Therefore, our aim is to incorporate as many relevant keywords as possible and fit them naturally in the content of the app description.

Let's emphasize the two primary metrics crucial for our work:

  1. Google Cloud Natural Language’s Confidence score evaluation of the description. А factor that assigns a fitting category to it, ensuring Google's algorithm places our app in the appropriate category alongside relevant competitors.
  2. Density keyword coverage.

Tips. To create a high-quality description, collect competitor examples from ChatGPT. Avoid including too many similar texts. The ideal number depends on your app's niche. For similar niches, around 5 texts should suffice; for diverse niches, use 10 or more to cover all variations.

It turns out that the finalized text should achieve high scores in both metrics. For Confidence, the target value is 0.8 or higher, and for Density (keyword frequency), our standard is a value of 2%. We will check the Confidence and Density values using Google Natural Language and WordCounter.

Let's proceed with text generation using ChatGPT. Here are the statistics for the original description:

Stage 2: Editing of the Text

The current text requires revision because it has a low Confidence score. On the other hand, Density has reached the optimal value for many keywords. However, this is influenced by the text's relatively small length, which in our case is only 500 characters.

Our initial step was to request a new description for the app.

The outcome was a significantly larger text that greatly improved all the existing metrics.

Similarly, we add new metrics here:

After the first text iteration, which did not include competitor examples or keyword selection, we prepared another description and increased search visibility.

Our work continues as we focus on enhancing individual metrics. Our first priority is to request ChatGPT to increase the Confidence value of the new text:

Here is the Confidence value of the new variant:

We've achieved remarkable results, but our text still has imperfections, particularly regarding keyword coverage. Fortunately, AI can help by increasing keyword density and frequency.

In our case, we utilize the keywords previously provided by ChatGPT but request an increase in their frequency. We select queries from the keyword list and incorporate them into the text.

Following this, we received a new text. The Confidence score dropped slightly to 0.95, still a high value for our set of criteria. Despite the text's increased length, keyword coverage improved.

Here is the result of the final text analysis:

Conclusion. ChatGPT excels in text generation, consistently delivering clear and predictable results with our 'advanced' requests.

Stage 3: Proofreading and Translation

More significant tasks where ChatGPT can streamline our work:

  1. Localizing already prepared English texts into other languages. While the neural network provides translations of reasonable quality, it's advisable to double-check with a translator.
  2. Proofreading prepared texts for grammatical errors, typos, and inaccuracies.
Device or context
Icon size
iPhone
60x60 pt (180x180 px @3x)
60x60 pt (120x120 px @2x)
iPad Pro
83.5x83.5 pt (167x167 px @2x)
iPad, iPad mini
76x76 pt (152x152 px @2x)
App Store
1024x1024 pt (1024x1024 px @1x)
Source language
Target language
How much longer (+) or shorter (–) the text in the target language is
English
French
21.18%
English
Spanish
19.52%
English
Italian
17.91%
English
Deutsch
16.67%
English
Dutch
13.80%
English
Portuguese (Portugal)
14.29%
English
Portuguese (Brazil)
12.96%
English
Polish
9.33%
English
Russian
9.11%
English
Czech
3.70%
English
Arab
–6.25%
English
Japanese
–39.68%
English
Korean
–44.04%
English
Chinese (Simplified)
–61.97%
English
Chinese (Traditional)
–63.80%
*Search Ads Popularity (SAP)shows the popularity of the search term from 5 to 99.

Tips to Improve ChatGPT's Performance in ASO

However, these tips represent just one part of the equation when it comes to utilizing AI effectively. The ChatGPT user community has uncovered various 'cheat codes' that can enhance its usability. While there are many more such tricks, we've highlighted the key ones:

step-by-step (to help point out the solution to your query)
with references (links) / examples (ChatGPT will add sources and examples to your query)
continue, keep going (continue to generate current query)
let's imagine (help to describe the abstract situation in terms of AI)
summarize it: link (ChatGPT will analyze given information and summarize it)
rephrase, try again (re-generate the current request)
Device
Spotlight icon size
Settings icon size
Notification icon size
iPhone
40x40 pt (120x120 px @3x)
29x29 pt (87x87 px @3x)
38x38 pt (114x114 px @3x)
40x40 pt (80x80 px @2x)
29x29 pt (58x58 px @2x)
38x38 pt (76x76 px @2x)
iPad Pro, iPad, iPad mini
40x40 pt (80x80 px @2x)
29x29 pt (58x58 px @2x)
38x38 pt (76x76 px @2x)
Search term
Translation
SAP* English (U.S.)
SAP* English (U.K.)
truck games
games with trucks
62
39
lorry games
games with trucks
32
jail games
prison games
29
8
prison games
prison games
32
24
*Search Ads Popularity (SAP)shows the popularity of the search term from 5 to 99.
App Store
Google Play
What can be tested?
  • Short description
  • Long description
  • Icon
  • Feature graphics
  • Screenshots
  • Videos
  • Screenshots
  • Videos
  • Icon (has to be uploaded to the build*)
Number of simultaneously running tests
5 tests (each test is valid within a single country.

You can choose a default country test (details below): then it will run in all countries where there are no localized graphical or textual materials.
1 test (the test can be immediately extended to all countries where the application is available or opt for specific countries as needed)
The number of test variants that can be tested with the current version in the store
Compared with a maximum of 3 new variants
Can a test be launched while another item is under review?
Yes
No
Mandatory formats for screenshots uploaded to the store
6.5
  • 6.5
  • 5.5
  • 12.9 (if there is an iPad version)
*Build – is a new version of the application. Updating the icon is only possible when updating the application version in the store. In other words, the term "build" refers to a specific version or variant of the application that is ready to be downloaded and installed on the users' devices. It contains all the necessary files and data for users to install and use the application.
More about optimizing graphic elements in the App Store and Google Play can be found in the article 'Graphics in Mobile App Promotion in the App Store and Google Play (ASO) – How to Optimize Graphic Elements.'

Device size or platform
Screenshot size
Requirement
Screenshot source
6.5 inch (iPhone 13 Pro Max, iPhone 12 Pro Max, iPhone 11 Pro Max, iPhone 11, iPhone XS Max, iPhone XR)
1284 x 2778 pixels (portrait)2778 x 1284 pixels (landscape)1242 x 2688 pixels (portrait)2688 x 1242 pixels (landscape)
Required if app runs on iPhone
Upload 6.5-inch screenshots
5.8 inch (iPhone 13 Pro, iPhone 13, iPhone 13 mini, iPhone 12 Pro, iPhone 12, iPhone 12 mini, iPhone 11 Pro, iPhone XS, iPhone X)
1170 x 2532 pixels (portrait)2532 x 1170 pixels (landscape)1125 x 2436 pixels (portrait)2436 x 1125 pixels (landscape)1080 x 2340 (portrait)2340 x 1080 (landscape)
Required if app runs on iPhone and 6.5 inch screenshots are not provided
Default: scaled 6.5-inch screenshotsAlternative: upload 5.8-inch screenshots
5.5 inch (iPhone 8 Plus, iPhone 7 Plus, iPhone 6s Plus)
1242 x 2208 pixels (portrait)2208 x 1242 pixels (landscape)
Required if app runs on iPhone
Upload 5.5-inch screenshots
5.5 inch (iPhone 8 Plus, iPhone 7 Plus, iPhone 6s Plus)
2048 x 2732 pixels (portrait)2732 x 2048 pixels (landscape)
Required if app runs on iPad
Upload 12.9-inch iPad Pro (3rd generation) screenshots
5.5 inch (iPhone 8 Plus, iPhone 7 Plus, iPhone 6s Plus)
2048 x 2732 pixels (portrait)2732 x 2048 pixels (landscape)
Required if app runs on iPad
Upload 12.9-inch iPad Pro (2nd generation) screenshots
Price
Traffic Source (Self Reporting Networks)
Facebook
Google
X (Twitter)
Apple Search Ads
Cohort reports
Impression tracking
Audience segmentation
At extra charge
Custom Dashboards
Custom Reports
At extra charge
Advertisement Cost
DAU/MAU (Stickiness)
Raw Data Export
At extra charge
At extra charge
At extra charge
API Reporting
Search term
Translation
SAP* French
SAP* French (Canada)
soldes
sale
25
5
aubainerie
sale
14
*Search Ads Popularity (SAP)shows the popularity of the search term from 5 to 99.
Device
Icon size
iPhone
180px × 180px (60pt × 60pt @3x)
120px × 120px (60pt × 60pt @2x)
iPad Pro
167px × 167px (83.5pt × 83.5pt @2x)
iPad, iPad mini
152px × 152px (76pt × 76pt @2x)
App Store
1024px × 1024px (1024pt × 1024pt @1x)
Icon format
PNG
Color models
sRGB or P3 (see "Color Management")
Layouts
Aligned without transparency
Sizes
Shape
Square without rounded corners
All the current requirements for icons are specified in the specification.
Icon format
32-bit PNG (with alpha channel)
Color models
512 х 512 pixels
Layouts
1024 КБ
Sizes
square, without rounding and shadows (Google automatically rounds the corners and adds shadows)
All the current requirements are specified in the specification.
  • AI can significantly aid ASO specialists with routine tasks. However, it's important to note that, at this stage, it cannot perform all tasks single-handedly as some level of analysis is still required.
  • ChatGPT excels in multiple ASO tasks, with its primary emphasis being on template-based keyword collection. We can utilize its generated keywords to enrich our existing keyword list and initiate our initial data analysis.
  • While it aids in generating ideas and metadata options with the necessary keywords for the App Store, comprehensive compilation is best entrusted to professionals.
  • An invaluable assistant for ASO in Google Play. Particularly, when you have limited funds and resources available for expert-level text compilation, localization, and proofreading.
  • ChatGPT is more suitable for text-related tasks and guidance. When dealing with graphical elements, it is advisable to explore the capabilities of specialized image-generating AI models.
Home
Up