Automating ASO: Building a No-Code Keyword Research System with AI

App Store Optimization (ASO) is essential for increasing an app’s visibility. One of the key elements of ASO is selecting the right keywords, but this process can be time-consuming and complex. To solve this, we developed an automated system using no-code tools to streamline keyword research with AI. This article outlines how traditional ASO tasks can be enhanced through automation, creating a foundation for more sophisticated optimization approaches.

 

Process Overview:

There are three major steps to go from one app name to a shortlist of the most relevant and promising <100 keywords, leveraging keyword research with AI to focus on the right opportunities.

1. Research – Collect keyword data from multiple sources.
2. Evaluation – Consolidate and analyze keywords.
3. Refinement – Fine-tune scoring and prioritization.

The tools we use include data from Google Search Console, AppTweak, and Apple Search Ads, with automation powered by Make.com (a no-code automation tool) and Google Apps Scripts (a cloud-based JavaScript platform that integrates with Google services). The entire process takes less than an hour, with most time spent on API calls and AI processing.

Step 1: Collecting Keywords

In the first step, we use a Make.com scenario to populate a Google Sheet with keyword suggestions from multiple sources. These sources include search terms from Google Console, suggestions from AppTweak, insights from Apple Search Ads, and AI-driven research. Additionally, we incorporate AppTweak and Apple Search Ads data not only for the focus app but also for its competitors.

This saves time and ensures a comprehensive starting list.

Step 2: Filtering and Shortlisting Keywords with AI

This initial collection generates thousands of keywords, including many duplicates, misspellings (which we consider to some extent), and irrelevant terms. To refine this list, we use Google Apps Scripts to consolidate and systematically narrow it down. After the first consolidation step, we typically end up with 1,000-2,000 keywords, depending on the number of competitors included and their activity on Apple Search Ads.

Next, we leverage keyword research with AI to further refine the list by:

• Removing irrelevant competitor terms
• Filtering out overly generic, low-intent keywords
• Eliminating “brand-heavy” keyword phrases, except when the brand name directly represents the app’s function

We introduced this last step after early iterations resulted in a keyword list dominated by “brand + function” phrases. To better capture user intent, we prioritize keywords that users are likely to search for when looking to download and use the app. This includes:

• Category-related terms
• Generic but relevant product/service keywords
• Keywords related to user problems
• Terms reflecting app features and functionality

To improve AI understanding, we provide a prompt with these criteria along with examples of high-quality keywords across different apps.

This process typically reduces the list to fewer than 200 keywords or phrases. At this stage, we pull keyword metrics from AppTweak and prompt AI to assess each keyword’s relevance based on four key dimensions:

1. Name alignment – How well the keyword matches the app’s brand
2. Category alignment – Relevance to the app’s primary category and functions
3. Value proposition alignment – How closely it relates to the app’s core features and benefits
4. User motivation alignment – How well it addresses user needs and intent

These dimensions collectively form a comprehensive relevance score, helping to filter out irrelevant terms and prioritize those most likely to attract high-quality users.

Final keyword prioritization is based on multiple factors, including:

• Relevance score (weighted most heavily)
• Search volume
• Keyword difficulty
• Current ranking
• Historical install data

Each factor is normalized and weighted to generate a final priority score, which classifies keywords as high, medium, or low priority.

Step 3: Refining Keyword Scores with AI

Adaptive Scoring System

The difficulty, search volume, and other keyword metrics vary significantly between apps and markets. A fixed scoring system failed to produce consistently good results. Similarly, arbitrarily designating the top 10 keywords as high priority did not yield the desired outcomes.

Our solution is to allow AI the flexibility to refine the scoring system dynamically. The AI can adjust priority bands and reweight factors (e.g., increasing the influence of search volume while decreasing the weight of difficulty) based on the specific context. The prompt instructs the AI to analyze patterns in the initial scoring, identify potential issues, and make adjustments to improve accuracy.

While we’ve implemented logging to track these changes, the AI’s decision-making process remains somewhat opaque due to its complexity. Nonetheless, this adaptability has significantly enhanced the final results.

Critical Success Factors

Through iterations, we’ve identified key factors that separate mediocre from high-quality results:

1. Ensuring AI understands the brand
   User motivations, though only briefly mentioned earlier, play a crucial role in assessing keyword relevance. Properly capturing these motivations significantly impacts scoring and prioritization.
   
2. Granting AI flexibility in scoring adjustments
   

 

• Dynamic adjustments to weights and priority bands are essential to accommodate varying metrics across different apps and markets.
• This adaptability allows the process to evolve alongside advancements in AI, improving over time.
  

The AI-Driven ASO System

AI plays three distinct roles in this process:

1. Initial Keyword Research – Identifying relevant keywords based on user intent and app functionality.
2. Relevance Scoring – Evaluating keywords using our four key dimensions.
3. Score Refinement & Prioritization – Adjusting scores based on market-specific keyword metrics to ensure practical prioritization.

Each stage builds upon the previous one: relevance scoring integrates insights from the research phase, while final prioritization combines relevance scores with real-world data. This structured approach maintains consistency while ensuring that each step adds meaningful value.

As we refine our prompts and leverage newer AI models, we continue to see improvements in accuracy and efficiency. While the system is not yet perfect, it is already transforming our approach to ASO. We are confident that future iterations will further enhance precision and scalability.

Balancing Automation & Manual Oversight

Although this MVP includes manual verification steps, this is intentional. These steps allow us to review and validate each stage of the process, ensuring accuracy and reliability. Once we achieve a consistently high-quality output, full automation will be a straightforward next step.

Next, we move on to keyword placement and analysis, where we strategically integrate the prioritized keywords and evaluate their impact.

 

Before You Go

• Curious to learn more about how we craft great user journeys for our clients at Phiture? Discover how our PressPlay platform leverages AI to automate A/B testing and enhance App Store Optimization in our PressPlay overview.
• Explore how AI is integrated into ASO strategies by reading about our ChatGPT Plugin for ASO, an experiment from our AI Labs, in this blog post.
• Learn about the ASO Stack, Phiture’s innovative framework that holistically considers all factors impacting app visibility and ranking in the app stores, in our ASO Stack overview.