AI in Marketing Strategy

What AI Should Do, and What People Shoul

“Should we turn this ad off, or add more budget?”

This is a familiar question for anyone working in game publishing. When agencies or UA teams deliver campaign reports, they often read like this:

"CPI has gone down, but ROAS also dropped."

“New user acquisition increased, but churn is high.”

And the usual response is:

“Let’s test a bit longer before deciding.”

But the first one to two weeks after launch are the “golden time.” Often, there is no time for extended testing. Decisions about whether an ad campaign survives or dies must sometimes be made within hours. At this point, what truly helps practitioners make those quick decisions is AI-based marketing automation tools.

ℹ️ How Does AI Judge and Act?

Modern automation tools go beyond reporting. They read data, analyze it, and improve performance on their own. At the core are two AI technologies: Deep Learning (DL) and Reinforcement Learning (RL).

■ Deep Learning (DL): Finds ‘performance drivers’

AI analyzes thousands of variables — creative elements, traffic timing, country, device, user behavior — to automatically identify the factors influencing outcomes. For example:

Which background tone in an image increased CTR

Which copy delivered higher conversion in each market

Whether users acquired between 12–3 p.m. retained better than others

In this way, AI uncovers correlations among variables that human practitioners might easily miss.

■ Reinforcement Learning (RL): Trains strategy through experimentation

While deep learning is based on past data, reinforcement learning continuously experiments in real time to train and refine strategies. AI repeatedly tests countless combinations of targets and creatives, adjusting strategy based on outcomes. This is why campaigns can keep improving performance even as they run.

ℹ️ The Core of Automation Is “Connected Data”

For AI to function properly, accurate and well-structured marketing data is essential. Game business practitioners often rely on the following tools:

Appsflyer/Adjust: Collect acquisition paths and payment history (e.g., “A user from Google Ads did not make a purchase within three days”)

Amplitude: Analyze drop-off points (e.g., “Users dropped out right after the tutorial”)

Braze: Automatically send personalized messages based on conditions (e.g., “Send a beginner’s guide message to churned users”)

When these data sources connect, an automation loop is completed — from ads → behavior analysis → messaging response.

ℹ️ AI Marketing Automation: Redesigning Cost Efficiency

Explaining AI-based marketing automation as merely “reducing operational workload” is not enough. Its real significance lies in redesigning cost efficiency — “achieving better results with the same budget.”

In the past, having a small ad budget meant fighting at a disadvantage. Now, with AI’s precise adjustments and real-time optimization, you can acquire more high-value users, reduce churn, and free marketers from repetitive work.

Ways AI-based operations enhance cost efficiency:

(1) Automatic Budget Optimization
AI analyzes live campaign data, concentrating spend on high-performing channels and targets. Underperforming ads are paused or reduced automatically, cutting waste. No more guesswork in “where to spend the budget.”

(2) Automated Creative Generation and Testing.

Unsure which tone or message works best? AI generates multiple creative combinations, tests them in real time, and scales only top performers — reducing test costs and repeated production burden.

(3) Resource Reallocation
AI takes over repetitive tasks — targeting, retargeting, analysis, and messaging — allowing marketers to focus on planning and strategy.

(4) Hyper-Personalized Targeting
Instead of demographic targeting like “women in their 20s,” AI builds behavioral segments in real time (e.g., “users who logged in 3 times in the past 7 days and completed the tutorial”). This increases efficiency while lowering user acquisition costs.

(5) Performance-Based Targeting (nCPI)
Going beyond CPI installs, AI predicts behaviors such as spending, long-term play, or inviting friends, then focuses on high-LTV users.

(6) Real-Time Retargeting to Prevent Churn
AI immediately detects churn signals and responds — sending a personalized message to an at-risk user or a last-minute offer to a hesitant spender. The result is higher return and conversion rates

ℹ️ Cost Efficiency Re-designed by AI

AI marketing automation is not just about operational efficiency; it’s an innovation that redesigns marketing performance itself. By freeing practitioners from repetitive tasks, it allows them to focus on their real work: strategy and judgment.

ℹ️ Strategy Still Needs Human Judgment

The more AI takes over execution, the more critical strategic judgment becomes in deciding what to automate.

Which KPIs should guide automation?

In which situations is manual intervention required?

At what point should automation logic be updated?

AI is a tool for the mind, not the hands. The direction is still set by business professionals.

ℹ️ After Acquisition Comes Retention

Running ads well is not the end. The greater challenge is ensuring users do not leave — preventing churn and driving them back. Naturally, this flow extends into CRM automation.

ℹ️ "Why Did Users Churn Despite Rewards?”

“DAU is down another 3% since yesterday… should we run a winback promotion?”

For many game operators, these are familiar thoughts that come up every morning while reviewing DAU reports. The real issue is that churn is usually detected only after it has already happened. By then, the users are gone, and the business team is left scrambling to respond.

Now, however, the flow is shifting with the adoption of AI-based CRM systems.

Take Nexon as an example. With many long-running live titles, Nexon faces an extremely diverse set of user segments and targets. In the past, it was difficult to design, and time return campaigns manually for each segment. To solve this, they introduced AI-driven CRM automation. The process began with simple steps:

Automatic detection of churn signals such as declining spend or longer login intervals

Automatic generation of tailored messages by user tier (non-paying, light-spending, VIP)

Automatic dispatch of targeted return promotions, such as “Come back now to receive a limited character”

AI automatically determined both the timing and the content. The outcome was striking return rates more than doubled. Practitioners also noted, “We no longer need to craft messages manually.” (Source: Nexon Game Operation Solution GameScale Introduction, 2023)

This demonstrates that CRM has now evolved beyond being a simple messaging tool. It has become a system where AI can anticipate:

Who is likely to churn soon

Which message will trigger a response

Which reward will be most effective

ℹ️ How Does AI CRM Work?

AI-driven CRM automation generally operates in three stages:

① Prediction – e.g., “This user has an 87% chance of leaving within three days”

② Automation – Automatically create and deliver personalized messages based on spend history, preferences, and player level

③ Experiment and Optimization – Continuously analyze response rates by time and channel in real time, and automatically scale the approaches that perform best

Previously, people had to manually select targets, draft messages, and conduct tests. Today, AI repeats this entire cycle automatically, consistently improving performance.

ℹ️ AI CRM for Small Studios

AI-based CRM systems are no longer the exclusive domain of major publishers. Today, SaaS-based tools make advanced automation readily accessible to small studios. Solutions such as Meta Advantage+, Mailchimp, and Google Smart Bidding enable sophisticated automation at a relatively low upfront cost.

As a result, even a single operator can now respond swiftly to DAU declines and manage automated reward programs.

AI-driven CRM is no longer optional — it has become the foundation of competitiveness. Regardless of size, every game company can now leverage AI to manage users more intelligently and efficiently.

♻️ Final Thoughts

AI marketing automation is not merely about improving efficiency. It is a tool for understanding users, predicting their behavior, and responding with strategic precision.

AI can execute but determining “what to automate” and by “what criteria” remains the responsibility of business professionals

In this sense, AI is not a tool for the hands, but for the mind

Now that the groundwork for the launch campaign is complete, the next step is to move forward with launch readiness — this time, together with AI.