On May 24th, Desmond Kwok—Head of Marketing at AfterSchool and a graduate of the PolyU Diploma in Digital and Social Media Marketing—returned to his alma mater to deliver an insightful guest lecture to students of the newly launched Professional Certificate in AI-Driven Digital and Social Media Marketing at PolyU SPEED. In his session, Desmond shared how his marketing team leverages Generative AI to solve real operational challenges and drive results at one of Hong Kong’s most successful online tutoring platforms.
The Rise of Generative AI in Marketing
In today’s rapidly evolving digital landscape, Generative AI is reshaping the way marketers operate. It offers tools that go far beyond content creation — enabling better decision-making, higher efficiency, and more scalable strategies.
Desmond opened the session by contextualizing AI’s practical value in marketing. At AfterSchool, his team doesn’t use AI for novelty, but as a deeply integrated part of their day-to-day workflow. The company’s AI adoption journey offers a grounded case study for anyone seeking to apply the technology to real marketing functions.
About AfterSchool
AfterSchool was established in 2018 and has since grown into the largest online tutoring platform in Hong Kong. The platform currently serves:
- Over 100,000 student members — roughly one in every two DSE candidates
- An annual revenue of HK$50 million
- A 20-person core team driving both education and marketing innovation
AfterSchool’s marketing team is responsible for three key domains:
- Social Media Engagement – managing accounts for the platform and its tutors on Instagram and Facebook
- Paid Ad Campaigns – executing performance-driven campaigns through Google and Meta
- Data Analytics – continuously refining strategies using traffic, sales, and conversion data
Daily Marketing Challenges
Despite a strong in-house team, Desmond pointed out that many challenges remain familiar to most marketers:
- Overwhelming ad reporting tasks – Campaign results come in daily, but making sense of large volumes of data consumes valuable time.
- Intensive ad copywriting requirements – Each campaign may need dozens of ad variants for testing and segmentation.
- Pressure to maintain social media output – Instagram’s algorithm demands consistent and engaging content, creating a production burden.
Three Generative AI Tools That Made a Difference
Tool 1: Digital Ads Report Insights Generator
This tool simplifies the process of reviewing ad performance by:
- Reading ad reports exported from Google or Facebook
- Automatically identifying patterns and calculating metrics like ROAS, CPA, and conversion rate
- Summarizing the results into 3–5 insights, each with a clear suggested action
The tool empowers marketers to focus on decisions rather than data extraction, allowing them to respond to trends quickly and effectively.
Tool 2: Google & Facebook Ad Copy Generator
To streamline ad production, this tool:
- Accepts a product URL or written description
- Analyzes the content and target audience
- Prompts the user to confirm selling angles and keywords
- Then automatically generates multiple ad variations tailored for search or social formats
This has been particularly helpful in running A/B tests across campaigns, and is now widely used by the AfterSchool team to support their tutors’ personal promotion pages.
Tool 3: Instagram Carousel Post Generator
This tool helps the content team transform long-form articles into visual, mobile-friendly content by:
- Accepting raw content or blog URLs
- Splitting the narrative into slide-based content with headlines and concise copy
- Following a preset structure that aligns with Instagram carousel best practices
The team now uses it regularly to publish educational or thought leadership content with consistent tone and style, while significantly reducing time spent on drafting.
From Prompt to GPT: A Systemized Approach
Desmond emphasized that these tools didn’t start as polished apps. Instead, they evolved from a simple practice: refining ChatGPT prompts until they became reusable.
He walked through a step-by-step process:
- Prototype tasks using ChatGPT manually
- Consolidate repeated actions into a prompt structure with clear instructions
- Turn that into a custom GPT with role definitions, formatting rules, and even reference examples
- Share it across the team or deploy on platforms like POE for broader use
By formalizing these prompts, the team ensured consistency, reduced training time, and built scalable internal AI capabilities.
Practical Advice: Start Small, But Build with Purpose
To help marketers take the first step, Desmond shared a simple but effective framework for building AI into their work habits.
- Spot the friction – Identify which part of your daily work is repetitive or time-consuming
- Prototype a solution – Use tools like ChatGPT or GPTs to quickly test how AI could help
- Optimize over time – Don’t expect perfection at first — refine and iterate
Most importantly, Desmond reminded students that AI is meant to empower people, not replace them.
Extending the Learning: From Tools to Ownership
Beyond showcasing tools, Desmond generously provided students with the full instructional prompts behind his three GPT-based systems. These included detailed guidance such as user input expectations, output structure, formatting requirements, and real examples — offering attendees a ready-made foundation for their own marketing automation tasks.
He further demonstrated how to transition from OpenAI’s GPTs interface to the more student-accessible Poe platform, ensuring everyone could implement the tools regardless of technical background.
This hands-on component emphasized that having access to tools is only the beginning — true value comes from creating and iterating your own prompts.
Desmond urged students to:
- Treat prompts as living systems — edit, test, and evolve them
- Adapt templates to their own work challenges
- View GPTs not as one-off helpers but as trainable team members
His message was clear: you don't just use AI — you design it to work for you.
Prompt Sample: Digital Ads Report Insights Generator
Desmond shared a complete prompt structure used to build the Digital Ads Report Insights Generator GPT. The prompt includes instructions such as:
- Ask the user to upload a Google or Facebook Ads report
- Break down the report into sections with clear summaries
- Generate 3–5 insights per section
- Use a fixed format for output:
- Section Name
- Insight Title
- Insight
- Recommended Action
Example:
“Campaign A shows the highest conversion rate at 2.47% and a total conversion value of HK$147,740. Consider increasing budget allocation.”
Students were also shown full templates for Google Ads copywriting and Instagram carousel generation, each structured to match platform requirements and user intent.
👉 A full version of these prompt templates will be published in a companion article. Stay tuned for the link.
Business Impact Summary
The adoption of these three AI tools has not only optimized day-to-day tasks, but also redefined how the AfterSchool marketing team approaches strategy, speed, and scale. Each tool brought tangible improvements:
- Faster campaign turnaround – Repetitive analysis and copywriting tasks now take hours instead of days, freeing up time for strategic thinking
- Higher ad performance – Data-informed insights and precise ad targeting improved ROI across multiple campaigns
- Scalable content creation – Content teams can now maintain high-frequency posting schedules without creative burnout
More importantly, these outcomes illustrate a deeper shift — from isolated marketing efforts to an AI-assisted, feedback-driven system. This evolution sets the stage for a new kind of marketer: one who doesn’t just execute campaigns, but designs intelligent workflows around them.
Final Reflection from Dr. Ken FONG
"What Desmond shared goes far beyond tools. He has shown us how AI can become a structured habit, a creative partner, and even a virtual teammate — if we design it thoughtfully. In the age of automation, your competitive edge won’t just come from using GPTs, but from learning how to shape and evolve them. That is the real art of intelligence design."
— Dr. Ken FONG
中文摘要
2025年5月24日，AfterSchool 市場總監 Desmond Kwok 回到母校香港理工大學，為香港理工大學專業進修學院「AI 驅動的數碼及社交媒體營銷專業證書課程」的學生帶來一場內容豐富且具實戰價值的專題講座。 他本身是 PolyU 數碼營銷文憑課程首屆畢業生，現時領導香港最大網上補習平台的市場發展。
他分享了 AfterSchool 如何應用三款生成式 AI 工具：
- 數據報表洞察工具：自動分析廣告成效並提供具體策略建議
- 廣告文案生成工具：快速產出 Google 和 Facebook 廣告文案，支援 A/B 測試
- IG Carousel 貼文生成器：將長文轉化為社交平台視覺內容
Desmond 並公開分享三個工具的 GPT prompt 寫法，並現場指導學生如何將這些 GPTs 移植到 Poe 平台操作。 他指出，懂得設計與修改 prompt，才是真正能讓 AI 成為你工作夥伴的能力。
Section：Campaign A
Insight Title：高轉換與低 CPA
Insight：轉換率 2.47%，總值 HK$147,740
Action：建議增加投放預算
他鼓勵學生主動建立自己的 AI 工具，並持續優化應用方式。他強調：「你不是在用 AI，而是在設計它為你工作。」
📌 完整 prompt 將於延伸教學文章中發布，敬請留意。
Keywords
Generative AI, GPTs, ChatGPT prompts, marketing automation, AfterSchool, ad copywriting, performance analytics, Instagram content generation, AI-powered tools, PolyU AI marketing, Poe platform, AI prompt engineering, social media marketing, digital education, A/B testing automation, custom GPT tools, campaign reporting, carousel post automation, AI literacy, content workflow automation
Posts
0 Comments