Ah, the good ol’ days of 2020-2021. Back when ChatGPT and Bard were just whispers in the tech ether, and building an AI tool felt like pioneering the Wild West.
Forget Google VertexAI and fancy templates – this was pure, unadulterated curiosity and hustle.
Innovation on a Shoestring Budget (and Our Free Time)
Charis Maragkakis (go give him a Follow!), my former colleague (now friend!), and I were itching to do something “wow.” Something that didn’t exist yet, something that would leave us saying, “We built that!”
We craved a customer-centric solution that could impact a key metric: Net Promoter Score (NPS). We had zero prior AI knowledge, limited resources, and zero pressure from our roles – just pure passion and a healthy dose of “let’s see what happens.”
From Reviews to Results: The Power of Text Analysis
We dove headfirst into a massive dataset of customer reviews from E-Satisfaction (shoutout to those guys!).
We built a custom algorithm, like training a puppy to sniff out good vibes (positives) and frustrations (negatives) in the reviews.
It wasn’t perfect, but with some optimization, it learned to distinguish between a “you’re awesome, keep it up!” and a “my order is late, you’re the worst!” review.
Here’s the magic: We created a “keyword cloud” to visualize the most frequently mentioned words in each sentiment category.
This helped us spot patterns. For example, if “late” and a specific “courier name” were frequently paired in negative reviews, it became clear which delivery company needed improvement. 80% of negative reviews pointed to just 4-5 key issues!
Talk about a Pareto Principle epiphany!
Validation and Impact: From Hobby to Case Study
We validated our findings with mystery shopping and other methods. Low and behold, the algorithm was spot-on!
Addressing these identified issues led to a dramatic decrease in negative reviews and a significant NPS boost.
This wasn’t just a fun side project; it became an internal case study for a major global e-commerce player (you know who you are ;)). It was a testament to what can be achieved with a little curiosity and a lot of hustle.
The Takeaways: Don’t Wait for the Future, Build It
The story doesn’t end there. Here’s what I want you to remember:
- Don’t wait for the perfect tools or the “right” time. Take action! Experimentation and innovation thrive in the face of “I don’t know, let’s try it!”
- Focus on the problem, not the technology. We used AI because it fit our needs. The key was finding a solution, not just using a fancy tool.
- Passion and perseverance are powerhouses. We had limited knowledge, but our drive to create something valuable fueled our success.
The Future of AI (and My Humble Opinion)
AI has been around for a while, but its accessibility is sparking a new wave of innovation. While the hype is real, I haven’t seen many concrete applications that deliver a clear return on investment (ROI) or revolutionize decision-making. The jury’s still out on some areas, like chatbots, in my honest opinion.
Looking Ahead: Where to Start with AI Today
For those starting their AI journey, here are some of the biggest e-commerce pain points I see (based on my experience and industry colleagues):
- Speeding Up Content Creation
- Bridging the Data Analysis-to-Decision Gap
- Leveraging AI to automate things and processes
The Final Word (and a Few Shameless Plugs):
Innovation isn’t just for Silicon Valley giants with bottomless budgets. It starts with curiosity and a willingness to experiment. So, what are you waiting for? Let’s build something cool together!
P.S. Huge thanks again to E-Satisfaction, Madlen Balatlian, Evangelos Kotsonis, and the International Ecommerce Director who championed this project globally.
P.P.S. Want to learn more about AI? Tune in to the next episode of Let’s Get Digital (it’s just me talking to myself, another innovation apparently!).
P.P.P.S. Check out this article for more insights: https://www.grow-digital.gr/