A good recommendation system increases user engagement, make your sales grow and beat your competitor. Everyone knows that. However, imagine you being a small shop-owner, offline or online, what you need to do in order to implement the recommendation system that tailored to your shop is discouraging. That's where openR steps in.
What it does
openR is an open platform that let you utilize the power of recommendation without owning any server. All you need to do is tell us what kind of products you have, your customers and transactions. openR will do its best to find a good recommendation for your customers. We focus on three things,
- Openness - we're an open platform. our mission is to make a platform for everyone (with internet)
- Accuracy - our engineers are doing their best to create a system that are customized just for you
- Measurable - we do not just return random result. we'll make sure that our recommendation is well trained and you can test that yourself.
How we built it
Challenges we ran into
Creating a recommendation system is easy. Creating recommendation systems for hundreds of store, however, need a good mix of skills from database design, machine learning skill, to autonomous cloud deployment. To create a platform that scale, we run into challenging tasks like automating the task of modeling and deployment, ensuring that our recommendation is accurate or convincing our target audience to use our service.
Accomplishments that we're proud of
Make this happen in one and a half day
What we learned
- we need more data
- collaborative filtering is more suitable for niche category
What's next for openR
- Build API client to let user call programmatically via REST
- lnwshop / woocommerce integration
- POS integration
- Tracking recommended result
- Add more feature and shift mission from recommendation system to "Analytic Platform"