Inspiration

Creators we know and love recently exposed Honey's alleged malpractice of taking commissions from creators' affiliate links and withholding savings from consumers.

Our team saw a gap in improving the affiliate link space, and rebuilding trust by helping e-commerce store owners, creators, and consumers.

We want to give power back to store owners and creators by enabling them to personalize experiences for their customers.

What it does

Imagine you are buying a laptop after watching a 15-minute video from Marques, MKBHD. You click on the affiliate link, enter the online store, and get the same shopping experience as any customer.

Why doesn't the store know about the video you just watched and the creator you follow?

With Flink, every affiliate link routes to a personalized landing page using AI to understand your context.

On our website, e-commerce store owners generate affiliate links by indicating the content source where the link will be used (i.e., a YouTube video). Our AI parses the content to create a link that routes to a personalized landing page for the store. Store owners can track the performance and usage of their affiliate links using our website.

How we built it

Backend: FastAPI with SQLite database (Dockerized).

Frontend: React and Vite (hosted on Netlify with a GoDaddy domain).

Demo Store: Shopify.

Challenges we ran into

Most brands use a standardized URL format that includes UTM parameters in the domain for tracking and analytics. Generating a new link, however, would disrupt the original tracking.

While we considered offering our own tracking service, this would create unnecessary challenges for store owners, requiring them to migrate their existing systems.

We also explored leveraging existing URL formats as a unique identifier for our custom affiliate links, but we found that current URLs lack the granular segmentation needed for video-level personalization. For example, creators often use the same affiliate link across all their videos. This meant we could not use existing URLs to decipher which customized store to route to.

Accomplishments that we're proud of

When building our service, we aimed to maintain ease of use to allow store owners and creators to onboard with Flink with minimal overhead. During our design and implementation, we made pivots to achieve this goal.

For store owners, our original solution would have overwritten the existing tracking that brands use. Although this would fulfill our technical feature to personalize the landing stores, this would harm the integration and ease of use for store owners. We modified our approach to support existing analytics with no overhead or conflicts.

For creators, the process is still one step: get the affiliate link. Nothing else is required, allowing them to focus on their craft.

What we learned

We used FastAPI for the first time which automatically generated Swagger Documentation. This documentation showed all of the API endpoints of our backend and the associated schema.

Using this technology allowed the frontend and backend development to be parallelized as it simplified the technical communication.

Typically, we would use services like Postman to test our backend before integrating with the frontend. Using FastAPI taught us a way to automate our development for documentation and testing.

What's next for Flink

Flink's unique advantage is that it understands the creator and the customer. This data comes from videos, blogs, or any online post.

Currently, Flink supports parsing YouTube videos using AI. Next, Flink will support blogs and short-form social media content.

Built With

Share this project:

Updates