Fleads

• 

  startup

• 

  leads - reveal on hover

• 

• 

  virality analysis

• 

  collaborators list

• 

  caption, hashtags and other insights

Inspiration

The inspiration of this project is pretty interesting. So, my father owns a cosmetic brand and he has someone who does his social media. what i noticed one day was that nowadays instagram is filled with hate comments. These hate comments are divided into two types. the first type are unnecessary hate, which is baseless, while the other is actual hate, which comes from criticism for the brand and its products. this second type of haters are people who actually tried the brand, but were upset. these are also the people that have so much potential of being customers for other brands. This is where i got the idea of designing a way of finding the right leads through this process, which led me to Fleads.

What it does

Fleads is a smart B2B tool that converts your competitors’ unhappy customers into your future loyal users. It begins by analyzing your company's social media to identify direct competitors in the space. Once the competitors are locked in, Fleads scrapes their most viral and trending posts, dives deep into the comment sections, and applies filters to detect genuine dissatisfaction—not baseless hate, but real feedback from users who were let down by the product or service.

These users are pure gold—they already care about the space, they’ve made a purchase decision, and they’re looking for better alternatives. Fleads captures these profiles and serves them to you as high-potential leads.

How we built it

Fleads is built using a blend of modern tools for scraping, AI-based sentiment analysis, and data visualization. We used Node.js for backend scraping with Instagram Apify API's. Sentiment and context-based comment filtering was handled using Sonar Reasoning Pro model for finding the right competitors and differentiating between trolling and actual product criticism.

Challenges we ran into

One of the challenges we faced was that at the start i was trying to get viral reel urls from sonar API, but i later realised that the API couldn't directly fetch accurate reel links. Then i switched to getting only the instagram usernames of the right competitors, which was done accurately by sonar, and further used a different instagram API for finding viral reels.

Another major challenge was differentiating between baseless hate and genuine customer dissatisfaction. Basic sentiment analysis models were not enough. We had to fine-tune our classifiers to identify not just negative tones but relevant contextual signals indicating experience-based reviews.

Accomplishments that we're proud of

We’re proud that Fleads has turned a simple observation into a powerful SaaS tool. It’s not just about scraping data—it’s about understanding intent, finding patterns, and giving brands actionable intelligence.

The fact that it all started from a real-world observation of a cosmetic brand’s social media makes the journey even more meaningful. We've built a pipeline that finds business opportunities in places no one else ever has.

What we learned

Through this project, we learned how critical context is in data. Sentiment alone isn’t enough—you need why and where behind it. Our analytical module taught us how much brands can learn from competitors’ captions, hashtags, and collaborations.

Above all, we learned that each complaint for your competitor can be a conversation starter for you, and if you outdo their past experiences, it can be the start of something valuable.

What's next for Fleads

I faced a problem in finding contact details of the leads because it posed privacy concerns, my next step would definitely be finding someway of contact- which could be automated DMs to the leads or any other more efficient way.

Next, we plan to introduce automated engagement—think auto DMs to filtered leads, guided by custom tone templates. We’re also working on integrating more platforms like YouTube and LinkedIn for B2B brands.

Built With

• apify-instagram-api
• nextjs
• node.js
• sonar-reasoning-pro