INTERNAL TOOL · GROWTH ENGINEERING
Social Comment Assist ("Fairway Radar")
A browser-based **"data washing machine"** that turns messy LinkedIn reactions and Instagram lists into structured, **enrichment-ready spreadsheets**—without using API keys.
Case Study Sections
* Select a section to view the technical breakdown of the 1-Week build.
The "No-API" Scraper
We needed a way to extract high-value leads from viral LinkedIn posts (people who "Liked" or commented) without risking our accounts using sketchy Chrome extensions or paying for expensive enterprise APIs.
The Core Problem
- Copy-Paste is broken: Highlighting a list of names on social media copies "garbage text" (buttons, timestamps, connection degrees) and—crucially—loses the profile URL.
- Extensions are risky: Platforms actively ban users who use unauthorized scraping extensions.
The Solution: "Fairway Radar" — a utility that uses advanced Regex and Clipboard API manipulation to reconstruct clean data from raw HTML, running safely inside your own browser tab.
User Profile
Growth Marketers & Recruiters needing to extract high-intent leads from viral social posts instantly.
Timeline
**1-Week Build** covering R&D, complex Regex logic, and browser clipboard manipulation.
Privacy Model
**Zero-Server Architecture.** All parsing happens in-browser using JavaScript, ensuring GDPR/CCPA safety.
The Challenge
Social platforms intentionally make data hard to extract. When a post goes viral, the "Likes" list is a goldmine of leads, but capturing them usually requires:
- Risky browser extensions that can get accounts banned.
- Expensive enterprise APIs that are overkill for quick tasks.
- Manual copy-pasting which results in unstructured garbage text (e.g., mixing up "titles" with "companies").
We needed a **safe, local tool** to clean this data instantly.
The 1-Week Build Journey
This wasn't just a UI wrapper; it required reverse-engineering the browser's Clipboard API and developing a heuristics engine to understand human job titles.
Decoding the "Copy-Paste" Chaos
- **The Discovery:** Standard `ctrl+v` loses the underlying profile URL (href), making the lead useless for outreach.
- **The Fix:** I wrote a custom `handleLinkedInPaste` function that intercepts the `onPaste` event. It ignores the plain text and parses the `text/html` clipboard data directly.
- Implemented `DOMParser` logic to extract hidden `linkedin.com/in/...` anchors and re-inject them into the raw text stream.
Teaching code to read Job Titles
- Raw data is messy: "VP of Sales at Acme | ex-Google | Dad".
- Built `splitTitleCompany`: A set of Regex rules that identifies separators like " at ", " @ ", " | ", or " - " to intelligently split the **Role** from the **Company**.
- Added noise filtering to strip out reaction keywords ("Like", "Love", "Celebrate") that pollute the name fields.
Generating Files in the Browser
- Integrated `SheetJS` to generate real `.xlsx` and `.csv` files entirely client-side (no backend required).
- Added a **Multi-Mode Architecture**: The tool can switch between "LinkedIn Reactions" (HTML parsing) and "Instagram Lists" (Text pattern matching) instantly.
- Implemented automatic deduplication (`seen` Set) to ensure that pasting the same scroll block twice doesn't result in duplicate contacts.
Making it feel like "Radar"
- Moved away from standard dashboards to a **Dark Mode / Glassmorphism** aesthetic using pure CSS-in-JS.
- Designed the "Capabilities Grid" and dynamic status badges to give the user immediate visual feedback on the state of the parser.
- Finalized legal disclaimers and privacy notices to ensure the tool is compliant with usage policies.
Net result: A robust internal tool that respects privacy, requires zero server costs, and turns a 2-hour manual data entry task into **15 seconds** of work.
Need a custom internal tool to automate your workflow?
Book a Strategy Call