About

A research project by Parkly

CarItch catalogues the everyday friction of owning a car in India — and ranks each problem by how painful, frequent, and unsolved it is. We publish the index openly so builders, journalists, regulators, and curious owners can use it.

Why this exists

India has over 300 million vehicle owners. The problems they face — insurance claim runarounds, service-centre overcharging, RTO bureaucracy, hidden dealer fees, parking scarcity, used-car fraud — are widely felt but poorly catalogued. Complaints scatter across public forum threads, social posts, news articles, and RTI filings; none of it is ranked, none of it is queryable.

CarItch is our attempt to fix that. Think of it as a living, AI-scored leaderboard of what actually frustrates Indian car owners — refreshed continuously, open to anyone.

What we publish

The index

A ranked list of itches — distinct problems scored 0-100 on severity, frequency, market size, and whitespace.

Raw complaints

Deduplicated source material — the forum posts, articles, and user submissions that each itch is built from.

Brand claims

When a company claims to solve a specific itch, their pitch goes on the page for community judgment (yes/no/neutral votes).

Trends & stats

Weekly movers, category breakdowns, source freshness — the shape of the data, in public.

Methodology

1. Collection

A scheduled job pulls from public RSS feeds, public forum APIs, and news aggregators covering the Indian automotive space. Anyone can also submit a problem directly via the site form.

2. Clustering

An AI model groups complaints that describe the same underlying problem. Fuzzy string matching and content fingerprints catch near-duplicates before they hit the clustering stage.

3. Scoring

Each itch gets a composite score from 0 to 100:

itch_score = (severity × 0.3 + frequency × 0.25 + tam × 0.2 + whitespace × 0.25) × 10

Severity (30%) — financial loss, safety risk, time cost
Frequency (25%) — how often this occurs across the 300M-owner base
TAM (20%) — size of the addressable market
Whitespace (25%) — how unsolved it is today

4. Quality gate

Before publication, each itch passes a PostgreSQL trigger that checks title format, length, description, category assignment, and fuzzy-duplicate detection. Failed itches go to manual review; they don't appear on the public leaderboard.

5. Re-ranking

Scores are re-computed every Sunday. New complaints arriving on an existing itch auto-boost its frequency dimension in real time via a database trigger.

What we don't do

Run ads, retargeting pixels, or third-party analytics.
Sell user data or brand contact details.
Make legal, safety, or purchasing recommendations — itch rankings are informational signals, not endorsements. See our Terms §10 and §8 for the full disclaimer.
Verify the commercial claims of brands that appear in the index. Brand pitches are reviewed for plausibility and relevance, not endorsed.

Transparency & limits

AI summarisation is imperfect. Titles are rewritten, descriptions are synthesised, and scores are probabilistic. If an itch misrepresents a specific incident, brand, or person, email [email protected] and we'll review.

Source code and migrations are open for inspection; the data layer runs on Supabase with row-level security, and the frontend is static HTML served by Cloudflare.

Who we are

CarItch is operated by the CarItch Editorial Team, c/o Parkly — a research project by Parkly, based in Sambhal, Uttar Pradesh, India. Parkly funds the infrastructure and hosts the project; editorial and ranking decisions are made by the CarItch team independently of any commercial relationship Parkly has with vendors in the automotive space.

Contact

General enquiries, corrections, press, partnerships — email [email protected].

Explore the index

The data surfaces as ranked itches, research write-ups, long-form explainers, and calculators. A few starting points: