← Sticker Wall AGENT · DETAIL

AI Real Estate Lead Broker: Automated Property Scraping and Buyer Intent Scoring & Distribution, $3.5K/Month

Workflow: Use n8n scheduled scrapers daily to collect updated data from public property sites like Ke, Lianjia, as well as buyer i

AGENT

Key Fields

FIELD STAMPS

IndustryMarketing / Advertising

RegionMulti-region（北美/中国一二线）

ScaleSME

ChannelOnline

🔧 Workflow

Use n8n scheduled scrapers daily to collect updated data from public property sites like Ke, Lianjia, as well as buyer intent posts on Xiaohongshu, Weibo, and local real estate forums. Feed this into the Claude Sonnet API to extract 5 dimensional features: budget range, target region, layout requirements, home-buying time window, and historical viewing records. AI automatically assigns an intent score of 0-100 to each lead, and outputs a batch of 30-50 high-intent lead cards every morning at 9:00 AM. These are pushed to the contracted agency's WeChat Work group, complete with preliminary matching property suggestions and agent communication scripts. Every Monday, generate a conversion rate report for the previous week, adjust the scoring model weights based on agent feedback, and input raw property text and buyer intent content to output standard lead cards featuring scores, matching suggestions, and follow-up tips.

🛠 Setup Requirements

Requires a 7x24 cloud server to host the n8n self-hosted version, basic Python scripting skills, familiarity with Claude API calls and prompt tuning, and a basic understanding of real estate transaction terminology to communicate effectively with agents. The prototype system can be set up in 3 days for basic scraping + scoring workflows, and the first batch of 3-5 pilot agencies can be onboarded for trials within a week. No front-end development skills are required; the main tasks are n8n workflow node assembly and scoring prompt iteration. In the early stages, there is no need to purchase professional real estate data interfaces. Prioritize scraping data from public sites and social platforms, keeping compliance costs extremely low.

🧰 Toolchain

🔧 n8n
🔧 Claude Sonnet API
🔧 Apify Scraping Actor
🔧 WeChat Work API

💰 Revenue

① Domestic lower-tier markets selling leads per item (main revenue): Local small and medium-sized chain brokerage companies pay 30-50 RMB per lead. Based on the Shenzhen case in the card at 35 RMB per lead with 800 leads delivered monthly, this yields a monthly income of about 28,000 RMB (source case, independently unverified); ② Overseas mid-sized brokerage companies billing per item (USD): A mid-sized brokerage in Austin, USA pays per lead at $50-$80 per lead. The card's case shows a monthly income of $3,500, though the exact delivery volume is untraceable and the proportion is unknown (case material, missing verification step); ③ Commission sharing on closed deals: Regional chain agencies deduct 10% of the final transaction commission. The Hangzhou case in the card signed 5 agencies and supplied leads monthly, with an average monthly commission of 12,000 RMB. The transaction base corresponding to the commission cannot be verified, and the proportion is unclear (self-reported case, no independent verification); ④ Opportunity item - Turning the lead scoring system into a SaaS product for agency stores to self-subscribe: A JLL report states that the global commercial property AI pilot adoption rate rose to 92% in 2026 (media figures, lacking independent verification); revenue data for this path is not yet available.

💸 Cost

Claude Sonnet API call fees are about $50-$80/month, Apify data scraping quotas are about $20/month, and cloud server hosting fees are about $10/month, bringing total costs to under $100/month.

⏱ Time Investment

Invest 1-2 hours daily to handle agent feedback and adjust scoring models, about 10 hours per week, with weekends dedicated to optimizing workflows and onboarding new agencies.

🚀 Getting Started

First, target local mid-sized chain brokerage firms with 10-50 employees. These entities have a strong willingness to pay for high-quality leads, short decision-making chains, and lower qualification requirements than large brokerages. Spend 3 days focusing on running through the property scraping and intent scoring process for a core residential complex, output 10 valid leads as a Demo, and use them to pitch 3 brokerage firms for pilot cooperation. Agree on post-payment via commission sharing or per-lead settlement without advancing any costs. If agencies have doubts about the model's scoring, offer 10 free leads for trial first, and sign a formal cooperation agreement once the conversion rate meets expectations.

🔑 Keys to Success

✅ The intent scoring model must go beyond simple keyword matching, focusing on 4 high-weight features: budget range, home-buying time window, historical viewing frequency, and whether specific property details have been queried. This is Claude's core advantage over general models.
✅ Initial cooperative agencies must sign a formal contract before leads are delivered. Free trials exceeding 3 days are strictly prohibited to prevent leads from being exploited without payment.
✅ Property and intent data scraping should prioritize public sites and social platforms, avoiding paywalls and non-public databases. Compliance risks are far lower than scraping internal system data.
✅ Each lead delivered must come with 2-3 matching properties and standardized communication scripts to improve the agent's first-touch conversion rate; otherwise, orders will be canceled due to poor lead quality.

⚠️ 风险

⚠️ Anti-scraping mechanisms on public property sites and social platforms will trigger bans as scraping frequency increases. Preparation is needed in advance for User-Agent rotation, distributed proxy pools, and scraping frequency downgrade contingency plans.
⚠️ If AI intent scoring shifts and leads to low actual conversion rates for high-scoring leads, agencies will refuse payment and terminate cooperation. Scoring model weights must be iterated weekly based on transaction data.
⚠️ Starting in 2026, some cities require real estate agency service providers to complete qualification filings. Providing leads to uncertified brokerage entities may carry administrative penalty risks. Verify the other party's business qualifications before signing.

📌 Real Cases

📌 Austin Real Estate Prospect Scoring Project Case: A developer used n8n + Claude to build a real estate prospect scoring system for a mid-sized brokerage in Austin, USA. It automatically scraped MLS public listings and local forum buyer posts, generating $3,500/month with lead conversion rates 40% higher than traditional manual screening.
📌 Domestic AI Automation Service Provider Case: This provider built a Xiaohongshu buyer listing scraping + scoring system in Hangzhou, signed 5 regional chain agencies, delivered high-intent leads monthly, and earned an average monthly commission of 12,000 RMB.
📌 Shenzhen Independent Developer Case: This developer built a buyer intent scraping system for the local second-hand housing market, signed 3 agency stores, charged 35 RMB/lead, delivered an average of 800 leads per month, and generated a monthly income of about 28,000 RMB.

Sources