Something Automated Setup for The Florida Keys
The Florida Keys The Wreckers John Viele is a scripted template built for Something Automated (SA), Chris B.'s web automation platform. It targets listing sites and data aggregators in the Florida Keys market — mainly real estate, rentals, and occasionally boat/property auction feeds. It automates the collection, parsing, and daily delivery of listing data into CSV, Google Sheets, or JSON depending on how you configure the output. SA runs headless Chrome instances. The script navigates to the target site, applies filters, scrolls through result pages, and scrapes the relevant fields. The Florida Keys template handles some of the common structural quirks in Keys-area property sites: pagination that loads via JavaScript, lazy-loaded images, and scattered price formatting (some sites use "$3,250/mo" while others just say "3250"). I set this up on a VPS with 4GB RAM and two simultaneous SA workers. It pulls roughly 800-1,200 listings per run across the three main sources in the script. Each run takes about 20 to 35 minutes depending on the target site's server response time. You'll want at least 2GB of free disk space per week for logs and output files.
Installation and Setup
First, you need Something Automated itself. Download it from the official site and install it on a machine that stays on — preferably a cloud VPS. The Windows version is the most stable. You don't need to run it directly; SA can be managed remotely via its web interface, which is handy if your local machine sleeps or goes offline. Once SA is running, you import The Florida Keys The Wreckers script. This is usually a .sa2 file or a folder with a manifest. Open SA, go to Import, and point it at the file. The script will appear under your templates. Before running it, open the script and review the configuration section at the top. You'll need to set: Target URLs — the base search URLs for the areas you want. The script comes with pre-built URLs for Key West, Marathon, and Key Largo, but you can add any sub-area.
Output path — where SA writes the CSV or sheet. I use a Google Sheet connected via the SA Google Sheets extension. It's more reliable than local CSVs because you don't have to move files around. Scroll depth and delay — the default values work for most sites, but if you hit rate limits, drop the scroll depth to 3 and increase the page delay to 4 seconds. This reduces your chances of getting temporarily blocked.
Get the Full Details
Running the Script for the First Time
Save the configuration and hit Run. The first run will take longer because SA caches the browser profile. Expect the initial crawl to take 45 minutes to an hour. After that, subsequent runs complete in 20 to 30 minutes if the target sites haven't changed their DOM structure. I encountered a specific issue on my second week of using this: Zillow's Florida Keys pages started returning a Cloudflare CAPTCHA challenge after about 80 page views. The script would hang on the captcha page and then fail silently because the selector for the listing grid no longer matched. I fixed it by adding a conditional wait step after the page load — if the URL contains "/captcha" or the listing grid element is missing, the script logs a warning and moves to the next search term instead of crashing. I also spaced the runs to once every 12 hours instead of every 4. That solved the blocking problem entirely.
Data Output and Post-Processing
The script outputs several fields: property address, price, beds, baths, square footage, listing URL, and date scraped. The price field sometimes comes through as a formatted string like "$425,000" and sometimes as a raw number. If you're feeding this into another system, normalize the price field immediately. A simple replace of "$", ",", and non-numeric characters with nothing, then cast to float, handles it. One thing beginners miss: the script does not deduplicate. If the same property appears on three different aggregator sites, you'll get three rows. I built a simple dedup step using the listing URL as the key. Properties that show up multiple times across runs are filtered on import into the spreadsheet. This cut my active leads from about 1,400 per day down to roughly 600 unique properties.
Limitations and What It Can't Do
The script won't handle sites behind login walls. If a Keys listing site requires an agent account to view details, this template isn't going to work unless you manually inject cookies into the SA browser profile, and even then it's fragile. I tried it with one local MLS mirror and spent three days debugging session expiration before giving up. Another limitation: the script assumes a certain HTML structure. When sites update their templates — which happens more often than people admit — the selectors break. The Florida Keys template hasn't been updated since mid-2024, so if a target site changed its DOM after that, you'll need to inspect the page manually and update the relevant CSS selectors in the script. The hardest part is finding which selector corresponds to which field. I keep a reference table in a note: field name, XPath, and a screenshot of the inspected element. It saves about 20 minutes per broken field. The biggest bottleneck is actually the headless Chrome process itself. SA's default memory usage per worker is around 600MB. If you run more than three workers on a 4GB machine, the system starts swapping and performance degrades badly. I learned this the hard way when my third worker caused the entire VPS to become unresponsive. Dropping back to two workers stabilized everything.
Alternatives to Consider
If you need data from sites that don't render via JavaScript — like some older listing platforms — SA isn't your best tool. A simple Python script with requests and BeautifulSoup will be faster and less resource-intensive. I switched one of my secondary feeds to a lightweight Python scraper and it completed the same data pull in about three minutes instead of 25. For sites that do require JS rendering, the alternative to SA would be tools like Puppeteer or Playwright written as custom scripts. They offer more control but require significantly more development time. For someone who just wants to pull Keys listings daily without maintaining code, The Florida Keys The Wreckers template is a reasonable starting point, provided you're willing to maintain it when targets change.
Final Thoughts
Set up the script, monitor the first few runs closely, check your output for missing fields, and adjust the selectors if anything looks off. The learning curve is maybe two hours if you're familiar with SA. After that, it runs mostly on autopilot except when a site updates its layout. Keep your VPS updated, don't overwork the machine, and check the logs weekly. That's about it.
