Five steps from a public web page to a ranked list you'll actually work. Vet the source before you touch it, write down what a good prospect looks like, then let the sheet do the scoring. You still make every call about who's worth your morning.
Fill it in right here, it scores as you go. Use Download CSV or Print to keep your copy. Nothing you type is saved to a server, it lives in this browser tab only.
Before anything else: is it public, and do its terms allow this? Sixty seconds that saves your reputation.
Point AI at the page you're already looking at and get structured rows out instead of a wall of text.
The roster gives you names. Research fills in the fields that decide fit: size, footprint, asset mix.
Apply your filters. Most rows should die here. A shorter list is the point, not a failure.
Score what survived, sort it, and start at the top on Monday. A list you don't work is a hobby.
You are licensed to look, not licensed to reproduce the database. Reading a result set you are entitled to see is what the subscription is for; walking the whole index into a file is what these agreements exist to stop. Run these checks once per platform and log the answer so you never do it twice.
Ctrl-F the terms for six words: robot, scrape, crawl, automated, data mining, artificial intelligence. One hit and you read that clause properly.
Add /robots.txt to the domain. Look for Disallow lines covering the pages you use, and for AI agents named directly.
One result set you filtered and were entitled to see is reading. Systematically capturing the whole index is not. Be honest about which.
Claude screenshots the tab and cannot filter what it sees. Use a separate browser profile with no client accounts signed in.
Public, no login, no anti-automation clause, no AI-bot block. Go, and stay on the pages you checked.
Yellow: ambiguous, personal data, or you couldn't verify. Red: any explicit prohibition. Red means find another source.
| Platform / source | URL | Login? | robots.txt OK? | Terms OK? | What the terms actually say | Verdict | Checked |
|---|
Row one is yours to fill in. Row two is the research source we use in step 3.
You cannot qualify a list against a feeling. Write the filters down before you look at a single name, or you will talk yourself into the wrong prospects at 4pm on a Friday. Be specific enough that someone else could apply your filters and get your answer.
The threshold below which it isn't worth your time. A number, not a vibe.
Your territory, or the markets where you can actually service the account.
The property types where your offer actually fits. Everything else is a distraction.
She asks for this every time. It separates a name from an account worth servicing.
Global companies you cannot service well are a real disqualifier, not a nice-to-have.
The things that kill a row on sight, no matter how good it looks otherwise.
The trigger that makes this the month rather than someday. Optional, but it's what makes outreach land.
If you can't finish this sentence, the list won't save you.
One row per company. Fill in what the roster gave you, then let research fill the rest. Fit and Reach are 1 to 5; the sheet multiplies them into a score and sorts on it. Rows that fail a hard filter get marked Cut and drop out of the count, but stay visible so you remember you checked them.
| Company | Source | HQ / market | Asset type | Portfolio | Meets size? | In geo? | Asset fit? | Why now (trigger) | Who to reach | Fit | Reach | Score | Cut | 1st |
|---|
Score = Fit × Reach, out of 25. Anything with a hard No in the three filter columns should probably be Cut. Three sample rows are pre-filled to show the shape; clear them and use your own.
Four moves, in order. Paste them into Claude in Chrome with the source page open, then into your regular chat for the enrichment and scoring passes. Swap the bracketed parts for your own filters from Step 2.
Starting points by category, with an honest flag on each. Green means public with no anti-automation clause found. Yellow means check it yourself, terms are ambiguous or the data includes named individuals. Red means an explicit prohibition; don't. Verify before you rely on any of this, terms change and this was checked in August 2026.