AI lead research
AI lead research that reads every company's website
Before a good rep calls a small business, they read its website. Ripple does that for every company in a scan: Claude, Anthropic's AI, reads up to four pages, pulls out the facts, judges the fit against your ideal customer and writes the brief.
No subscription · $10 of free credit · Pay only for the leads you get
Pages read
- gorgehollow.example
- gorgehollow.example/about
- gorgehollow.example/services
- gorgehollow.example/projects
Found
- Owner
- Callum Reid (About page)
- Team
- About 6 (“our team of six”)
- Software
- Google Workspace; website on an old template
- Fit
- Good: matches 2 of your good signs
Scored B · 78 · call-ready
What it reads, and what it won't guess
Facts from the pages
The owner or manager, phone, email, team size, services, service area, up to four recent projects and the software the company uses, as named on its pages or spotted in the site’s code.
Blank, not guessed
When a site doesn't say something, the field stays empty. Phone numbers and emails must appear on the pages, and team size is only estimated from a clear basis, like a team page.
Sources you can check
Every profile lists the pages Ripple read, so a caller can check a fact before they rely on it. Photos of the company's work are linked from its site, never copied.
Your ideal customer, in plain words
Ripple scores companies against what you tell it about your market, written the way you'd brief a new rep:
- What you sell: a summary and your main selling points, so briefs speak to your product
- Business types: what to search for, and how good a fit each one usually is
- Team size: the range you sell to best
- Good signs and deal breakers: anything a website could show, in your own words
- Competitors and partners: software that counts against a lead, and software you work with that counts for it
Claude copies the good signs and deal breakers it finds exactly as you wrote them, so the reason behind every score reads in your team's language. A deal breaker only counts when the pages clearly show it.
| Points | |
|---|---|
| Ideal-customer fit | Up to 25How well the company matches the business types and good signs you describe, as Claude judges from its website. |
| Company size | Up to 20Full marks inside your ideal team size; fewer points the further outside it. Unknown sizes score as average. |
| Current tools | Up to 20Highest when no competing software shows up; low when a competitor does. A bonus for software you work with. |
| Activity & growth | Up to 15Recent projects and posts, and signs of growth such as hiring. |
| Reputation | Up to 10How well reviewed the company is on Google. Only the score's effect is kept, never Google's rating itself. |
| Reachability | Up to 10Whether there is a named owner, an email and a website to work from. |
A call brief your callers will actually read
Two to four reasons to call
One sentence each, specific to this company and to what you sell. No generic talking points.
An opener
One or two spoken sentences: the owner by first name when known, something specific from their site, then a question that leads to the problem you solve.
Cautions, only when needed
Added when something points to a poor fit, such as a competitor already in use or one of your deal breakers.
Deep dives for the leads that matter
A deep dive goes past the website. Claude searches the web for reviews, social pages, news, job posts and who runs the company, then combines that with up to ten pages of the company's site.
Everything it adds is marked in the profile, so callers know what came from the deep dive, and the lead is scored again with the new evidence. Run one from any profile for $3, or let a Deep scan run one on every call-ready lead.
Scores that learn from your team
When a caller would score a lead differently, they correct it and say why. Their score replaces the AI's everywhere (lists, map, call queue), with the AI's kept beside it, and it can be undone.
Claude sees your team's latest corrections whenever it scores a company, so its judgment leans toward yours. On request, it can also read all the corrections and suggest specific changes to your ideal customer. Nothing changes until someone applies a suggestion.
FAQ
AI research questions
Which AI does Ripple use?
Claude, made by Anthropic. It reads each company's website, judges the fit against your ideal customer and writes the call brief. Ripple's own scoring then turns those findings into the 100-point score.
Does the AI make things up?
It's told to use only what the pages say and to leave a field blank when a site doesn't say it: it never guesses a phone number, email, person, year or team size. Like any AI it can misread a page, so each profile lists the pages it read and your team can correct any score.
Can I change how leads are scored?
Yes, through your ideal-customer settings: the business types you sell to and how well each fits, your ideal team size, good signs, deal breakers, competitors and the software you work with. Update them any time; new research uses your latest settings.
What happens if a website is down or can’t be read?
The scan's activity feed notes that the company couldn't be researched and why, and the scan moves on. You aren't charged for websites Ripple couldn't read.
What if a website tries to give the AI instructions?
Website text is treated as untrusted data. Claude uses it only as a source of facts about the company and ignores any instructions it contains.
Let AI do the pre-call research
Draw a territory and get every company in it researched, scored and briefed before your team picks up the phone.
No subscription · $10 of free credit · Pay only for the leads you get