Overview
LeadReacher is a Windows desktop app that finds local businesses worth pitching. You describe the kind of customer you want; it discovers matching businesses from Google Maps, enriches each with public web and social signals, scores how good a lead they are, and keeps your whole pipeline organized in one place.
It runs entirely on your machine — the scraping happens from your own computer and your leads live in a local database. Nothing is sent to a server to be processed.
Getting started
- Create a hunt. Pick a what to find preset, a category (e.g. “hair salon”), and a location.
- Watch it work. Leads stream into the table live as the scraper discovers and enriches them — scores fill in as they go.
- Open a lead. Click any row to see its full profile: contact info, source links, both scores, and every scraped signal.
- Work the pipeline. Set a status as you go (reviewed → contacted → won / archived) and pin the best ones to the map.
- Save the hunt to keep it in your workstation. Unsaved hunts are session-only and cleared on the next launch, so test runs don’t pile up.
Export to CSV any time, and re-import it later — it merges with de-duplication so you never get duplicates.
Presets — what each “what to find” means
A preset is a lens: the same businesses rank differently depending on the problem you solve.
| Preset | Finds businesses that… | Good for |
|---|---|---|
| No social presence | have no Instagram or an abandoned account | social-media services |
| Old or broken website | have a stale, insecure, or thin site | web design / dev |
| Hard to find on Google | are weak on on-page SEO basics | SEO services |
| Bad reviews or unclaimed | have low ratings or unclaimed listings | reputation management |
| All local businesses | match the area, no problem filter | general prospecting |
Two scores: Opportunity vs Fit
Every lead carries two separate 0–100 scores, so you always see the difference between “matches the rules” and “matches you.”
Opportunity
A transparent rules score for the active preset — how well a business matches that problem. Available from day one; the same for everyone.
Fit
A personalized score from your own on-device model — how much a lead looks like the ones you win. Grows more accurate as you label outcomes.
Until your model has learned enough, Fit shows a dash rather than a guess — we never show a misleading number.
The AI fit model — how it learns
This isn’t an “AI” label slapped on a rules engine. It’s a real, if lightweight, machine-learning model that trains on your own outcomes and gets measurably better the more you use it.
- It learns from you. Leads you mark Won are positive examples; Archived are negatives. The model finds the pattern separating them and scores new leads by it.
- It starts neutral. It needs a small floor of labeled wins and losses before it switches on — until then, ranking uses the transparent rules only.
- It earns trust gradually. The more you label, the more weight the model gets in ranking. You can watch that influence grow inside the app.
- It explains itself. It’s an interpretable model — it shows the signals behind your wins (“you tend to win: many reviews, no Instagram, a free email domain”), never a black box.
- It stays yours. The model is trained on your device, from your data only. It is never uploaded, pooled, or shared between users.
What the model looks at
Each lead is scored on more than a dozen signals across four data sources — all gathered from public pages, on your machine:
Working your leads
Statuses & the pipeline
Move each lead through new → reviewed → contacted → won / archived. Won and archived are also what train your fit model, so working your pipeline makes the AI smarter.
Map & pins
Every hunted business is auto-pinned and colored by score. Pin the ones you care about to keep them on the Explore map across all your hunts — handy for working a neighbourhood.
Saved hunts & change detection
Save a hunt to keep it. Re-run it later and LeadReacher flags what’s new and what changed (reviews jumped, a website appeared) — so your list is a living pipeline, not a one-off dump. Stable identity per business means no duplicates on re-run.
Deep Scan
Run a heavier on-page SEO / performance / accessibility audit on just your shortlist, then the leads re-score with those deeper signals folded in. Fast hunts by default; depth only where it counts.
CSV in & out
Export your leads to CSV for your email tool or CRM, and import a CSV back — it merges into the workspace with de-duplication.
Privacy & your data
- Scraping runs on your machine from public pages — your leads never pass through our servers.
- Your data is local and per-account, with automatic backups taken before risky operations and on a timer.
- Instagram is handle-only — we read the handle from pages we already have and never visit instagram.com or scrape activity.
- Your fit model is on-device. It trains from your outcomes and is never pooled or uploaded.
- No API keys required. Maps and geocoding use open sources — there’s nothing to configure.
Glossary
| Hunt | One search run: a preset + category + location that discovers and scores businesses. |
| Lead | A single business found by a hunt, with its scraped signals and scores. |
| Opportunity score | Rules-based 0–100 fit for the active preset. |
| Fit score | Personalized 0–100 from your on-device model, learned from your won/archived leads. |
| Signal | One scraped fact about a business (e.g. “no HTTPS”) that feeds the scores. |
| Status | Where a lead is in your pipeline: new, reviewed, contacted, won, archived. |
| Pin | Marks a lead to show on the Explore map across all hunts. |
| Saved hunt | A hunt you keep; unsaved ones are session-only. |
| View | A saved filter across all your leads (e.g. “contacted salons in Manitoba”). |
| Deep Scan | An on-demand SEO / performance / accessibility audit on a shortlist. |
FAQ
Do I need any API keys or accounts to scrape?
No. Maps discovery and geocoding use open sources, and scraping runs from your own machine.
Will the AI work on day one?
It ranks with transparent rules immediately. The personalized fit model switches on once you’ve labeled a small floor of won and archived leads, then keeps improving.
Is my lead data uploaded anywhere?
No. Leads and the trained model live on your device. Backups are local files.
Does it work outside North America?
Yes — hunt in any country and city. Locale-aware parsing pulls ratings, reviews, and addresses wherever the businesses are.
What platforms are supported?
Windows desktop today. See the waitlist for availability updates.
Feedback & support
LeadReacher is in active development and we read everything — bugs, missing data, a preset that mis-ranks, or a feature you wish existed.
Not on the beta yet? Join the waitlist — early-access invites go out first, and that’s where we’ll share how to send feedback once the beta is in your hands.