Lead Scoring for Local Businesses: A Framework That Works
Most lead scoring is arithmetic theatre — arbitrary points assigned to arbitrary attributes, summed into a number that feels objective and predicts nothing. A useful score for local B2B answers one question: if I contact this business today, how likely is a reply? Everything that doesn't move that probability is noise.
Here's a framework you can run in a spreadsheet, and how to make it actually predictive rather than decorative.
Two things a score must combine
Every useful local lead score is the intersection of:
- Is there a reason to contact them? — a visible problem you fix
- Can they act on it? — a real, trading business with money and a decision-maker
Miss the first and you're spamming. Miss the second and you're writing beautiful messages to dormant listings. Most scoring systems measure only the first.
The signals that actually matter
Reason signals — why contact them
| Signal | Strength | Why |
|---|---|---|
| No website at all | Strong | Binary, verifiable, easy pitch |
| Social page used as website | Strongest | They already decided presence matters |
| Free page-builder site | Strongest | Proven willingness to pay |
| Site broken or years stale | Strong | Concrete and demonstrable |
| Rating below 4.2 with many reviews | Medium | Real problem, delicate to raise |
| Very few reviews for years trading | Medium | Fixable, visible win |
| Google profile unmanaged | Medium | Cheap fix, good entry point |
| Many reviews, good rating, good site | Negative | They're doing fine. Deprioritise |
That last row is where most people go wrong. A thriving, well-marketed business is a worse lead than a struggling one, not a better one. Success on paper is not buying intent.
Capacity signals — can they act
| Signal | Meaning |
|---|---|
| Review count 15+ | Actually trading. The single best proxy |
| Phone number present | Listing is maintained |
| Independent, not a chain | The person you reach can decide |
| Trading 3+ years | Past survival mode, has budget |
| Email discoverable | Reachable without a phone call |
Review count is the most underrated filter in local prospecting. It's a proxy for revenue, longevity and customer volume all at once, and it's free to check. A business with two reviews and no website isn't a lead — it's often not a business.
The grades
Keep it to four, and define them by action rather than points:
A — contact this week. Strong reason signal and strong capacity. A salon with 87 reviews, 4.9 stars, Facebook as its website.
B — contact when you run out of A. Reason present but weaker, or capacity slightly thin. Site exists but looks abandoned; 20 reviews.
C — no clear reason. Real business, nothing obviously wrong. Don't contact. Revisit in six months.
D — not a fit. Chain, dormant, wrong category, out of area. Delete permanently.
Healthy distribution from a raw export: roughly 15% A, 25% B, 35% C, 25% D. If you're getting 60% A, your criteria are too generous and the grade means nothing.
Calibrating it to reality
This is the step everyone skips, and it's what separates a real score from decoration.
After 100 contacts, go back and check which signals actually predicted replies.
Track it simply:
| Signal | Contacted | Replied | Rate |
|---|---|---|---|
| No website | 30 | 9 | 30% |
| Facebook as website | 22 | 8 | 36% |
| Weak site | 25 | 4 | 16% |
| Low rating | 23 | 2 | 9% |
Now you know something real about your market and your offer. In that example, low rating is barely worth contacting and should drop out of A entirely, while the Facebook-as-website segment deserves more of your week.
Your numbers will differ from anyone else's — by country, niche, and what you sell. That's precisely why a generic scoring model out of the box is worth less than one you've calibrated over a hundred contacts.
Two mistakes that make scores useless
Too many factors. Twelve weighted criteria produce a number nobody understands and nobody can debug. Three or four signals plus a capacity floor is enough. If you can't explain in one sentence why something is an A, the model is too complex.
Never revisiting it. A score set once and never checked against outcomes is a superstition. Recalibrate every hundred contacts or whenever you change what you sell — the signals that predict replies for websites are not the ones that predict replies for SEO.
Doing it by hand
You don't need software. A spreadsheet with columns for business, review count, rating, website status, signal and grade will do it. Sorting by grade takes seconds.
The cost is time: about 20–30 seconds per business to check and classify. For 200 businesses that's roughly ninety minutes, every campaign.
That's the part Lead Radary automates — it applies this framework to Google Maps data and returns the grade with the reason attached. The calibration step, though, stays yours. Nobody else can tell you which signal predicts replies in your market.
Summary
- A score must combine a reason to contact and the capacity to act.
- Social-page-as-website and free-builder sites are the strongest reason signals.
- Review count is the best single capacity filter. Set a floor of 15.
- A thriving, well-marketed business is a worse lead, not a better one.
- Four grades defined by action. Recalibrate every 100 contacts.