A local business needs an AI agent when enquiries are arriving faster than a small team can answer them and the cost of a missed one is high. The seven signals below are all visible before you speak to the owner — which is what makes them useful for prospecting. Each is deterministic: traceable to the business record or its trade, not a guess.
This is part of the AI for client acquisition pillar and follows where to find AI automation clients.
The Seven Signals
| # | Signal | Where you see it | Weight | |---|---|---|---| | 1 | Enquiry-driven trade | Category: clinic, law firm, salon, real estate, trades, hospitality | Sets the ceiling | | 2 | Review complaints about response time | Recent reviews: "couldn't get through", "no reply", "rang twice" | High | | 3 | Phone-only or manual booking | No booking widget; "call to book" | High | | 4 | Contact form into a shared inbox | Generic info@ address, no routing, no auto-reply | Medium | | 5 | Small, owner-operated team | Photos, "meet the team", review mentions | Medium | | 6 | High transaction value | Trade norms: a booking worth hundreds or thousands | Raises priority | | 7 | Strong reputation to protect | High review count and rating | Raises urgency |
First, Check the Business Is Worth Pitching At All
Before any of the seven signals matter, one question comes first: is this a real, healthy business that could actually pay you?
This ordering is not a stylistic preference. A dormant business with no website, no booking system and no recent activity looks like a perfect opportunity on paper — every gap you could name is present — and it is a worthless lead. Score the opportunity before the business and exactly that listing floats to the top of your list.
So check for evidence of real trade first:
- Review volume. Reviews are the one public, hard-to-fake measure of whether customers are actually showing up. A business with 100+ reviews is demonstrably trading.
- Rating alongside volume. Volume with a poor rating is a business with problems, not a business doing well. The two have to be read together.
- Recency. A listing whose last review is fourteen months old may be running on old momentum. Recent reviews — within the last quarter — are what tell you the business is still active today.
- Someone tending the listing. A phone number, complete opening hours, an operational status. These are small things, and their absence together is a pattern.
In Runvax this runs as a hard gate rather than a scoring input: businesses below a strength threshold are cut from the list entirely, not ranked last. The list is a cut, not a sort — if every business in a search survives, you have a directory with better labels, not a shortlist. That distinction is the whole difference between a lead list and a prospect list.
Every Signal Should Be a Fact, Not an Interpretation
The discipline that makes these signals usable is separating what you can point at from what you conclude.
"Their reviews mention nobody answering the phone" is a fact. "They need a booking system" is an interpretation. "They get around 200 enquiries a week" is a guess, and guesses are what get you caught out on a first call.
Keep the two layers apart and every claim in your outreach can survive the obvious follow-up question — why do you think that? — because you can quote the review. Collapse them and you are pitching a diagnosis you cannot defend. In Runvax this is enforced structurally: the signal layer extracts only facts, each one carrying the evidence string shown on the opportunity card, and interpretation happens one layer up.
Signal 2 in Detail: What Customers Actually Write
Review text is the richest business-specific evidence available, and the only place a customer tells you directly that enquiries are being dropped. It clusters into five distinct complaints, and the distinction between them matters because they point at different products:
| Complaint | Phrasings that indicate it | What it points to | |---|---|---| | Missed enquiries | "nobody answered", "called three times", "couldn't get through", "phone is always off" | Answering / routing | | Slow replies | "waited days", "still waiting", "had to chase them" | Response automation | | Booking friction | "hard to get an appointment", "no online booking", "waited over an hour" | Booking system | | Information gap | "no prices listed anywhere", "had to call just to find out", "wish they had a website" | A website, not an agent | | Staff stretched | "understaffed", "only one receptionist", "too busy to help" | Coverage outside staffed hours |
That fourth row is the one most people get wrong. A customer complaining they could not find prices online is not asking for an AI agent — they are describing a missing page. Pitch an enquiry-handling system to that business and you are solving a problem they do not have. It is the single most common mis-pitch in this niche, and reading the complaint carefully is all it takes to avoid it.
Worth noting on method: matching these phrasings should be deliberately conservative. It is better to miss a business whose reviews hint at a problem than to flag one whose reviews do not support the claim — because the entire value of the signal is that you can quote it back. Recall costs you a prospect; a false positive costs you credibility on the call.
How to Read the Signals Together
Signal 1 tells you whether an AI agent could plausibly help at all. A busy dental clinic has a high ceiling; a sole-trader bookkeeper has a low one. But category alone cannot rank two businesses within the same category — every clinic in a city shares the same trade, so the category tells you nothing about which clinic to call first.
That is what signals 2–5 are for. They describe this specific business rather than its type, and they are the only signals that should move a business up your list or raise your confidence. Signals 6–7 then tell you how urgent the case is and how much the business can pay.
A business that hits 1, 2, 3 and 6 — enquiry-driven, review complaints about replies, phone-only booking, high-value bookings — is a strong prospect. A business that hits only signal 1 is an industry match and nothing more, and pitching it is a coin flip.
This is the same two-layer logic that lets AI find businesses without a website reliably: category sets the range, business-specific signals place the business inside it.
What the Agent Actually Does for Each Signal
- Signals 2 & 3: answers the enquiry and books the slot, day or night.
- Signal 4: routes, acknowledges, and captures every enquiry instead of dropping it in a shared inbox.
- Signal 5: covers the hours a small team cannot.
- Signals 6 & 7: protects revenue and reputation by making sure no high-value enquiry goes cold.
Using the Signals to Prospect
Reading these signals for one business takes about five minutes: check the category, skim recent reviews for the five complaints above, look for a booking widget, check whether a real inbox or a generic one is listed. Doing it for eighty businesses in a city takes a day you do not have — which is the actual reason most people pitch on category alone and wonder why the reply rate is poor.
Runvax reads them for every business in a search, applies the strength gate first, scores the opportunity, and drafts a first message built on the specific signals it found — opening on the missed enquiry, never on the agent. For how to pitch once you have the list, see the client acquisition guide for AI agent builders, and next in this pillar, the AI agency client-acquisition playbook.
Score your first market free — no card needed.