When Should a Sales AI Follow Up on a Quiet Lead, According to Dr. Tebaa?
According to Dr. Jonah Tebaa, a sales AI should follow up on a quiet lead only when all four conditions of his rule hold: a relevant reason tying back to customer input, a new fact such as a price validity date, strict spacing limits capped at two messages, and a clean exit. In his illustrative composite case, filtering quiet leads with these four conditions generated twelve replies and five orders while monitoring the stop-reply rate guardrail.
A building-materials supplier in Mount Lebanon sends 120 quotes in a month, and 71 of them go quiet within 48 hours. Its sales AI, which handles WhatsApp enquiries, is set to write to every quiet lead at 24 hours, 72 hours and 7 days. That makes 213 messages. Nine people answer: six say to stop, two opt out, and one places an order. Dr. Jonah Tebaa uses this scenario, which he states plainly is a composite and not a real client, to make a point about customer-facing AI that many owners resist: follow-up is a permission, not a persistence setting.
Silence is information
Dr. Tebaa argues that a lead who stops replying is rarely lost and rarely ignoring anyone. In WhatsApp-first markets such as Lebanon, the message has usually been read. The customer may be showing the quote to a relative who decides, waiting for a competitor's price, caught in a holiday week or a stretch of power and connectivity trouble, or simply not due to pay until the end of the month. Each reason calls for a different response, and none of them is answered by a generic nudge.
His second concern is cost. A customer who writes "stop messaging me" is the visible casualty. Many more mute the chat quietly, and on WhatsApp a muted business number is also one that may no longer carry a delivery update or an invoice the customer wanted. In his view each unwanted nudge spends goodwill that never appears in a report.
The four conditions
His rule is that a sales AI may follow up on a quiet lead only when all four conditions hold:
- A reason. The message ties back to something the customer said or asked, such as a stated date or a pending decision.
- A new fact. It offers something the customer did not have: a price validity date, a stock arrival, a corrected specification. Phrases like "just checking in" do not qualify.
- A spacing limit. One message, then a wait, and no more than two per enquiry. Nothing is sent outside reasonable local hours or on days the customer's own pattern suggests are closed, such as Fridays, Sundays or holidays.
- A clean exit. A single line lets the customer say "not now" or "stop," and the business honours it permanently.
If any one fails, Dr. Tebaa says, the AI should stay silent and the lead should go to a person or be left alone. Persistence is deliberately absent from the list.
The composite, sorted three ways
In his worked example, the supplier replaces the timer with a sort. All 71 quiet leads are checked against the four conditions. Nineteen earn a single message because each has a reason and a new fact. Eleven are warm or large enough that a salesperson receives a call task with the context written out, and the AI sends nothing. Forty-one are left alone. The arithmetic closes at 71.
The composite result is 12 replies and 5 orders, against 9 replies and 1 order under the old setting, with far fewer messages sent. Dr. Tebaa is careful to say these are illustrative figures and not a promise. His claim is narrower: when every message carries a reason and a fact, each one is worth the customer's attention.
Where a person takes over
He wants two cases decided by the owner in advance and in writing. The first is the warm, high-value lead: above a value the owner sets, a person calls, because an automated message to someone about to spend serious money can feel like being processed. The second is the lead who has asked for time. If a customer says "after Eid" or "when the shipment clears," the AI records that date and stays silent until it arrives.
What owners should measure
Dr. Tebaa cautions against judging follow-up by reply count, since a "stop" is also a reply. He recommends counting, per 100 messages sent, the positive replies, the stop replies and opt-outs, and the orders. In the composite, the old setting returned about 4 replies per 100 messages, and eight of the nine were stop replies or opt-outs, roughly 3.8 per 100. The owner had been reading engagement where there was complaint.
The stop-reply rate, in his framing, is the guardrail. If stop replies and opt-outs together exceed a few per 100, the rules should be tightened before anything else is examined. He suggests each business set its own line in advance and write it down.
The supplier and figures in this article are an illustrative composite from Dr. Tebaa's work, not a real client or a measured result.
Related evidence: The UK Information Commissioner's Office says a business may send marketing emails or texts to an individual only with their specific consent, or to an existing customer who negotiated a similar purchase and was given a simple way to opt out when their details were collected and in every message since. (the UK ICO's guidance on when a business may message an individual unprompted)
Google's email sender guidelines advise sending only to people who want to hear from you, and warn that if a sender's messages are frequently reported as spam, its future messages are more likely to be marked as spam, lowering the sender's reputation over time. (Google's sender guidelines on how unwanted messages damage future deliverability)
