Yelp’s rules on auto-responders and AI replies

What Yelp's Leads Agreement actually says

Two clauses matter if you're thinking about automating Yelp replies. Yelp's Leads Agreement discourages auto-responders — the blast-a-template-at-everything kind — and it sets expectations around AI: output produced with AI should be human-reviewed, and AI use should be disclosed to the consumer.

Read plainly, that's not a ban on using AI. It's a ban on pretending a machine is a person, and on letting a machine talk to consumers unsupervised. The distinction matters, because a lot of tools in this space are built on exactly the pattern the agreement discourages.

Why a pure auto-responder is the risky option

An auto-responder fires the same message at every lead the moment it arrives. From Yelp's side that's a bad consumer experience: someone describing a specific job gets a generic reply that doesn't engage with what they wrote, and they can't tell whether a human ever saw it. From your side, the account carrying the risk is yours, not the vendor's. If Yelp takes issue with how replies are being sent from your profile, it's your advertising account in the conversation.

That's worth weighing carefully when a tool advertises fully hands-off replying as the headline feature. Hands-off is the thing the agreement specifically discourages.

Human-in-the-loop, in practice

The compliant shape is straightforward: the AI reads and drafts, a human decides, then it posts. Concretely, in Yelp Lead AI:

• New accounts default to approve mode. Every drafted reply is sent to you for a one-tap approval. Nothing posts until you tap.
• You can edit before approving. The draft is a starting point, not a fait accompli — approving sends exactly what's in front of you.
• Auto-posting is opt-in, and discloses. The Instant plan can post a qualified reply directly into the thread; when it does, it carries an AI-disclosure line rather than passing as hand-typed.
• Junk leads get no reply at all by default. Silence on a lead you'd never take is safer, and more honest, than a template that implies you're interested.

The trade-off is real and worth stating: approve mode means a reply waits on you. If you're on a roof for three hours, it waits three hours. Some businesses would rather have speed than supervision — that's a legitimate preference, and it's why the Instant option exists with disclosure attached.

Disclosure: what it looks like

Disclosure doesn't mean a legal paragraph. It means the consumer isn't misled about who or what they're talking to. In practice, a short line noting the reply was AI-assisted is enough to be straight with someone, and it costs you very little — most consumers are less bothered by an AI-drafted reply that answers their question than by a human-typed one that clearly didn't read it.

Questions worth asking any vendor

Whoever you end up using, these four answers tell you where the risk sits:

• Does anything post without me approving it? If yes, under what conditions, and can I turn it off?
• Is AI use disclosed when a reply is posted automatically? If not, ask why not.
• Does it reply to every lead, or does it decide? Replying to everything is the auto-responder pattern by another name.
• Whose account is at risk if Yelp objects? The honest answer is always yours — but a vendor that says so unprompted is telling you they've thought about it.

The honest caveat

Yelp writes and revises its own agreements, and it enforces them at its own discretion. Nothing here is legal advice, and no vendor — us included — can promise you an outcome with a platform we don't control. What a tool can do is stay on the defensible side of a rule that's written down: draft with AI, keep a human in the decision, disclose when the machine posts. Read Yelp's current Leads Agreement yourself before you automate anything; it's the document that actually governs your account.