Run Google listings for 10 clients alone with AI agents
A one-person agency can run Google listings for 10 local clients by giving each job to an AI agent: review replies, weekly posts, listing audits, and a monthly report that reads each client's reviews and their competitors' reviews. You write one file per client, reuse the same free prompts for all of them, and approve everything before it goes live.
If you run a marketing agency on your own, Google listings are some of the most useful work you sell. Local businesses need them, they're easy to explain, and the results show up in calls and bookings. The catch is that the work is weekly, and it's weekly for every client. Reply to reviews, post, check hours, check the menu, write a report. Do that for 10 clients by hand and there's no time left to find client 11.
ZeroCrew built this use case for one-person agencies and freelancers who manage Google Business Profiles for local clients: restaurants, cafes, salons, dentists, gyms, and shops. The idea is simple to say. You keep the client relationships and the judgment. AI agents do the reading, the checking, and the first drafts, for every client, every week.
Google says what helps a listing. According to Google's Business Profile Help, local results are mainly based on relevance, distance, and popularity, and businesses with complete and accurate information are more likely to show up. The same page says replying to reviews shows customers you value their feedback, and that photos help tell a business's story. That's the work. This article gives you six jobs for an agent and a free prompt for each.
Why do solo agencies hit a ceiling on Google listings?
Solo agencies hit a ceiling because the work grows with every client and you're the only one doing it. One client takes a couple of hours a week. Ten clients take your whole week. The work that slips first is the work that makes clients stay: answering reviews on time and telling them something useful each month.
The clients' competitors are usually no better. Search for a client's kind of business in their neighborhood and look at the top 3. You often find old photos, unanswered reviews, and no posts for months. That's what you're selling: the business that does the boring things every week starts to stand out.
How do AI agents run Google listings for many clients?
AI agents run Google listings for many clients by doing the same jobs with a different client file each time. It works in three moves:
- Each client gets one file. You write it once: their services, hours, area, competitors, and how they talk.
- Each job gets one prompt. The same prompt works for every client. Only the client file changes.
- The agent drafts, and you approve. Nothing goes live until you read it. The agent does the legwork, and you make the calls.
You can run this in a general AI assistant such as Claude or ChatGPT. Keep one project or folder per client with their file inside, so every job starts with the right context. If your assistant can run tasks on a schedule, set the weekly jobs for Monday morning and the client reports for the first Monday of the month.
Step zero: write a file for each client
A client file is what turns generic AI output into output that sounds like that client. Write it during onboarding and paste it at the top of every job. Include:
- The business name, address, phone, and website, written exactly as they appear on Google.
- What they sell: services or menu, signature items, and price range.
- Their hours, including holidays.
- What they offer: bookings, delivery, parking, accessibility, dietary options, or whatever their customers ask about.
- Two or three true reasons customers pick them.
- Their top 3 competitors on Google Maps.
- How they talk: warm and casual, formal, or somewhere in between.
- Who approves what: which replies and posts the client wants to see first.
Update the file when something changes. A new menu or new hours goes in the file first, then on the listing.
Which 6 jobs can AI agents do for each client?
Each job below has a short prompt. Paste the client file first, then the prompt, and swap the words in capitals for that client's details. The examples use a restaurant client, because restaurants live and die by their reviews.
1. The monthly review intelligence report
This is the job that keeps clients paying. Once a month, the agent reads every new review for your client and for their top 3 competitors, and writes a one-page report on what customers actually say.
Reviews are the most honest market research a local business gets, and they're public. A competitor's customers say what they love there, what annoys them, and what they wish someone offered. Business owners don't have time to read all of it. Neither do you, for 10 clients. An agent does it in minutes.
The report answers six questions:
- What did the client's customers praise and complain about most this month?
- What do customers praise and complain about at each competitor?
- What is each business known for, in customers' own words?
- Does that match what the client thinks makes them different?
- What do customers ask for that nobody nearby does well?
- Are the client's rating and review count going up or down, and how do they compare?
Here's the kind of thing it finds for a restaurant client. Customers at the place across the street keep complaining about a 40-minute wait on weekends. Three of your client's reviews mention their quick lunch. Nobody nearby has a vegan main that customers rave about, and several reviews ask for one. That's three decisions for your client in one page, and a reason to keep you for another month.
The full prompt for this report is in the copy box at the end of this article.
2. The prospect audit: win the next client
Ask: "Search Google Maps for PROSPECT'S TYPE OF BUSINESS in THEIR NEIGHBORHOOD. Compare PROSPECT NAME with the top 3: categories, photos, attributes, booking or ordering links, how recent their reviews are, whether they reply, and whether they post. List every gap, most important first, in plain words for the owner."
Run it before a sales call and bring the gaps with you. A one-page list of what a business is missing next to its competitors starts a better conversation than a pitch. Once they sign, run the same audit every quarter.
3. The review agent: reply to every review
Ask: "Here are this week's new reviews for CLIENT NAME: PASTE REVIEWS. Draft a reply to each in the client's voice. Thank happy customers and mention what they wrote about. For unhappy customers, apologise once, don't argue, and invite them to contact the business at CLIENT EMAIL OR PHONE."
Read each reply and send it to the client for approval if they asked to see them. A reply that names the dish or the service shows a real person read the review.
4. The weekly post agent: one post per client
Ask: "Here's what's new at CLIENT NAME this week: SPECIALS, EVENTS, OR NEWS. Write one Google post about the best one. Under 150 words, plain words, and end with one call to action like booking or calling."
Ask each client for one real photo a week. One honest post with a real photo beats a burst of ten and then silence.
5. The details check: make each client's information match everywhere
Ask: "Search for CLIENT NAME on Google, Apple Maps, Yelp, Facebook, and any booking or delivery apps they use. Compare the name, address, phone, hours, and menu or services on each with the client file. List every place where they don't match."
Fix each mismatch in that site's settings. Wrong holiday hours on one site are enough for a one-star review, so run this before every holiday for every client.
6. The customer questions agent: answer what people ask before they buy
Ask: "Read CLIENT NAME's reviews and their competitors' reviews and list the 10 questions customers ask or wish they'd known, such as parking, waiting times, kids, or booking for groups. Draft a short, true answer to each using only the client file."
Put the answers on the client's website and in their listing's details. If the agent can't answer from the file, that's a question for the client, not a guess to make.
What should AI agents never do with a client's listing?
AI agents should never write reviews, reward reviews, or pick which customers get asked. Google's review policy forbids fake reviews, offering incentives such as discounts or free items for a review, and selectively asking happy customers for positive reviews. Breaking those rules can get reviews removed or a client's listing restricted, and it's your name on the work.
Keep three more rules:
- Approve everything. The agent drafts. You, or the client, read and post.
- Use real facts only. No made-up products, prices, or awards. Google's guidelines ask that a listing represents the business as it really is.
- Don't stuff keywords into a business name. Use the client's real name, exactly as it appears on their sign.
How do you show clients it's working?
Check each client's Business Profile performance once a month and write down three numbers: calls, direction requests, and website or booking clicks. Add the rating and review count from the review report. Put all of it on the first page of the monthly report. Reviews, posts, and photos compound over weeks, so show the trend over two or three months, not one week.
What stays yours
Your clients, your judgment, and the relationship stay yours. The agents read the reviews, find the gaps, write the drafts, and remember to do it every week for every client. You decide what's true, what goes live, and what to tell each owner. That's what they pay you for.
To find the right agent for every other job in your agency, browse the agent directory.
Copy the prompt
You are the review intelligence agent for {client name}, a {type of business} in {neighborhood, city}. I run their Google listing as their agency.
Read the client file first:
- What they sell and their price range: {services or menu, price range}
- What they think makes them different: {two or three reasons}
- Their top 3 competitors on Google Maps: {names}
On the first Monday of every month, read every Google review from the last 30 days for the client and for each competitor. If you can't open all of them, ask me to paste them.
Then write a one-page client report with:
1. The client: what customers praised most and complained about most, with the number of reviews behind each point and two short quotes.
2. Each competitor: the same, in three lines each.
3. Differentiators: what each business is known for in customers' own words, and whether that matches what the client thinks makes them different.
4. Gaps: anything customers ask for that none of them offer well, such as a product, an hour, a service, or a price point.
5. Trend: the client's average rating and review count this month versus last month, and the same for each competitor.
6. Actions: one thing to fix, one thing to tell customers more about, and one idea to test next month.
Write it in plain words for a business owner, not a marketer. Use only what reviews actually say, and quote them. Don't guess or invent numbers. Never write reviews and never suggest rewarding customers for reviews.The prompt below writes the monthly review intelligence report for one client. Fill in the parts in curly brackets from the client file, paste it into your AI assistant, and run it for each client on the first Monday of every month. It's free to use and change.
Sources (5)
- Tips to improve your local ranking on Google, Google Business Profile Help
- Manage customer reviews, Google Business Profile Help
- Create and manage posts on your Business Profile, Google Business Profile Help
- Prohibited and restricted content, Maps User Generated Content Policy Help
- Guidelines for representing your business on Google, Google Business Profile Help
Questions founders ask
How many clients can one person manage with AI agents?
It depends on how much each client needs, but the weekly work per client drops from hours to minutes once agents draft it. The limit moves from doing the work to reviewing it and talking to clients. Start with the clients you have, time how long each one takes for a month, and add clients when the numbers say you can.
Can an AI agent read a client's competitors' Google reviews?
Reviews on Google Maps are public, so an AI assistant that can browse the web can read them. Some assistants can't load every review on the page. If yours can't, copy the newest 20 or 30 reviews for each business and paste them into the prompt. The analysis works the same way.
Is it against Google's rules to reply to reviews with AI for a client?
Replying with help from AI is fine when the reply is honest and the client agrees to it. Google's review policy forbids fake reviews, rewarding customers for reviews, and asking only happy customers for one. Agree with each client who approves replies, and read every draft before it goes out.
Do I need to pay for this?
No. The prompts in this article are free and work in a general AI assistant such as Claude or ChatGPT. Some assistants charge for higher limits or for running tasks on a schedule, which matters once you run jobs for many clients, but you can start on a free plan.
What should I charge clients for this?
Charge for the outcome and the report, not the hours. Your clients pay for a listing that's complete, reviews that get answered, and a monthly page that tells them what customers say about them and their competitors. Agents lower your cost to deliver it, which raises your margin rather than your client's price.
Zero employees. Never build alone.