How a Restaurant Owner Monitors What People Say About Her Business Online

It started with someone else’s bad day.

Maria runs a farm-to-table place in the suburbs of Denver. One night she was scrolling Reddit and stumbled into a thread about a restaurant across town. The post was brutal. Dozens of people piling on about cold food, rude hosts, and long waits. Someone who claimed to work there jumped in and started airing out how toxic it was behind the scenes. Management drama. Tip skimming. The whole thing.

The thread had thousands of upvotes. It was the top post in the city’s subreddit for two days.

Maria did not know anyone at that restaurant. But she sat with the thread for a while, and the thing that stuck with her was not the drama itself. It was the fact that the owner probably had no idea it was happening.

That is when the question hit her: what are people saying about my restaurant right now?

She had no way to know

Maria checks her Google reviews. She reads what comes in on Yelp. She looks at her Instagram comments. But that is all inbound. People tagging her, leaving reviews on her pages, writing to her directly.

The Reddit thread she found was not on anyone’s review page. It was a conversation happening in public, about a business, without the business knowing. And Reddit is just one place. People talk on Facebook groups, X, Instagram stories, local forums, Nextdoor, food blogs, and random threads she would never think to check.

There is no notification for that. Nobody tags you when they complain about your restaurant in a local subreddit. Nobody sends you a link when someone raves about your brunch in a Facebook group you have never heard of.

The information exists. It is just scattered across the internet in places you would have to manually check every single day to catch.

Nobody has time for that.

What she actually wanted

Maria did not need a social media dashboard. She did not need analytics or sentiment scores or engagement metrics.

She wanted something simple: if someone mentions my restaurant online, I want to know about it.

Good mentions, bad mentions, whatever. She just wanted to be in the loop.

The Reddit thread made it obvious that the worst version of this problem is finding out weeks later, or never finding out at all. A bad review you can respond to. A complaint you never see just festers.

And positive mentions are just as valuable. If someone posts a photo of your food and raves about it to 500 people, that is worth knowing too. You can thank them. Repost it. Reach out. At a minimum, you know what is landing with people.

She found BotHound and set it up in an afternoon

Maria is not technical. She does not write code. But the setup was straightforward.

She built a bot with two tasks:

1. Search the web for any mention of her restaurant name, her own name, and a few menu items people tend to talk about.
2. Email her a summary of everything found, with links to the original posts.

She attached her business context as a file so the bot understood what to look for and how to evaluate what it found.

Then she put it on a daily schedule. Every morning at 6:00 AM, the bot runs. By the time she gets to the restaurant, the report is in her inbox.

The business context file

This is the part that makes the bot useful instead of noisy.

Without context, a web search for a restaurant name will pull up your own website, your Yelp page, old press mentions, and a bunch of irrelevant results. The bot needs to know what it is actually looking for.

Maria’s context file tells the bot:

• The restaurant name and any common misspellings or abbreviations people use
• Her name and the names of a few key staff members
• The city and neighborhood
• What platforms to prioritize: Reddit, Facebook groups, X, Instagram, Nextdoor, local food blogs
• What counts as a mention worth reporting: someone talking about the restaurant, recommending it, complaining about it, posting a photo, asking for opinions
• What to ignore: the restaurant’s own social media posts, old press from more than 6 months ago, results that are clearly about a different business with a similar name

That last point matters. If your restaurant is called “The Kitchen” you are going to get a lot of noise without guardrails.

The bot soul

The bot soul sets the tone for how the bot approaches the work. Maria’s looks roughly like this:

You are a brand monitoring assistant for a local restaurant.

Your job is to find any public mention of the restaurant across the web, evaluate whether it is relevant, and report what you find.

You care about:
- Customer reviews and feedback on any platform
- Social media posts that mention or tag the restaurant
- Reddit threads, Facebook group posts, or forum discussions where the restaurant comes up
- Blog posts, local press, or food coverage
- Any mention of staff by name in a public context

You are thorough but not paranoid. You report real mentions, not false positives. When you find something, you include the source link, the platform, a short summary of what was said, and whether the tone is positive, negative, mixed, or neutral.

You do not editorialize. You report what you find so the owner can decide what to do about it.

Short, direct, and clear about what counts and what does not.

Task 1: Search for mentions

The first task does the actual searching.

Tool attached: WebSearch

The prompt tells the bot to search across multiple platforms for recent mentions. It looks for the restaurant name, the owner’s name, and a handful of signature dishes and menu terms that people tend to reference.

The key instruction is to cast a wide net but filter hard. A lot of web searches for a restaurant name return the restaurant’s own pages. Those are not useful. The bot is trained to skip self-published content and focus on what other people are saying.

It returns a list of mentions, each with:

• Platform (Reddit, Facebook, X, Instagram, blog, etc.)
• Link to the original post or discussion
• Date, or approximate date if not available
• Short summary of what was said
• Tone: positive, negative, neutral, mixed
• Whether a response from the restaurant might be appropriate

That last field is the one Maria checks first.

Task 2: Email the report

The second task takes everything from the first task and packages it into a clean email.

Tool attached: SendEmail

The email has a subject line that includes the date and a count of mentions found. The body is a simple list, sorted by tone. Negative mentions come first, because those are the ones she wants to see immediately. Positive mentions follow. Neutral ones at the bottom.

Each mention includes the link, so she can click through and see the full context herself.

On days when nothing is found, the bot sends a short email that says so. Maria says that is actually reassuring. No news is good news, and now she knows it for a fact instead of just hoping.

What she found in the first two weeks

The first week, the bot surfaced a Nextdoor thread where someone asked for restaurant recommendations in her area. Three people had mentioned her place. She had no idea the thread existed.

She also found a one-star Google review she had somehow missed. It had been sitting there for four days. She responded to it that morning.

The second week, the bot caught an Instagram story from a local food blogger who had eaten at her restaurant over the weekend. The blogger had 12,000 followers. Maria reached out, thanked her, and they ended up collaborating on a small event the following month.

None of those things were hard to act on. The hard part was knowing they existed.

Why she keeps it running

Maria told us the value is not really about catching a crisis. It is about staying connected to what people actually think.

When you run a restaurant, you get a very filtered version of feedback. People who leave reviews are either thrilled or furious. The people in between just eat and leave. But those middle-ground opinions show up in casual conversations online. Someone asks “where should I eat tonight?” and your restaurant does or does not come up. Someone posts a photo of a dish and says “this was fine, nothing special.” That kind of signal is invisible unless you go looking for it.

The bot goes looking for it every day. Maria does not have to.

She compared it to having a friend who reads every local food thread and sends you the highlights before you wake up. That is basically what it is. Except the friend never forgets and never takes a day off.

The pattern behind this

This is not really a restaurant story. The same setup works for any local business.

A dentist, a gym owner, a boutique, a law firm, a landscaping company. If you have a business with a name and customers who talk online, this exact bot works for you. The only thing that changes is the context file and the search terms.

The underlying need is always the same: people are talking about your business in places you are not checking, and you have no way to know unless someone tells you or you go look.

Most business owners do not go look. Not because they do not care, but because there is no time and no system for it.

What it actually costs

Maria’s bot has two tasks. At 50 credits per task, each run costs 100 credits. That is one dollar a day.

Thirty dollars a month for daily brand monitoring across the entire public web.

She said the Nextdoor thread alone, where three people recommended her restaurant to someone actively looking for a place to eat, was worth more than a year of that cost.

The real lesson from that Reddit thread

The restaurant that got roasted on Reddit probably never saw it. Or they saw it weeks later, after thousands of people had already formed an opinion.

That is the thing about online reputation. It is not something you build in one place. It is something that forms in dozens of small conversations you do not control and usually cannot see.

You cannot control what people say. But you can make sure you hear it.

That is all Maria wanted. And it took one bot, two tasks, and a daily schedule to get there.