From Buried in Bots to Back in Control: How an Independent Restaurant Took Charge of Its Social Comments

Client: The Anchor & Vine (independent restaurant, South West England)
Sector: Hospitality & Independent Restaurants
Service: AI Moderation Assistant (Social Media Comment Moderation)
Timeline: 4 week build and go-live, plus 30 days of monitoring and ongoing support

The Challenge

The Anchor & Vine had built a genuinely loyal following on Facebook and Instagram. Weekly specials, behind-the-scenes kitchen shots and event nights were pulling in steady engagement, and the comments section had become one of the busiest parts of the business' online presence.

Growth brought a problem the owners hadn't planned for. Every popular post attracted a wave of spam within minutes: fake giveaway accounts, crypto and investment bots, and the occasional obviously fake review. More frustrating still, a handful of local competitors and rogue traders had worked out that dropping a comment with a link to their own booking page underneath a busy Anchor & Vine post was an easy way to intercept diners who were actively looking for a table that night.

Comment moderation had always been handled informally by the owner and the assistant manager, checking in whenever they had a spare few minutes between services. That worked when volume was low. It stopped working once weekend posts were regularly pulling in hundreds of comments. Genuine enquiries got lost in the noise: a message asking about a table for eight on a Saturday night once sat unanswered for two days, buried beneath a run of promotional spam and a handful of unpleasant remarks left after a single bad review.

There was no record of what had been actioned, no consistency in how decisions were made, and no way to tell at a glance whether a comment had been dealt with or simply scrolled past. There was also a harder problem underneath the obvious one: not every negative comment was spam. Genuine criticism about a slow table or a disappointing dish needed to be seen and handled properly, not swept away alongside the bots. Deleting real feedback risked looking like the business was hiding from it.

The brief to Ready Made Digital was clear: strip out the spam and the competitor interference automatically, protect genuine enquiries so they get a timely response, keep a clean audit trail of every action taken, and make sure real criticism still reaches a human rather than disappearing.

The Solution

Ready Made Digital designed and built a dedicated AI Moderation Assistant for The Anchor & Vine's Facebook and Instagram channels, delivered through our standard project process from first meeting through to ongoing support.

1. Scoping

We started with a working session with the owner and the assistant manager, reviewing three months of historical comments across both channels. That gave us a real picture of the volume, the categories of unwanted content, and the tone of genuine customer interaction we needed to preserve.

2. Design

From that review we mapped a clear moderation policy: what gets hidden automatically, what gets flagged for human review, and what always stays visible. Spam, obvious bot activity and competitor links sit in the automatic bucket. Genuine criticism, ambiguous feedback and anything referencing an individual staff member sit in the human-review bucket by design.

3. Build

Our team built the AI Moderation Assistant on top of the Meta Graph API, wired into a private dashboard for the restaurant. Every incoming comment is scored against the policy in real time, with a clear reason recorded for every decision. Nothing is deleted; unwanted comments are hidden, which keeps a full record intact for the owner and complies with Meta's own guidelines.

4. Prototype

Before touching a single live comment, the assistant ran in shadow mode for a week. Every decision it would have taken was logged and reviewed with the owner, so anything the policy had missed could be tuned before it went anywhere near a real customer.

5. Live

Once the policy was signed off, the assistant went live across both channels. Spam and competitor interference started dropping out within minutes of each new post going up, and genuine enquiries stayed visible and prominent. Anything ambiguous was routed to a simple review queue the owner could clear in a few minutes over morning coffee.

6. Monitoring

For the first 30 days we monitored every decision the assistant took, meeting the owner weekly to review edge cases and refine the policy where the assistant had been too strict or too lenient. Two small policy tweaks were made in that window; no genuine enquiries were incorrectly hidden after week two.

7. Handover

At the end of the monitoring window, we handed over a short written playbook covering the dashboard, the review queue, and the escalation path for genuine criticism. Training took a single 45-minute session with the owner and the assistant manager.

8. Ongoing Support

The assistant now runs continuously in the background, with Ready Made Digital providing ongoing support, policy updates and platform maintenance as Meta's APIs evolve. The restaurant pays a predictable monthly fee and no longer thinks about comment moderation as a task.

The Results

What the Restaurant Said

“We were spending hours a week trying to keep our own comments section tidy, and still missing bookings because a real enquiry was buried under twenty bits of spam. Now the noise is gone, the real messages come straight to us, and we have a clear record of everything the system has done. It gave us back control of our own pages. — Owner, The Anchor & Vine”

Key Takeaways

Social comment sections are a real commercial channel

For independent hospitality businesses, the comments under a busy post are where bookings are won or lost. Leaving that channel to informal, ad-hoc moderation means real enquiries are missed and competitors are handed an open door.

Automation should protect feedback, not erase it

A good moderation policy hides spam and competitor interference automatically, and deliberately keeps genuine criticism visible so it can be handled properly. That distinction is the difference between a moderation tool and a reputation risk.

An audit trail matters

Every automated decision should be logged with a clear reason. That protects the business, keeps the owner in control, and makes it easy to tune the policy over time.

About Ready Made Digital

Ready Made Digital builds practical AI systems for UK service businesses, from lead management through Risponda to bespoke tools like the AI Moderation Assistant described above. To talk about a similar build for your business, arrange a discovery call or explore more work on the Ready Made Digital blog.