You open a one-star review. You read the first sentence. Something tightens in your chest before you finish the second. By the third, you are already composing a response in your head that you will almost certainly regret sending.
Every hospitality operator knows this moment. Your brain’s Protector has just taken the wheel.
I have explained this in more detail in a webinar for TiCSA and David Olney and I explore it more broadly in our recent podcast episode: Working With Your Whole Brain Costs Peanuts.
This article is a summary of this approach and how an 80/20 rule applies in asking an AI tool like Claude to help shoulder the load.
Four characters, one response
Dr Jill Bolte Taylor’s Whole Brain Living framework describes four neurological characters operating inside each of us. Understanding which one is driving when you read a difficult review changes everything about how you respond.
Character 1, the Planner. Left-brain thinking. Wants facts, order, and accuracy. Knows exactly which room the guest stayed in and can recall the maintenance log from that week.
Character 2, the Protector. Left-brain emotion. Fires before you finish reading. Wants to defend, deflect, or disappear. Writes every response you have drafted and then deleted at 11pm.
Character 3, the Explorer. Right-brain emotion. Lives in sensation, delight, and the present moment. Most useful when you are responding to a glowing review or crafting a warm closing line.
Character 4, the Connector. Right-brain thinking. Sees the bigger picture. Recognises that the person who left the review is a human being with unmet needs. Allows you to respond with genuine empathy rather than strategic damage control.
A well-constructed review response draws on all four. It opens with Character 4, moves through Character 1, and closes with Character 3. Character 2 is acknowledged, thanked for its concern, and gently asked to sit this one out.
Easier said than done when your stomach is in knots.
This diagram from SequenceWiz does a great job of explaining them:
Where Claude actually carries the load
Here is where the 80/20 split takes place.
Claude does not simply generate a draft while you watch. Claude does the cognitive and emotional heavy lifting: reading the review carefully, classifying what kind of complaint it is, identifying the unique detail that will make the response feel personal, and then asking you a targeted set of questions before a single word of the response is written.
Those questions matter enormously. Was a complaint raised during the stay, or only in the review? Has your team investigated and found anything? Has anything changed in your process since? Which room, which date, which crew member was on?
Your 20 percent is two things. First, you answer those questions, supplying the ground-level truth only you possess. Second, once the draft arrives, you read it aloud, add the human detail no AI can supply, and adjust anything that does not quite sound like you.
The result is a response that is specific, grounded, and written in your voice. Without the knot in your stomach driving it.
Building your own Claude skill
The workflow above is not a one-off prompt. It is a reusable skill: a structured set of instructions that lives inside a Claude Project and runs every time you paste in a review.
Here is what makes a skill worth building.
Start with phases, not prose. A good skill breaks the task into numbered steps with clear instructions for each. Ours has seven: acknowledge the emotional reality, collect the review, classify it, gather context, mine for uniqueness, draft, and polish. Each phase has a specific job. Claude does not move to the next until the current one is complete.
Separate your instructions from your context. The skill file tells Claude how to think. Your Property Document tells Claude who you are. Brand voice, marketing promises, policies, recurring topics, example responses you are proud of: all of that lives in a separate document uploaded to your Project. When both are present, Claude skips the broad business context questions entirely and goes straight to the specifics of the review in front of you.
Mine for uniqueness before every draft. The single most important instruction in our skill forces Claude to identify something genuinely specific in each review before writing a word. The reviewer’s emotional register. What they praised despite the complaint. The scale of their inconvenience. A phrase from their own text, reflected back naturally. This is what separates a response that is technically different from one that actually feels personal.
Name the hard cases explicitly. Review blackmail, suspected competitors, repeated complaints with no operational fix, guests who were warm in person and scathing online: these scenarios need their own guidance because the default instinct handles them badly. A hard-cases reference file, linked from the main skill, means Claude approaches them with the right framework rather than the standard empathy-plus-facts template.
Build in the human handover. Every draft should end with a prompt that returns control to you. Here is the 80 percent. Here is what only you can add. Read it aloud. If it sounds like you, it is ready. This keeps the response human. It is not a courtesy. It is the point.
Start here if you want to build your own
To get Claude working this way, paste the following into a plain text file and upload it to a Claude Project as your starting skill. Then add a second file: your Property Document, with your brand voice, policies, recurring topics, and three to five example responses you are proud of. That combination is all you need.
You are a tourism and hospitality review response assistant. Follow this workflow every time a review is provided.
Phase 0: Before anything else, briefly acknowledge that reading a difficult review can be stressful. Keep this to two sentences. Then move on.
Phase 1: Ask for the full review text, the star rating, the platform, and the approximate date of stay. Do not proceed until you have the review and platform at minimum.
Phase 2: Silently classify the review as one of the following: genuine negative, mixed, glowing positive, discrepancy guest (warm in person, scathing online), vague or generic, or possible bad faith.
Phase 3: Before drafting anything, ask the operator a targeted set of clarifying questions based on the review type. For a negative review: do you know which room or area this relates to? Was a complaint raised during the stay? Has anything changed in your operations since? Is there anything about this guest or booking you recall? If a Property Document is present in this Project, skip general business questions and go straight to these situation-specific ones.
Phase 4: Identify one detail in this review that exists nowhere else. The reviewer’s emotional register, what they praised despite the complaint, the scale of their inconvenience, or a phrase from their own text. This detail must appear in the draft.
Phase 5: Draft a response in three parts. Open with genuine acknowledgment using Character 4 (empathy and connection). Address the substance with specific detail using Character 1 (facts and accuracy). Close with warmth and a forward invitation using Character 3 (delight and possibility). Mirror the brand voice from the Property Document if present.
Phase 6: Present the draft, then say: this is your 80 percent. Add the detail only you would know. Read it aloud. If it sounds like you, it is ready.
That prompt will not produce a perfect result on its first run. Refine it as you go. Add your hard cases. Build your Property Document carefully, particularly the section with your example responses. The skill improves as you teach it.
The complete set, ready to use
If you attended our workshop and would prefer to start with the fully developed skill documents rather than build from scratch, email us and we will send you the complete set: the skill file, the hard-cases reference, the property document guide, and the fillable property document template. All plain files, ready to upload. Email [email protected].
The response is the product now
In 2026, AI platforms read your review responses as source material when deciding which properties to recommend to travellers who are ready to book. Traffic referred by AI converts at roughly 14 percent, compared to under three percent from traditional search. A poorly handled response is a data quality problem as much as a customer service one.
Your responses are working for you around the clock, whether you are managing them or not.
The operators getting this right are not necessarily the ones with the most reviews. They are the ones whose responses are specific, honest, and written in a voice that sounds unmistakably like theirs. They use AI to remove the emotional friction. They use their own judgement to add the human detail no AI can supply.
Start with one review. Run the workflow. Read the draft aloud.
You will know immediately if it sounds like you.