AI Review Analysis

When a guest review lands in Vanio, an AI doesn't just store it — it reads it, breaks it down, and tells you what's actually wrong (or right) and what to do about it. Within seconds you get categorized issues, severity levels, suggested action items per role, sentiment per category, business-impact flags, and cleanliness-specific feedback you can hand to your housekeeper.

Quick reference: AI review analysis, review insights, review breakdown, review categorization, review sentiment, review issues, recommendation risk, cleanliness analysis, multi-unit issues, repeat problems, what's wrong with my reviews, why is my rating dropping — all live in the review itself in your dashboard inbox, generated automatically.

Where to find it

Open any review in your dashboard inbox (or click any review notification). Below the review text you'll see an AI Analysis section with the breakdown. It runs automatically the moment a review syncs in — no setup, no toggle.

What it gives you

For every review the AI produces a structured analysis with seven sections:

1. Summary

For long reviews (200+ characters), a 1-2 sentence summary at the top. Useful when you're catching up on a week's worth of reviews and don't have time to read every word.

2. Issues

A list of every distinct problem the guest mentioned. Each issue has:

• Category: amenity / maintenance / cleanliness / service / safety / communication / other
• Description: what specifically went wrong, in plain language
• Severity: critical / high / medium / low
• Affects multiple units?: true if the issue is likely a building-wide or property-group-wide problem (broken elevator, no hot water, pests) rather than a single-unit issue
• Action items: a to-do list for resolving the issue, each item assigned to a role (maintenance, housekeeping, host, property manager, customer service) with a priority (immediate, next 24h, this week, when possible)

So a review saying "the AC didn't work and the bathroom was dirty" produces TWO separate issues — one maintenance (high severity, assigned to maintenance, immediate) and one cleanliness (medium severity, assigned to housekeeping, next 24h) — each with its own action items.

3. Sentiment

The AI's read on the guest's overall feeling, plus per-category scores (1-5 for location, cleanliness, communication, value, accuracy) even if the channel didn't ask for them. This catches mismatches: a 5-star overall rating that's actually mixed sentiment is a leading indicator that the next review from the same guest type might be worse.

4. Guest Satisfaction

• Score (1-100): a single number the AI uses to rank reviews. Useful for sorting your inbox by "worst first" when you have a backlog.
• Likely to return: true/false. Based on tone, complaints, and stated intent.
• Recommendation risk: high / medium / low. How likely the guest is to actively warn other people about your property.

5. Business Impact

• Reputational risk: high / medium / low. The AI's read on whether this review is likely to damage your search ranking or future bookings.
• Potential revenue loss: true if the review describes something likely to repeat and cause future problems.
• Requires immediate action: true if there's a safety, health, or critical maintenance issue that can't wait.
• Affects multiple units: true if the issue likely shows up in other reviews from the same building or group.

6. Cleanliness Deep-Dive (when applicable)

If the review mentions cleaning, hygiene, tidiness, or anything related, the AI extracts a separate cleanliness section:

• Direct quotes from the guest about cleanliness
• Specific issues (e.g. "hair in shower", "dust on shelves", "stained sheets")
• Positives (e.g. "spotless bathroom", "fresh towels")
• Recommendations for the cleaning provider — actionable, specific
• Recommendations for management — process or policy improvements
• Overall assessment — a constructive paragraph you can copy-paste into a message to your housekeeper

Use this section to give your cleaning team feedback they can actually act on, instead of just telling them "the guest complained, do better."

Why this matters

Manual review reading at scale doesn't work. If you have 30 properties, 200+ reviews a month, and each review is 100-300 words, that's 30-60 minutes of reading per week just to keep up. And that's BEFORE you've categorized them, escalated the problems, or done anything about them.

The AI analysis cuts that to about 5 minutes:

• Sort by satisfaction score (worst first)
• Open the top 5
• Read the AI's summary + action items
• Click "create task" to dispatch the action items to your housekeeper, maintenance team, or yourself
• Move on

Issues that affect multiple units get auto-flagged so you spot building-wide problems (HVAC, elevator, pests) before they show up in five more reviews.

How it picks up multi-unit patterns

If your property is part of a Listing Group (Settings → Listing Groups), the AI knows about the other units in that group. When a review describes something that's almost certainly building-wide (broken elevator, no hot water, loud construction next door), it flags affectsGroup: true and includes the group context — group ID, unit count, location key.

This means a single review can prompt action across every property in the group at once. Useful when one guest at unit 3B mentions the elevator being out — you can react before guests at 5A, 6C, and 7B post the same complaint.

Where the analysis lives

Every analysis is stored alongside the review. You can:

• See it inline when you open the review in the dashboard
• Filter your reviews list by analysis fields (e.g. "show me all reviews with reputational risk = high")
• Export it via the CSV export on the reviews page
• Use it as a workflow trigger ("when a review comes in with severity ≥ high, notify me on Slack")

For deep-dive trend analysis, the per-property review dashboard rolls up the analysis fields into category trends — see Reviews Management for the dashboard side.

Common scenarios

"I have 50 properties and I want to know which ones have the worst trending issue right now." Reviews → Filter by date (last 30 days) → Sort by guest satisfaction score (low to high) → look at the issues categories. Patterns will jump out within seconds.

"I want to be alerted on Slack when a review comes in with critical severity." Workflows → Trigger: Review received → Condition: severity = critical → Action: Send Slack message. The AI analysis fields are available as workflow conditions.

"My cleaning team won't take vague feedback. I need specific items." Open the review → scroll to the Cleanliness Deep-Dive → copy the Recommendations for Provider section → paste into a message to your housekeeper. They'll get the exact issues and direct quotes, not a vague "be more careful."

"Why is my rating dropping at one specific property?" Open the property → Reviews tab → look at the most common issue categories from the last 90 days. The AI analysis aggregates them so you see "cleanliness mentioned in 8 of last 12 reviews, severity trending up" instead of having to read every review to spot the pattern.

"I want a summary of all the issues mentioned across all my properties this week." Reviews → Date range = last 7 days → Export. The CSV includes the AI analysis fields. Pivot in a spreadsheet by category and severity. Done in 30 seconds.

What it can't do (yet)

• The AI doesn't AUTO-FIX issues. It surfaces them and suggests action items. You (or your team) still have to act on them.
• It doesn't write the response to the review — that's a separate AI feature, see Reviews Management for the auto-reply side.
• It doesn't predict future ratings. The reputational-risk score is based on the current review, not a forecast.
• It analyzes English and major European languages well; less common languages get a more basic analysis.