AI Sentiment Analysis
Most owners read their last 5–10 reviews and call it a feel for the business. AI Sentiment Analysis reads all of your reviews and tells you, in plain English, what customers actually love and what's quietly hurting you.
Where to find it
Open the dashboard and stay on the Home page. The right column shows three panels: Feedback Summary, What's Going Well, and the negative-sentiment companion below it. Together they're your weekly "what would I want to know if I only had two minutes?"
The three panels, and what each one is for
Feedback Summary
A short paragraph synthesizing your entire review history into the honest version of your customer experience. It calls out what works, what's inconsistent, and which operational issues show up across reviews — even ones a casual read would miss.
Use it to: Brief a new manager. Share with your team. Quote in an investor or franchise application. It's the most concise honest summary of your business that exists anywhere.
What's Going Well
A ranked list of the specific things customers consistently rave about — products, people, moments, attributes. Each item shows how many reviews mention it.
Use it to: Inform your marketing copy. If "fresh, hot donuts" is the #1 thing customers mention, that's exactly what should be in your headlines, social posts, and Google Business Profile description. Stop guessing what your USP is — your customers already told you.
What Needs Attention (negative themes)
The mirror image: the recurring issues that drag your rating down. Inconsistent product quality, specific staff interactions, a policy that frustrates people, slow service at certain times — whatever's showing up across multiple reviews.
Use it to: Prioritize one fix per month. The themes are ranked by frequency, so the top item is genuinely the highest-leverage operational problem to solve. Solve it, watch your rating climb.
How it actually works
Every time a new review lands, we re-run the analysis across your full review history (Google + any other platforms you've connected). The model groups reviews by topic, scores sentiment, and surfaces themes that appear in 3+ reviews. Single-review issues stay out of the summary so you're not chasing noise.
Counts and themes refresh within minutes of a new review hitting your inbox.
The most-undervalued feature in the product. Most owners check ratings and reply to recent reviews — fine, but those are lagging indicators. The sentiment panels are the only place you can read the long-term truth about your business in 30 seconds. We strongly recommend opening the Home page every Monday morning.
Common questions
Does it use my reviews to train an AI model that other businesses see?
No. The analysis runs on your data for your account only. Your reviews are not used to train shared models, and the themes for your business stay in your account.
Can I export it?
The Feedback Summary text is copyable directly from the panel. Selecting and copying works the same as any text on the page.
How many reviews do I need before it's useful?
The model needs around 15–20 reviews to produce reliable themes. Below that you'll see lighter output (or a "not enough data yet" state). Once you cross 50 reviews the analysis becomes genuinely sharp.
What if a theme is wrong or misleading?
Reply to anything we've miscategorized via anthony@newreview.co and we'll improve the model. We do not allow themes to be manually edited (that would defeat the point of an honest mirror), but the analysis re-runs continuously, so a misread will usually correct itself once more reviews come in.