How to analyse Trustpilot and Yelp reviews in one place

Most businesses collect feedback in more than one place without meaning to. An online shop gets Trustpilot reviews after delivery. A restaurant or salon picks up Yelp reviews from people searching nearby. Both may also run a post-purchase survey, and a few customers email in. Each source gets read on its own, by a different person, on a different day.

The result is several partial pictures. This guide explains how to analyse Trustpilot and Yelp reviews properly, and how to put them together with your own survey replies so you can see what customers are telling you as a whole.

Each platform hears from a different crowd

Before combining anything, accept that each source has its own audience and its own habits. None of them is a neutral sample of your customers.

This is not a criticism of any platform. It simply means that a topic can look large on one source and small on another for reasons that have nothing to do with your business. Delivery complaints will dominate an online shop's Trustpilot reviews partly because that is what people go there to talk about.

The point of combining sources is not to average away these differences. It is to see each topic from several angles, and to notice when sources agree.

Why combine reviews with your own survey replies

Public reviews have two weaknesses as evidence. They are written by people who chose to write, who are often the very pleased and the very annoyed. And they are written about whatever the reviewer wanted to say, not about the questions you need answered.

Your own survey replies fix both. You invite everyone, so you hear from the middle. You can ask a direct question ("What is one thing we could do better?") and get an answer about it. A survey with a would you recommend us question and one open text box is enough to start.

Reviews, in turn, fix the survey's weakness: people say things in public, in their own words, that they would never put in a form. Used together, each corrects the other.

Step 1: Bring everything into one table

Whatever tool you use, the goal is a single list where every piece of feedback has the same fields:

• Source: Trustpilot, Yelp, survey, widget, email.
• Date: when it was written, or when the order or visit happened if you have it.
• Rating: if there is one, kept on its original scale for now.
• Text: the full words, plus any title.
• Location or product: whatever lets you split the results later.

Keep the source on every row. You will want to filter by it constantly.

Step 2: Put ratings on a common footing, carefully

Trustpilot and Yelp both use five stars. Survey questions might use 0 to 10, 1 to 5, or a thumbs up and down. You can convert them, but be careful about what you are claiming.

A simple and honest approach is to group every rating into three bands rather than converting to a single number:

Bands are easier to compare across sources than averages, and they do not pretend a 3 on Yelp means exactly the same as a 6 on your survey. Even so, compare a source with itself over time before you compare it with another source.

Step 3: Use one set of topics across every source

This is the step that makes the combination worthwhile. Tag every mention in every source against the same list of topics: delivery, product quality, returns, customer service, prices, and so on for a shop; food, staff, waiting time, atmosphere, booking, prices for a restaurant.

Let topics nest so you can get specific (delivery → late, delivery → damaged packaging) and let one review carry several topics. Give each mention its own tone, positive, negative or mixed, rather than one tone for the whole review. Our guides on analysing open-ended survey responses and sentiment analysis cover both steps in detail.

If you tag Trustpilot with one set of words and your survey with another, you will never be able to line them up. One list, used everywhere, is the rule.

Step 4: Compare share and tone, source by source

With every source tagged the same way, build a simple grid: topics down the side, sources across the top, and in each cell the share of that source's feedback that mentions the topic, plus whether those mentions are mostly positive or negative.

Use share, not counts. If Yelp sends you thirty reviews a month and your survey sends three hundred replies, raw counts will make the survey look like it cares about everything more. Share of conversation (the percentage of feedback in a source that mentions a topic) puts them on the same footing.

Then look for three patterns:

• Agreement. A topic that is negative across Trustpilot, Yelp and your survey is almost certainly real and worth fixing first.
• Only in public. A topic that shows up in reviews but barely in your survey may be something people only bother to mention when they are upset, or something your survey questions steer around.
• Only in private. A topic that appears in your survey but not in reviews is often a quieter, everyday friction that never becomes a one-star review but still wears customers down.

Step 5: Watch the change, not just the snapshot

Once a month, compare each topic's share and tone with the previous month. A hypothetical online shop might see delivery → late rise from a small share to a noticeably larger one on Trustpilot while staying flat in its own survey. That points to a problem with a particular courier or region that affects people who review, and it is a lead worth chasing.

Volume across sources will move independently. A Trustpilot invitation campaign can double your Trustpilot reviews overnight. Measuring share keeps that from looking like a sudden rise in every topic.

Step 6: Decide, act and check

Rank negative topics by how many customers they touch, how strongly they appear in unhappy feedback, and whether they are getting worse. Give the top few an owner and a date. Then keep watching the same topics after the fix, across every source, to see whether the change landed.

Reply to public reviews as you go, especially the negative ones. Our guide on responding to negative reviews has a structure for that.

Common mistakes to avoid

• Treating one platform as the truth. Each one is a slice. Agreement across slices is what you are looking for.
• Averaging stars across sources. It hides the differences that tell you the most.
• Different topic lists per source. It makes comparison impossible.
• Counting mentions. Volume changes for reasons unrelated to quality; use share.
• Reading only the extremes. Three-star reviews and middling survey replies often hold the most specific suggestions.

Doing this with Userforms

On the Plus plan and above, Userforms imports Trustpilot, Yelp, Google and TripAdvisor reviews, and you can add CSV uploads or pasted text from anywhere else. Your own forms and website widget feed into the same library, so every review and reply sits together with its source attached.

Each piece of feedback is read as it arrives and filed into one shared topic hierarchy, with a tone for every mention. The What changed briefing shows which topics grew or shrank in share of conversation and which turned better or worse, and every chart opens onto the replies and reviews behind it. You can also ask plain-language questions, such as what customers say about delivery on Trustpilot compared with the survey, and see the replies the answer relied on.

You can start for free with your own forms, or see which plans include review imports on the pricing page.