If you sell a product that fixes a specific problem, Trustpilot reviews are one of the few places where customers of other companies describe that problem in their own words, in public, with a date attached. A customer writing "I waited three weeks for a refund and nobody answered my emails" is telling you two things at once. The company has a refund process problem, and the company has a support response problem. If you sell refund automation or support tooling, that review is a lead, and the company name is right there at the top of the page.
The method is simple. Pick the complaint pattern that maps to your product. Collect recent reviews from companies in the segment you sell to. Filter for reviews that mention the pattern. Count how often it shows up per company and whether it is rising. Then approach the companies where the pattern is frequent and recent, with the complaint as your opening context. The rest of this article explains how to do each step, what Trustpilot actually exposes to a visitor who is not logged in, where the data misleads, and when you should just read the pages by hand instead of collecting anything.
Why complaints are a better signal than ratings
A star rating tells you that customers are unhappy. It does not tell you why. A one star review from someone who hated the product is useless to you if you sell shipping software. A three star review that says "product is great but the tracking link never worked" is exactly what you want.
That is why the useful unit here is the review text, not the TrustScore. Trustpilot explains on its own support pages how the TrustScore is calculated. It is a weighted average that gives more weight to recent reviews and uses a Bayesian starting point so that companies with few reviews do not swing to extremes. You can read the details in Trustpilot's explanation of the TrustScore at https://support.trustpilot.com/hc/en-us/articles/201748946-TrustScore-and-star-rating-explained. The important consequence for sales work is that the score smooths exactly the thing you are looking for. A sudden burst of complaints about one topic barely moves the score, but it is a strong buying signal.
So ignore the aggregate. Work with the individual reviews, their dates, and their text.
What a visitor who is not logged in can see
Trustpilot business profile pages are public. Without an account you can open a company page and read reviews, see the star rating per review, the review title, the review body, the date, the reviewer display name and country, the number of reviews that person has written, and whether the review is marked as verified or invited. You can filter by star rating, sort by most recent or most relevant, and in many locales search within the reviews of a single company for a keyword. You can also page through older reviews.
Company replies are public too. A company that answers every complaint with a templated apology is behaving differently from one that never answers at all, and both patterns are visible to anyone.
What you cannot see without logging in is almost everything on the business side. You do not see the company's internal analytics, which reviews came from invitations the company sent, the review invitation volume, or the business account tier. You do not see reviews that Trustpilot has removed or that are pending moderation. You do not see reviewer email addresses or any contact details, and you should not try to get them. The reviewer is not your lead. The company is.
Trustpilot also publishes a transparency section on each profile that shows whether the company is actively asking for reviews and whether it uses paid Trustpilot features. That is useful context. A company that pays for Trustpilot is already investing in its public reputation, which often means someone inside owns the problem you are about to raise.
Where the data is incomplete
Be honest with yourself about the gaps before you build a pipeline on top of this.
First, reviews are not a sample of all customers. People who write reviews are disproportionately either delighted or angry. A company with twenty complaints about late delivery might ship ten thousand orders a week. The complaint rate you compute from reviews is not the true defect rate. Treat it as a relative signal across companies, not an absolute measurement.
Second, Trustpilot removes reviews that break its guidelines, and companies can flag reviews for investigation. Trustpilot describes this process in its guidelines for reviewers at https://legal.trustpilot.com/for-reviewers/guidelines-for-reviewers. A review you saw last week may be gone this week. If you store review IDs and re-collect, expect some to disappear. That churn is normal and is not evidence of anything on its own.
Third, invited reviews skew positive. When a company sends review invitations after a purchase, more satisfied customers respond than would organically. Companies that do not invite tend to have a profile dominated by organic reviews, which lean negative. Comparing complaint frequency between an inviting company and a non inviting company is comparing two different populations.
Fourth, not every company is on Trustpilot in a meaningful way. Trustpilot creates profiles for domains even when the company has never claimed them, and many of those have a handful of reviews from years ago. A company with three reviews gives you nothing usable. Set a minimum review count before you draw conclusions.
Fifth, language and locale split the data. The same company can have separate review streams on different Trustpilot country sites. If you only collect from one locale you miss the rest.
Building the complaint pattern
Start from your own product, not from the data. Write down the three to five problems you solve, then write down how a frustrated customer would describe each one without knowing your product exists. Customers do not write "lack of order status visibility." They write "no idea where my parcel is." They do not write "slow refund processing." They write "still waiting for my money back."
Build a small vocabulary per problem. Include misspellings and the obvious synonyms. Keep it short. Twenty phrases that are clearly about your problem beat two hundred that catch everything.
Then decide on a time window. For sales purposes the last ninety days is usually the right default. Complaints from two years ago may describe a problem the company has already fixed. Recency is the whole point.
Collecting the reviews
You have three realistic options, and the first two are free.
Read by hand. If your target list is under about thirty companies, open each profile, sort by most recent, use the one star and two star filters, and read. This takes an afternoon and you will notice things a keyword filter would miss, like a complaint that is phrased in a way you did not anticipate. For a short list this is the better choice, not a fallback.
Use Trustpilot's own API. Trustpilot offers a Business API for companies that have a business account, documented at https://developers.trustpilot.com/. It is designed for a business to pull its own reviews and manage invitations, not for pulling reviews about other companies. If you are a Trustpilot business customer and you want to analyse your own complaint patterns, this is the correct tool and it is already included in what you pay. It will not help you research prospects.
Collect the public pages programmatically. For a target list in the hundreds, reading by hand stops being practical and the Business API does not cover other companies. At that point you need something that fetches the public profile pages and returns the review fields as structured data. This is where a paid tool earns its place. What you want out of it is a flat table with one row per review: company domain, review date, star rating, title, body, whether the company replied, and the reply date. Everything downstream is spreadsheet work.
Whatever you use, respect the site. Trustpilot's terms of use for consumers are at https://legal.trustpilot.com/for-reviewers/end-user-terms-and-conditions and set out what Trustpilot permits. Keep request volume modest, collect only public fields, and do not store or use reviewer personal data for outreach. The reviewer did not write to you.
Turning rows into a ranked list
Once you have the rows, the analysis is short.
For each company, count the reviews in your window that match any phrase in your vocabulary. Divide by the total reviews in the same window to get a complaint share. A company where one in five recent reviews mentions your problem is a much better prospect than one where it is one in two hundred, even if the second has more total complaints.
Then look at trend. Split the window in half and compare the complaint share in the recent half against the earlier half. A rising share means the problem is getting worse or the company has recently grown into it. Both are good moments to call.
Then look at company replies. If the company replies to complaints about your problem and the reply says some version of "we are working on this," someone inside has acknowledged the issue publicly. That person or their manager is your entry point. If the company never replies, you may be dealing with a business that does not monitor Trustpilot at all, which changes your outreach. You might be the one telling them.
Finally, read the top twenty matching reviews per shortlisted company yourself. Keyword matches lie. "Refund was quick, thanks" matches the word refund. A human read of the top matches catches this in minutes.
Using the signal without being creepy
The complaint gives you context, not permission. Quoting a specific review back to a prospect in a cold email is a bad idea. It reads as surveillance and it puts the prospect on the defensive. Describe the pattern instead. "I noticed a number of your recent public reviews mention tracking visibility" is a factual observation about public information. Pasting a customer's review into the email is not.
Keep the reviewer out of it entirely. Never name them, never contact them, never imply you know who they are.
Also check the obvious before you reach out. Does the company already have a solution in place? Is the complaint about a subcontractor rather than the company itself? A delivery complaint about a retailer is often really a complaint about the carrier. Your read of the top matching reviews will usually tell you.
When free is the better choice
To be direct about it: if your target list is short, read the pages yourself. If you want to study your own company's complaints, use the Trustpilot Business API you already pay for. If you need one quick check on one prospect before a call, open the profile and filter to one star reviews from the last month. None of those need a paid tool.
A paid collection tool is worth it when the list is long, when you want to repeat the analysis monthly to catch rising trends, or when you want to combine review data with other signals like job listings or filings. At that scale, hand collection costs more in hours than the tool costs in money.
If you are at that point, the Trustpilot Reviews Scraper on this site returns the public review fields described above as structured rows, priced per review collected, so a monthly refresh of a few hundred company profiles stays predictable and you only pay for what you actually pull.
Want the signal instead of the raw filings? Get a free report preview. Prefer the tool to the write-up? Browse all data feeds or connect the free MCP server.