The star average is the weakest number on the page
Sorting contractors by star rating feels obvious and is almost always wrong. It is the number every platform puts in the biggest font, and it is the number that carries the least information. This is the reasoning True Level Media built its directories on, and it works just as well by hand on any review platform.
Consider two Connecticut painters. One has six reviews, all five stars, average 5.0. The other has two hundred and eleven reviews, average 4.5. Sorted by average, the first one wins. Read as evidence, the second one wins easily: two hundred people describing consistent work is a far stronger claim than six people describing six jobs, and six perfect jobs is a completely normal run for a business that has done a dozen.
This is the reasoning our ranking method formalizes with a Bayesian adjustment, and you can apply it by hand in about ten seconds: ask how many people are actually speaking before you look at what they said.
Volume and sentiment answer different questions
Keep them apart in your head.
Volume answers: how many people have dealt with this business and said something publicly? That is a question about consistency and about how long a business has been operating in the open.
Sentiment answers: of the people who spoke, how satisfied were they, and about what?
Rolling both into a single star average destroys the difference. A business with forty reviews at 4.6 and a business with four reviews at 4.6 are not the same proposition, and the one number they share tells you nothing about which is which.
What recency tells you
Trades change faster than their review histories suggest. Crews turn over, a good foreman leaves, an owner sells the business and keeps the name. A wall of excellent reviews from three or four years ago, with almost nothing since, is a real signal, and it is not a reassuring one.
Sort by most recent and read the last twelve months first. If the recent reviews look different from the older ones in either direction, that difference is the most useful thing on the page.
The reverse case matters too. A business with a sudden burst of reviews in a single month, and little before or after, is usually reflecting one event: a storm season, a promotion, or a campaign asking every past customer at once. Weigh it accordingly.
Reviews about a sales visit are not reviews about work
This is the trap that catches careful people, and it is most common in trades sold through in-home appointments: windows, doors, roofing, remodeling.
A review describing a sales visit reads like this: the rep was punctual, the presentation was thorough, the price was explained, the discount expired that evening. Every word of it may be true, and none of it is evidence about installation, because no installation happened.
Learn the vocabulary. Reviews about quotes, appointments, pressure, financing, and scheduling describe a sales process. Reviews about the finished result, cleanup, callbacks, and how it held up in the rain describe work. When a business has a lot of the former and few of the latter, that pattern is itself the finding.
A quick test
Read ten reviews and count how many describe a completed job you could inspect. If the answer is two, you have not learned much about the work, however many stars are attached.
Read the critical reviews first
Most people skim the five-star reviews and stop. The useful reading is the other direction: open the one, two, and three-star reviews first, because that is where the specifics live.
You are not looking for the existence of complaints. Every business that has done enough work has some. You are looking for two things. The first is whether the same complaint keeps recurring: three separate people describing a job left unfinished over a winter is a pattern, and one person describing a scheduling clash is a Tuesday. The second is how the business responded. A short, factual reply that acknowledges a problem and says what was done about it tells you something about how you would be treated if something went wrong on your job. A defensive reply arguing with the customer tells you something too.
Then check the dates on those critical reviews against the dates on the good ones. Complaints clustered in one period, followed by a year of steady positive reviews, often means something changed for the better. Complaints that are the most recent thing on the page mean the opposite.
What a strong review record looks like
- Enough reviews that a handful of outliers cannot move the picture.
- A steady flow over time rather than one dense cluster.
- Specific detail: what the job was, how long it took, what went wrong and how it was handled.
- Some critical reviews, answered like an adult. A business with two hundred reviews and no complaints is either extraordinary or curated.
- Descriptions of the parts you cannot see for yourself, which is where trade work actually succeeds or fails.
Reviews are half the job
Even a perfectly read review record does not tell you whether a contractor is registered to do residential work in Connecticut. That is a separate, factual check, it is free, and it takes two minutes at the state source: check a contractor’s registration walks through it.
Do both. Confirm the facts that are matters of public record, then weigh the evidence that is a matter of judgment. Our Connecticut directories run the second half of that at scale, using the same reasoning set out here, which is why the method is published rather than described as proprietary.