What a score is built from
Every ranking on a True Level Media directory comes from public customer reviews. Not from a survey we ran, not from a panel, and not from anything a contractor told us about themselves. Public reviews are the only body of evidence about local trade work that a homeowner can go and check independently, which is the whole reason the method rests on them.
Five things are weighed:
- Sentiment. What reviews actually say, not just the star attached to them. A four-star review describing a problem that got fixed properly reads differently from a four-star review describing a job nobody wants to talk about.
- Star rating. The raw average, used as one input rather than as the answer.
- Volume. How many people have written at all. This is evidence about consistency, and it is weighed separately from sentiment because the two answer different questions.
- Recency. When the reviews were written. A contractor’s work three years ago is weaker evidence about their work this month, especially in a trade where crews change.
- Completeness. How much of a record exists to read: whether reviews describe the job, the timeline, and the outcome, or just say “great job”.
The Bayesian adjustment, in plain terms
Raw averages lie when the sample is small. A contractor with five reviews at a perfect rating has an average of five stars. A contractor with two hundred reviews averaging four and a half has a lower number and a far stronger case. Sorting by the raw average would put the first one on top, which is exactly backwards.
So the method applies a Bayesian adjustment. In practice it means every business starts from a neutral expectation, and its own reviews pull it away from that expectation in proportion to how many there are. Five reviews move a score a little. Two hundred move it a lot. Nothing is thrown away, and no minimum review count is imposed as a hard gate; the adjustment simply refuses to treat a thin record as if it were a thick one.
The short version
A small sample is not evidence in the way a large one is. Five perfect reviews does not outrank two hundred good ones, and any ranking that says otherwise is measuring luck.
The same logic runs in reverse. A single bad review on a business with three reviews total drags an average through the floor in a way that tells you very little. The adjustment keeps that in proportion too.
Per-trade criteria
The signals above are common to every directory. What sits on top of them is not. Each trade is scored against what customers of that trade actually write about.
A tree service is scored partly on how reviewers describe cleanup and property damage, because heavy equipment on a residential lot leaves a record and reviewers describe it in detail: ruts across a lawn, a cracked fence panel, a driveway swept clean. A flooring installer is not scored on any of that. Flooring reviews are read for what they say about the substrate, moisture testing, and whether a floor stayed flat. Roofing is weighted toward follow-through, because a homeowner usually cannot see the work and the honest signal is what happened on the callback.
That is why the per-trade notes appear on each directory brief rather than being buried in one universal formula. If you want to see how a specific trade is judged, the directory’s own page says so.
What is never used
This list is short and absolute:
- Payment of any kind. No contractor can buy a position, and none has.
- Advertising spend. Not with us, not anywhere else.
- Any commercial relationship. The ranking engine has no access to client status of any kind. Where a commercial relationship exists anywhere in the business, it is disclosed rather than hidden: see our editorial standards for the full statement.
If paid placement launches in future, it will be labeled as paid and sit outside the ranked list, and it will not affect a score. That commitment is part of the standards page, not a footnote here.
The monthly re-check and the depth floor
Directories are re-checked every month. New reviews come in, old reviews age, businesses close, and scores move accordingly. Nothing sits frozen from launch day, and a good month two years ago does not hold a position indefinitely.
The re-check is also how listings come off. A business that has closed drops out at the next cycle, and if you know of one before we do, an email gets it removed faster.
There is a second rule that decides whether a directory page exists at all. We call it the depth floor: a page is published only when there are enough real contractors with enough public review evidence to make the ranking useful. A list of three businesses, two of which have four reviews between them, is not a directory. It is a page pretending to be one, and building thousands of those is precisely the practice that makes trade search unusable. Where the floor is not met, we do not publish.
That is also the honest answer to why a particular contractor might be missing. It is not a judgment about their work. It usually means there is not yet enough public review evidence to place them fairly.
What a score is not
A position on one of our directories is an editorial judgment produced by applying a stated method to public evidence. It is worth being exact about what that does not amount to.
It is not a certification. We do not inspect work, audit accounts, or verify insurance, and a ranking should never be read as us vouching for a particular job.
It is not a statement about licensing or registration. Those are matters of public record that you can and should confirm yourself at the state source, and a high ranking does not substitute for that check.
It is not a prediction about your project. Reviews describe work already done, often for houses and budgets unlike yours. A strong record makes a good outcome more likely; it does not make it certain.
It is not permanent. Scores move monthly, and a position held in March is not a claim about June.
What a score does tell you is narrow and useful: of the contractors in this trade with enough public review evidence to judge, here is how the evidence orders them, and here is the rule used to order it.
Checking the method yourself
None of this is worth much if you cannot test it. Two things help. First, how to read contractor reviews without being fooled walks through the same reasoning applied by hand, on any platform, so you can see whether a ranking matches the evidence underneath it. Second, if you are a contractor and you think a score rests on something incorrect, the correction process is free and the timeline is published: see if your business is listed.
A method you can only take on faith is a marketing claim. This one is written down so it can be argued with.