How we rank

Methodology v1.0.0 · effective 2026-08-10

The TRP Score is a 0–100 composite of 5 weighted pillars — reputation quality (45%), trust & verification (20%), volume & momentum (15%), longevity (12%), engagement & transparency (8%). Franchises and chains are excluded by policy. Position cannot be purchased, and no manual action anywhere in this system can set a score or a rank.

The pillars and their weights

Why a five-star newcomer does not come first

A perfect average over a handful of reviews is weak evidence. Every company is scored as though it began with 25 reviews at the market median for its list, so a long record of very good work outranks a short record of perfect work. The market median is published on every ranking page, so you can check the number we used.

Reviews also lose weight with age, halving every 730 days. Nothing is ever discarded; a review from five years ago still counts, just less than one from last month.

A worked example

Computed by the same engine that scores real companies. If the code changes, this changes with it.

What we verify, by trade

Texas licenses almost none of this work, so each checklist measures against what that trade can actually prove. Points are normalized within a category: a business holding every core item reaches the full pillar.

Who is eligible

• Independent. Not a franchise, franchisee, or corporate chain location. The full excluded-brand list is published.
• Local and operating. Not marked permanently closed, reachable by phone or website, and inside the metro we are ranking.
• A minimum record. At least 5 reviews to be listed and 10 to hold a rank, with a first review at least 6 months old.
• Rated at least 3.0 stars.
• Working in the trade being ranked. A ranking is of one trade, so a company holds a position in it only if that trade is the primary category on its own listing. Secondary categories do not count: almost every general contractor also files itself under decking, which is exactly how a lawn-care company came to lead a ranking of deck builders. Generalists who take the work occasionally stay listed with their score and their reviews — they simply do not take a number from the specialists, which their total review count would otherwise let them do. Some trades share their practitioners: most people who clean upholstery are listed first as carpet cleaners. Where that is true we say so in the category’s configuration, and a company from the neighbouring trade ranks if it also names this one among its own services — two statements from the company, not one.
• Entity records are checked but not required. Many independents trade under an assumed name that differs from their registered one, so we look every company up in the Texas franchise-tax register and say plainly when we could not find a match. A company we could not confirm still ranks — it simply forfeits those points and carries the disclosure.

How the list is capped

A ranked list holds at most 100 companies. Everyone else keeps a profile and a score; only the rank number is withheld. Positions trade places only when the margin exceeds 0.25 points, and entering the list needs a margin of 0.5 — otherwise the table would reshuffle every month on noise.

Two companies that finish level are separated first by whether they are based in the place being ranked, then by adjusted rating, then by effective review count, then by age, and finally by name so the order never wanders. Distance never costs a position on merit — a company that drives in is ranked on the same scale as one down the road — but when nothing else separates them, the local one goes first.

Why the published score is not the pillar total

Adding up the weighted pillars gives a company’s score for that week. The number we publish is an exponential moving average of the last 4 weekly scores, weighting the newest at 0.6 and the running average at 0.4. One unusually good or bad week therefore moves a rank part of the way rather than all of it, and a company cannot climb by arranging a single week.

Both numbers are shown on every profile: the pillar table sums to the week’s score, and the headline is the smoothed one. They differ most for a company whose score has just moved, and converge when it holds steady.

Review integrity

We look for review bursts against a company’s own history, clusters of accounts with no review history, near-duplicate text, and star distributions that differ sharply from the rest of the market. When two independent checks agree, the affected window of reviews is set aside and the score recomputed without it.

The arithmetic does not care which direction an anomaly runs. A business being attacked with fake one-star reviews is protected by exactly the same machinery that catches a business buying five-star ones. Any penalty beyond setting reviews aside requires a person to decide it, and every such decision is published in the corrections log.

Can a business pay to rank higher?

No. There is no mechanism in this system that sets a score or a rank — not for money, and not for us. Editors can correct wrong data, exclude a franchise, or record a verified credential, and the engine then recomputes from those inputs. Every one of those actions is logged publicly with its reason.

Where the data comes from

What the score bands mean

• 90+ — Elite
• 80+ — Excellent
• 70+ — Strong
• 60+ — Fair
• 0+ — Developing

What we cannot see

• Reviews that Google filters or removes are invisible to us, as they are to everyone outside Google.
• No public registry lists general liability insurance anywhere in the United States. When we show insurance, a business sent us a certificate and we checked the carrier and the date.
• Sole proprietors trading under a county assumed name do not appear in the state entity register at all.
• Service areas on Google Business Profiles are self-declared and cannot be independently verified.