How we rank landscapers

Every landscaper on a Yard Index list is scored by the same formula, from the same public data, on the same day of the month. This page is that formula in plain English. Weights version 2026-09.

Where the data comes from

We read public Google Business Profile data: the business name, primary category, address or service area, phone, website link, star rating, review count, the dates of recent reviews, whether the owner replied to them, the number of photos on the profile, and which profile fields are filled in. We refresh it on the first of every month and rank from that snapshot. We store review dates and reply flags, never review text, and we never republish photos from a profile.

We also visit the website linked on the profile to check three things: that it answers, that it answers over HTTPS, and that it renders on a phone. Nothing else on the site is read or scored.

Who is in the field

For each city we search Google for landscaper, landscaping company, lawn care service and keep every profile whose primary category is a landscaper category, or that lists one among its categories. From that pool we remove:

businesses in another trade, including the suppliers, nurseries and equipment dealers the search returns;
profiles with fewer than 5 reviews, because the rating is not yet meaningful;
national franchises and their local branches, which compete on a different footing than local firms;
profiles outside the 20 km radius we use for the city, or in a locality that belongs to another city on this site;
profiles marked permanently closed, and duplicate listings for a business already counted.

Each city page publishes its own counts: how many profiles we started with, how many were evaluated, and how many were removed for each reason.

The factors

Weights, version 2026-09
Factor	Weight
Rating quality	25%
Review volume	20%
Recent reviews (90 days)	20%
Owner replies to reviews	10%
Profile completeness	10%
Photos	10%
Website	5%

Rating quality — 25%

The visible star rating on its own rewards a business with four perfect reviews over one with three hundred at 4.8. We correct that with a Bayesian average: the rating is pulled toward the city mean until enough reviews back it up. A 5.0 from 4 reviews in a city that averages 4.5 scores near 4.6; the same 5.0 from 200 reviews scores near 5.0. The result is then placed on a scale running from half a star below the city mean up to 5.0.

Review volume — 20%

The total number of reviews on the profile, log-scaled and capped at the 90th percentile for the city. Log scaling means the step from 10 reviews to 40 counts for more than the step from 400 to 430, which is how homeowners read the number too.

Recent reviews (90 days) — 20%

Reviews inside the last 90 days. This is the factor that separates a business still collecting proof from one living on a reputation earned years ago. It saturates at twice the city median, so a steady stream scores full marks without needing to be the loudest.

Owner replies to reviews — 10%

The share of recent reviews that carry an owner response. It is measured on the newest reviews we read for each profile. Where too few recent reviews exist to measure it fairly, the factor is estimated from the city, and where most of a city cannot be measured the factor is dropped for that city and the remaining weights are renormalized.

Profile completeness — 10%

How many core profile fields are filled: hours, website, phone, categories and attributes. These are the fields a homeowner checks before calling, and they are free to fill in.

Photos — 10%

The number of photos on the profile, log-scaled and capped at the 90th percentile for the city, so a profile with 900 photos does not make everyone else look empty.

Website — 5%

Three checks on the website linked from the profile: that one is linked at all, that it answers over HTTPS, and that it renders on a phone. A social page in place of a website counts for less.

Profile completeness counts these fields: hours, website, phone, categories, attributes. A profile description is not scored, because the data source does not expose it reliably.

The arithmetic

Ratings are weighted by how many reviews back them

A raw star rating treats four reviews and four hundred as the same evidence. We use a Bayesian average instead: (C × m + R × n) ÷ (C + n), where R is the rating, n the review count, m the city mean and C = 20. In a city averaging 4.5 stars, a 5.0 from 4 reviews lands near 4.6, while a 4.9 from 200 reviews stays at 4.9. The adjusted figure is then placed on a scale from half a star below the city mean up to 5.0.

Counts are log-scaled and capped

Review counts and photo counts use log10(x + 1) ÷ log10(cap + 1), where the cap is the 90th percentile for that city. Going from 10 reviews to 40 therefore counts for more than going from 400 to 430, and one business with thousands of photos does not flatten everyone else.

Everything is normalized inside the city

Every factor is scored from 0 to 1 against the rest of that city's field before any weighting. A landscaper is only ever compared with others serving the same city.

The composite

The composite is the weighted sum of the factors, multiplied by 100 and rounded to one decimal. Ties are broken by review velocity: the business with more reviews in the last 90 days ranks higher.

When a factor cannot be measured

If we cannot measure a factor for a business, we do not guess in its favor: the value is estimated from the city and the page says so. If more than 50% of a city cannot be measured on a factor, that factor is dropped for the whole city and the remaining weights are renormalized, so every business in that city is scored on the same basis. The city page names any factor dropped that month.

What we do not use

Money. There is no paid placement, no sponsorship and no advertising on these lists.
Client status. Some ranked businesses are clients of D'Amico Digital. That is not an input to the formula, and no one at D'Amico Digital can move a business on this list.
Opinion. No editor picks, no hand-weighting of individual businesses, no "editor's choice".
Review text and photos. We count reviews and replies; we do not read, quote or republish them.
Claim status. Claiming a listing adds a Verified mark. It does not change the score.

Cadence and corrections

Lists are rebuilt on the first of each month and the month appears in every list title. Between builds, nothing on a list moves. The next re-rank runs on October 1, 2026.

If a fact about your business is wrong on a list, email noah@damicodigital.com with the business name and the city; we correct it from the source data on the next build. Ratings and review counts come from Google and change only when Google changes them.

Version history

2026-09: Initial weights.

When weights change, the version changes with them and the change is listed here. Lists published under an older version keep the version they were built with, shown in the footer of every page.