What should a kitchen remodeler's homepage say? We read the ones on Google's page one

Remodeling is sold on trust, and the usual advice follows from that: fill the homepage with the words homeowners want to hear. Best, trusted, premier, award-winning, five-star. The remodelers who win page one, the story goes, lean hard on that language, and a remodeler who does not is underselling.

We tested it. On September 28, 2026 we searched Google for “kitchen remodeling” in 22 US cities, 11 large metros and 11 small cities, and read the homepage of every remodeler on page one. One search failed twice, so 21 cities returned results: 177 page-one results, 59 remodelers after screening, and 48 homepages with readable text.

The belief did not hold. The median page-one remodeler used one distinct praise word, and one checkable number. Page one is not loud. It is thin on proof.

Your benchmark

Here are the two counts to take to your own homepage. First, the praise words: count how many different ones from this list appear (best, #1, number one, top-rated, highest-rated, leading, premier, trusted, award-winning, voted, five-star). Second, the checkable claims: count how many different numbered facts appear, such as years in business, projects completed or reviews received.

Measure (48 homepages)	Median	Share with none	The most on one page
Different praise words	1	21%	5
Different checkable numbers	1	44%	7
Proof types, out of the 6 below	1	6%	5

Counted by occurrence rather than by distinct term, the median homepage used a praise word twice. Two different checkable numbers put a homepage ahead of 81 percent of page one. Three different proof types put it in the top 13 percent.

What page one actually says

We grouped the checkable proof into six types a homeowner could verify, and scored each homepage out of six:

Proof type	Homepages carrying it
A link to a project gallery or portfolio	81%
A founding year (“since 1994”)	25%
A license or insurance statement	21%
A named third-party award, rating or membership (Best of Houzz, BBB A+, NARI)	17%
A star rating with its number (“4.9 stars”)	8%
A warranty with a stated term (“one-year warranty”)	4%

The median score was 1, and for most sites that one was the gallery link. Only 42 percent carried two proof types, and no homepage carried all six. Just 2 of 48 stated how long their warranty lasts, in a trade where the job is judged years after it is finished.

Praise and proof were not substitutes. Of the 48 homepages, 24 carried both a praise word and a checkable number, 14 carried praise with no number, 3 carried a number with no praise word, and 7 carried neither.

Did any of it sort page one?

We could not detect it. Against position on the page, the count of distinct praise words gave a rho of 0.128, checkable numbers 0.082 and the proof score 0.16, none significant. With 48 homepages this sample can only detect strong effects, so this is not evidence that wording has no effect. It is evidence that page one is not sorted by a large one.

The top three looked like the rest. Their 18 readable homepages had the same medians: one praise word, one checkable number, one proof type. None of the 18 showed a star rating or a warranty term. The four homepages that did show a rating sat at positions 6 to 9, and the rating’s correlation with position (rho 0.335) is significant but rests on those four sites, so read it as a description of this sample, not a rule.

Large metros and small cities wrote the same way. The median was one praise word and one checkable number in both. What differed was the site behind the homepage: a median of 107 pages in metros against 26 in small cities, a significant gap.

The bigger surprise: most of page one is not a remodeler

Before any of that, we had to find the remodelers, and they were a minority of the page. Of 177 page-one results, remodelers held 62 slots (35 percent). Directories, marketplaces and social platforms held 83 (47 percent). Ranking-list publishers, review platforms and data vendors held 12. And 14 slots (8 percent) went to lead-generation sites: templated city pages with a toll-free or out-of-state call-routing number and no named business behind them.

The split by city size is sharp. In large metros remodelers held 44 of 90 slots (49 percent). In small cities they held 18 of 87 (21 percent), while directories held 62 percent.

Nashville is the extreme case. Its page one returned six results and not one was a remodeler: two videos and four lead-generation pages, one of them on a domain named for a different city in another state. Remodelers were at position 1 in 13 of 21 cities.

Lead-generation sites are worth recognizing because they imitate exactly what this study measures. They use a city name in the domain or subdomain, carry “top-notch” and “best” in their headings, and repeat the same template city after city. The quickest tell is the phone number: a toll-free or distant area code on a page that claims to be local.

What to do with this

• Count your praise words and your numbers against the medians. One and one is page one’s middle. Moving to two different checkable numbers puts you ahead of 81 percent of it.
• Add the proof page one skips. A stated warranty term, a founding year and a license statement cost nothing to write, and most page-one competitors do not carry them.
• Look at who you are really next to. In a small city you may share page one with directories and lead-generation pages more than with other remodelers. Your homepage is being compared with a template.
• Treat praise as neutral, not required. Page one did not lean on it, and we could detect no advantage from more of it.

What changed in verification

The first pass let through every publisher, lead-generation page and unverifiable keyword domain in the tables above, because each has “kitchen” or “remodel” in its title or domain. We opened every kept site and classified it by what the page itself says: a named business with an address, a template with a call-routing number, or a list of other companies. A lumber yard, a marketing agency and a franchise sales page came out the same way. A pilot search before the study had already shown how loose the screen was: Google answered “kitchen remodeling” in one metro with pages about the word “North”, and the screen passed a payments company because “powerhouse” contains “house.” We replaced that city and tightened the screen before spending the budget.

Reading the matched phrases changed the measurement too. One homepage repeated the same few numbers so often that it counted 17 claims, and its distinct count is 3. Review widgets repeated “5 star” on every card. Two homepages numbered their menu (“01 Kitchens, 02 Bathrooms”) and those numbers counted as claims. We now report distinct counts, drop numbers with a leading zero, and widened the star rating rule after it missed “4.9 Rated” written after a row of stars.

Limits: what this can and cannot tell you

Measured: the homepage text of 59 kitchen remodelers holding 62 page-one slots for one query in 21 cities on one day, with every excluded result kept in the dataset with its reason.

Not measured: any page other than the homepage, so a warranty or license stated on an inner page does not count here. Six homepages were built by script and returned no readable text, and five did not respond; all are reported as missing, not as lacking proof. We did not measure reviews, backlinks or the Google Business Profile, which plausibly matter more for ranking than homepage wording.

Range restriction: every remodeler here already ranks on page one. The study describes how winners write; it cannot say whether a remodeler on page three would climb by writing differently.

Other explanations: our praise list counts phrases, not intent, so “Best of Houzz” counts as an award and also as the praise word “best.” A national brand’s local page resolves to the brand’s homepage, which we measured for its three locations.

The number to write down

One praise word, one checkable number, one proof type: that is the median page-one kitchen remodeler. The cheapest way to stand out on that page is not a louder adjective. It is a year, a warranty term and a rating a homeowner can check.