108
bail bond reviews read across 12 agencies.
They showed who actually picks up the phone: usually a parent or partner, not the person in jail.
30 reviewers were family or friends, 10 were defendants.
How we build them
Before we draw a single page for a new industry, we study the businesses already winning online, read what their customers say, and count what works and what doesn't. Then we build designs that keep the good parts and fix the gaps.
Who we study
Our newest industries start with 20 to 30 live US websites. Not random ones: they have to earn their place in one of three groups.
The brands that invest the most: many locations or a franchise, big review counts, national advertising, websites hundreds of pages deep.
Businesses that win mainly through their website, an app or online booking, because they've had to make the site do the selling.
Local owners with the best ratings at real volume, for example 4.8 stars or more from 150 or more reviews, in the markets where customers spend the most.
We pick 4 to 6 US markets per industry and write down why: high-income areas, and the places demand runs highest for that trade. For bail bonds, that meant the states with the most commercial bail, and leaving out states where it's banned.
Directories, marketplaces and lead-generation sites, unless they're the real competition. Businesses outside the US. And for product brands, giants that mostly sell in stores: their websites are an afterthought, and copying them teaches the wrong lessons.
Then we pick the 10 that matter most for a full teardown: at least 3 national leaders, 3 independents and 2 online-first brands, covering every way the trade sells.
What we record on every site
In our newest industries, each site we study gets its own written file. Every claim is something we saw on the live site, with the link and the date.
The order of every section on the homepage, desktop and phone
What fits on the first phone screen, where most of your customers land
How a customer gets a price, books or gets a quote, and how many taps it takes
Every way to reach the business: phone, text, form, chat, and any promise about response time
The proof they use: reviews, ratings, guarantees, licenses, and where it sits
The menu and footer, and how deep the site goes
How it looks: colors, type, photos, who appears in them, and what looks dated
What works, and what it does badly, specific enough to design from
What customers actually say
A website has to answer the questions customers really have. So we read reviews across the businesses we study and sort them into what people praise, what they complain about, and what worried them before they bought.
108
bail bond reviews read across 12 agencies.
They showed who actually picks up the phone: usually a parent or partner, not the person in jail.
30 reviewers were family or friends, 10 were defendants.
144
personal trainer reviews and testimonials read in full across 9 businesses, plus Reddit threads where people talk about choosing a trainer.
We never copy a review into a design. What we learn tells us what the sample reviews should cover, and which of your real reviews to feature.
What the leaders get wrong
We count patterns across every site, so a finding is a number, not an opinion. A few from our research:
1 of 17
bail bond sites we could check had a call button that stays on screen on a phone.
Bail bonds
10 of 18
show no rating or review count at all.
Bail bonds
19 of 27
hide the main price.
Personal trainers
22 of 27
have no call link on the first phone screen.
Personal trainers
7 of 10
operators hide turnover prices behind a phone call.
Vacation rental cleaning
0 of 16
show a sample of the photo report they promise.
Vacation rental cleaning
3 of 13
have a real heat or flavor guide.
Hot sauce brands
4 of 10
show the main button without scrolling on a laptop.
Food brands
Every gap like these is a reason a customer leaves. Our designs are built to close them.
How we decide what to build
We don't pick a number and fill it. Each design has to be a different way of selling that the research supports, and it only makes the cut if:
“A design added to reach a number is the same page with a new name.”
The thumbnail test: every homepage in a set should be recognizable at a glance without reading a word.
How we rank them
Our newest designs are scored out of 10 by a review that's separate from the build and told not to be generous. It looks at every page, on desktop and phone.
Clear at a glance, an exact grid, every section earns its place, a phone layout designed for the phone rather than squeezed onto it, and clearly better than the best competitor page it answers.
All of that, with one small weakness we can name. Ready to show a client as it is.
Solid, but generic in places. Goes back for fixes.
Rebuilt.
After the fixes, a new review that hasn't seen the earlier notes gives the final score. That score is the one that counts: never an earlier, higher one, and never an average.
What we check on every page
Every page of every design is checked at desktop and phone width before it goes in the library:
112 designs across 18 trades. Find yours and preview every page, on desktop and phone, before you buy.
Find your trade