The 7 things every small business website needs in 2026

The short answer
Seven things, and none are design: your details as real text, a mobile site that works one-handed, opening hours that are correct today, one obvious action per page, pages that answer the questions you get asked, structured data so machines can read you, and someone who can update it without emailing anyone. Everything else is optional.
On this page
The list, in priority order
Not a design checklist. These are the things that decide whether a website earns anything, and they are ordered by what costs you most when missing.
1. Your details as real text
Name, address, phone, hours, prices — as selectable text on a page, not inside an image or a PDF.
This is first because it is the most common failure and the most total. Text inside an image cannot be found by search, read by an assistant, tapped to dial, or read aloud by a screen reader. A menu as a photograph is, to a machine, a page with no menu on it.
The test: on your phone, try to select your own phone number with your finger. If nothing highlights, it is a picture.
2. A mobile site that works with one thumb
Most of your visitors are on a phone, and Google judges your site by its mobile version, not the desktop one you look at.
The test: on mobile data rather than office wifi, find your opening hours in under ten seconds and complete a contact form one-handed. Any step that makes you sigh loses customers who will not sigh — they will leave.
3. Opening hours that are right today
The single most-checked fact on a small business site, and the one most often wrong. Bank holidays, seasonal changes, the week you closed for a refit.
Wrong hours do a specific kind of damage: someone travels to you and finds a locked door. That is worse than having no website, because you converted a customer into a bad experience.
The test: whatever the next public holiday is, does your site say what you are doing? Right now, without you editing anything?
4. One obvious action per page
Every page should make one thing easy: book, order, call, get directions. Not five equally-weighted options.
Pages with several competing calls to action convert worse than pages with one, because a visitor who has to choose frequently chooses nothing.
The test: open a page and ask what you want a stranger to do next. If you cannot answer in one sentence, neither can they.
5. Pages that answer the questions you actually get asked
Every question you get twice a week is a page, or at least a clearly headed section: do you do set lunch, is there parking, do you take walk-ins, do you do gluten free.
This is not content marketing, it is the mechanism. A question answered on a page can be found and quoted. The same answer given in person cannot.
The test: list the five questions your staff answer most. How many are answered on the site, in words a search engine could return?
6. Structured data
Hidden labelling that tells search engines and AI assistants what your page means — this is a restaurant, these are its hours, this is a price.
It does not change what visitors see, and it is not a ranking trick. It makes you easier to cite correctly, which matters increasingly as more people ask an assistant rather than scrolling results.
The test: run any page through Google's Rich Results Test. It will tell you exactly what it can read.
7. Someone who can update it without emailing anyone
The one that quietly determines whether the other six stay true.
A site nobody can edit does not stay correct. Hours drift, prices go stale, the seasonal menu stays up until March. Not through negligence — because every change costs an email, a wait, and sometimes a fee, so small corrections do not get made.
The test: how long would it take you, personally, to change one price right now?
What is not on this list
A blog, if you will not maintain it. A chat widget nobody staffs. An animated homepage. A photo carousel — they are consistently ignored by visitors and they slow the page.
None of those are wrong. They are just not what decides whether this works, and they are where budget goes when the seven above have not been settled first.
Why the seventh decides the other six
Worth stating plainly, because the list reads as seven equal items and it is not.
Items one to six all decay. Hours change. Prices change. The service you added last spring is still not on the site. The photographs are from the old fit-out. Structured data drifts out of line with the page it describes. Every one of them is correct on launch day and slightly wrong a year later — unless somebody can fix it without raising a job.
That is why number seven is not the last item on the list but the mechanism for the rest of it. It is also the one most often traded away, because it is the least visible thing on a proposal and the easiest to defer to phase two.
Helm is built around precisely that trade. A designed site — by the team in days, generated as a unique one by Helm AI, or your existing site attached — with everything that changes wired to a dashboard: text, images, prices, menus, hours, and the SEO title and description on each page. The first six stop being launch-day achievements and become the ongoing state of the site.