It has to be findable
Every page needs its own title and description — not the same one repeated, which is what you get when nobody bothered. The title is what appears as the blue link in search results; the description is the grey text under it. They are your advert, written by you or written badly by a search engine.
You need a sitemap so search engines can see every page without guessing, a robots file telling them what to crawl, and canonical tags so the same page at two addresses does not compete with itself. You need the site registered in Google Search Console, which is free and is the only place you can see what people actually searched before they arrived.
Structured data is the one most small sites skip. It is a block of machine-readable information describing what your business is, what it sells and where it operates. It is how you become eligible for the richer-looking results with ratings, prices or expandable questions.
It has to be fast, and measured rather than assumed
Speed is not vanity. People leave slow pages, and search engines know they leave. The usual causes are dull: enormous uncompressed images, fonts loaded from somewhere else, and scripts for things nobody uses.
The important word is measured. "It feels fast on my laptop" is not a measurement. It should be tested on a mid-range phone on a normal connection, because that is what most of your visitors are using.
It has to work for people who are not using a mouse
Some of your visitors navigate by keyboard, some use a screen reader, some have low vision, and far more than you expect are reading your site one-handed on a bus in bright sunlight.
The basics are not expensive: real contrast between text and background, visible focus outlines so keyboard users can see where they are, labels on form fields, alternative text on images, and headings in a sensible order. Done from the start this costs almost nothing. Retrofitted afterwards it costs a great deal.
The contact route has to be tested end to end
A contact form that silently fails is worse than no form at all, because you never learn about the enquiries you lost. Someone should fill it in before launch and confirm the message actually arrives — in the inbox, not just in a success message on screen.
The same goes for whatever happens next: an email that lands in spam, a reply address that bounces, or a notification nobody is watching. Test the whole path, not the first step.
You have to be able to see what happens
Without analytics you are guessing. You want to know which pages people land on, which ones they leave immediately, where enquiries come from and what they searched for.
You do not need an elaborate setup. You need enough to answer "is this working" honestly, and the discipline to actually look at it once a month.
You have to own it
This is the one that costs people most and gets discussed least. The domain should be registered in your name, not your developer’s. The hosting account should be yours. The analytics, the search console, the email — yours.
Ask directly, before any money changes hands: when this is finished, what is in my name, and could another developer pick it up? A studio that answers that question clearly is one you can leave, which is exactly why you probably will not want to.
What you can skip, at least at first
Plenty of what gets sold alongside a website is worth having eventually and not worth having on day one:
- A blog, unless somebody has actually agreed to write it. An abandoned blog dated two years ago is worse than no blog.
- Live chat, unless somebody is there to answer it.
- A newsletter, until you have people to send it to.
- An elaborate content system, if the site changes twice a year.
- Multiple payment gateways, when one works.
If a quote covers the look of the site and nothing on this list, it is not a cheaper version of a professional website. It is a different and smaller thing, and the difference turns up later.