What a Good Website Actually Needs: The Non-Negotiables Checklist

Short answer: a good website is fast, works first on a phone, is usable by everyone, is set up for basic search visibility, gives visitors one clear next step, uses real content instead of stock-photo filler, runs on HTTPS with sane security defaults, leaves the code and domain in your name, and gets maintained after launch. Skip any one of these and you are not paying for a website, you are paying for a liability. Speed is not a developer's obsession, it is money Google measures three things on every page: how fast the main content appears (LCP), how quickly the page responds when someone taps something (INP), and whether elements jump around while loading (CLS). These are called Core Web Vitals, and they are a ranking factor, not a suggestion. But the real reason to care has nothing to do with Google directly: a visitor who waits more than two or three seconds for your homepage to load is already gone, and they rarely come back to check if it got faster. A slow site is usually not one big problem, it is a pile of small ones: unoptimized images, fonts loaded the wrong way, third-party scripts nobody audits, and a hosting setup that was never built for speed. Ask whoever builds your site what your Core Web Vitals scores actually are. If they cannot answer, that tells you something. Mobile-first, because that is where your visitors already are For most small businesses, more than half of traffic arrives on a phone, often a lot more. Mobile-first does not mean your desktop site squeezed into a narrow column, it means the phone experience was designed first: thumb-reachable buttons, readable text without pinching to zoom, forms that do not fight the on-screen keyboard, and images that do not eat someone's mobile data. Design for desktop second, not the other way around. Accessibility is not a nice-to-have Accessibility means your site works for people using a screen reader, navigating with a keyboard instead of a mouse, or dealing with low vision or color blindness. In practice that is: real alt text on images, enough color contrast to actually read the text, buttons and links that are reachable by keyboard, and semantic HTML instead of a pile of generic clickable divs. This is not a compliance checkbox for large corporations. It is a meaningful slice of your potential customers, and it is also, not coincidentally, what makes a site easier for search engines and AI assistants to understand. Basic SEO: the unglamorous stuff that actually works You do not need a growth-hacking agency to get the basics right. Every page needs its own unique title tag and meta description that actually describe what is on it. Structured data (schema markup) helps search engines and AI answer engines understand what your business is and does. A sitemap and clean, readable URLs help everything get indexed properly. None of this is exotic. It is mostly things that get skipped because nobody on the project owns them. One clear next step, not five competing ones A visitor who lands on your homepage should know, within a few seconds, what you do and what to do next. Call, book a slot, send a message, buy the thing. Pick one primary action per page and design around it. A site with six different buttons competing for attention in the hero section usually converts worse than one with a single confident call to action, even if the design looks busier and more impressive in a screenshot. Real content, not stock-photo slop Visitors and AI search assistants alike can tell the difference between a business that wrote its own words and took its own photos, and one that filled every section with the same handshake stock photo and vague filler copy that could describe any company in any industry. Real photos of your actual work, your actual space, your actual product build trust in a way no stock library ever will. If your copy could be pasted onto a competitor's site with a find-and-replace of the company name, it is not doing its job. HTTPS and basic security hygiene This should be table stakes in 2026, but it still gets missed: a valid SSL certificate so the connection is encrypted, sane security headers, forms that validate input on the server and not just in the browser, and no exposed API keys or secrets sitting in the frontend code where anyone can view-source them. None of this is visible to a visitor when it is done right. It becomes very visible, very fast, when it is not. Who actually owns the code and the domain This is the one people find out about too late. Some agencies keep your domain registered under their own account, or keep the source code on a platform only they can access, which quietly turns a website into a hostage situation the moment you want to switch providers or bring development in-house. Before you sign anything, ask directly: who owns the domain registration, do I get the full source code, and what happens on a full payment and migration handover. A website is not done at launch Launch day is the start, not the finish line. Dependencies need updates, backups need to exist and get tested, uptime needs monitoring, and content goes stale if nobody touches it. This is also where automation earns its keep: a site that quietly reminds you about a stale page, or an AI agent that answers questions and books appointments while you sleep, keeps working for you long after the launch party is over. Put together, none of this is exotic. It is the difference between paying for a website and paying for a business asset.

What a Good Website Actually Needs: The Non-Negotiables Checklist

Start a Project