---
title: "What makes a good website? Not the design | PostCliq"
description: "AI builders made polish the entry requirement. What makes a good website in 2026, a checklist that separates looking finished from working, with sources."
url: "https://postcliq.com/blog/good-looking-website-is-not-enough"
---

# What makes a good website? Anyone can build a good-looking one now, and that is the problem

AI builders ship a polished site in an afternoon, so looking good became the entry requirement. What makes a good website now is whether it works for the buyer. Here are the four things that still have not been automated.

By [Nirmal Gyanwali](https://postcliq.com/blog/author/nirmal-gyanwali) Updated 30 September 2026 9 min read

On this page

1. The short answer
2. What changed about building a website?
3. Polish is now the floor, not the ceiling
4. What makes a good website in 2026?
5. Where do AI website tools stop?
6. Why is looking finished not the same as working?
7. Does Google penalise AI-built websites?
8. The four things that did not get automated
9. How do you know whether your website works?
10. What to do next
11. Frequently asked questions

## The short answer

Key takeaways

- A good website is one that turns the right visitors into enquiries or sales, not one that looks current.
- AI website builders now produce a presentable site in an afternoon, so polish has become the entry requirement rather than the advantage.
- 95.9% of the top million home pages failed automated accessibility checks in 2026, so looking finished and working are separate measurements.
- What still separates two good-looking sites is what the buyer needs to see, the proof, a path that works, and knowing whether a change worked.

What makes a good website in 2026 is not how it looks but whether it works for the person trying to act on it. WebAIM found detectable accessibility failures on [95.9% of the top million home pages](https://webaim.org/projects/million/) in February 2026, up from 94.8% a year earlier, in the same period the tools for producing polished pages got dramatically better. Design is now table stakes. Conversion is where the difference is.

## What changed about building a website?

Building a presentable website went from months and a designer to an afternoon and a description. Website builders now generate pages, copy, images and colours from a brief, and most developers use AI tools in their daily work. The labour cost of looking professional collapsed at both ends of the market.

Wix markets its AI builder as creating a full site from a conversation about your business, "in minutes" ( [Wix AI Website Builder](https://www.wix.com/ai-website-builder) ). Squarespace describes the same process, producing "a personalized, multi-page website" with pages, copy, imagery, colours and fonts from a short brief ( [Squarespace AI website builder](https://www.squarespace.com/websites/ai-website-builder) ).

Those are vendor claims on vendor pages. Treat them as marketing. But you can test them in twenty minutes yourself, and most people who do come away agreeing that the output clears the bar their own site cleared in 2019.

On the custom build side the same thing happened. Stack Overflow's 2025 developer survey found 84% of respondents using or planning to use AI tools in their development work, up from 76% the year before, with 51% of professional developers using them daily ( [Stack Overflow Developer Survey 2025](https://survey.stackoverflow.co/2025/ai) ).

## Polish is now the floor, not the ceiling

Here is the commercial consequence.

When a good-looking site was scarce, having one was an advantage. When everyone has one, having one is a requirement. You do not get credit for meeting a requirement. You only get penalised for missing it.

That means the money you were about to spend making the site look more current is buying you a position you already hold. It is defensive spending dressed up as growth. Before you sign a rebuild, work through [the questions to ask before you rebuild your website](https://postcliq.com/blog/questions-before-you-rebuild-your-website) , because "it looks dated" is now the weakest reason on the list.

The advantage moved. It is now in whether the thing converts, because that part did not get cheap.

## What makes a good website in 2026?

A good website confirms the visitor is in the right place, shows the proof a buyer needs, makes the next step obvious, works end to end on a phone, loads quickly, can be used by everyone, and is measured. Only the first of those is mostly about design.

| Quality | Does an AI builder give it to you? | How to check it yourself |
| --- | --- | --- |
| Looks current and consistent | ✓ Largely | Open it next to two competitors |
| Confirms the offer in the first screen | ◐ Only if the brief was specific | Read the first sentence out loud |
| Shows proof a buyer trusts | ✕ It cannot invent real proof | Count named clients, case studies and substantiated numbers |
| An obvious, outcome-worded next step | ◐ Often "Submit" or "Learn more" | Ask a stranger what to do after ten seconds |
| A path that works end to end | ✕ Generation does not verify | Submit the form from a personal address |
| Loads fast on a phone | ◐ Depends on images and plugins | Time the first screen on mobile data |
| Usable by everyone | ◐ Contrast and labels often fail | Run a free accessibility checker |
| Measured | ✕ You define what counts | Write down what a conversion is |

Source: PostCliq analysis

Each row maps to a longer guide. Proof is covered in [the trust signals that make B2B buyers enquire](https://postcliq.com/blog/website-trust-signals) , the form in [contact form best practices](https://postcliq.com/blog/contact-form-best-practices) , and speed on WordPress in [why is my WordPress site so slow](https://postcliq.com/blog/why-is-my-wordpress-site-slow) .

## Where do AI website tools stop?

AI website tools stop at verification. They are good at structure, layout, responsive behaviour, plausible copy and a coherent colour system. They are weak at checking that the form delivers, the links resolve and the page serves the specific buyer. Developers name "almost right" as their top frustration with AI output.

Be fair to the tools. Structure, layout and plausible copy are pattern problems, and pattern problems are what this technology solves.

They are weaker at the step immediately after. In the same Stack Overflow survey, the single most common frustration, named by 66% of developers, was "AI solutions that are almost right, but not quite", and the second, at 45.2%, was that debugging AI-generated code takes more time (same source).

"Almost right" is a precise description of the failure mode you should expect on a generated website. The contact form exists and looks correct. Whether the enquiry arrives in a human's inbox is a different question, and nothing in the generation step answers it.

## Why is looking finished not the same as working?

Looking finished is a visual judgement; working is a behavioural one. A page can pass a client review and still fail people who try to read it, fill in its form or load it on a phone. The best long-running measurement of this, WebAIM's, got worse in 2026.

WebAIM tests the home pages of the top million websites once a year with automated tooling. In February 2026 it detected WCAG 2 A and AA failures on 95.9% of home pages, averaging 56.1 errors per page, up from 94.8% and 51.0 errors the year before ( [WebAIM Million](https://webaim.org/projects/million/) ).

Looking finished versus working, in four numbers

**95.9%** of the top million home pages had detectable accessibility failures, 2026 [WebAIM Million](https://webaim.org/projects/million/)

**83.9%** of those home pages had low-contrast text [WebAIM Million](https://webaim.org/projects/million/)

**51%** had at least one form field with no label [WebAIM Million](https://webaim.org/projects/million/)

**48%** of websites pass all Core Web Vitals on mobile [HTTP Archive Web Almanac 2025](https://almanac.httparchive.org/en/2025/performance)

Three things to hold onto.

First, the direction. A multi-year improvement trend reversed. Home pages got worse on this measure, not better, in the year the tooling got dramatically better at producing them.

Second, the honest caveat. WebAIM does not establish that AI tooling caused this, and neither does this article. The report notes page complexity rose over the same period, which is a plausible alternative explanation. Two facts sitting next to each other are not a cause.

Third, and most importantly, WebAIM states that "absence of detected errors does not indicate that a page is accessible or conformant". Automated detection finds only machine-findable failures, so the real figure is higher, not lower.

The point stands without the causal claim. A page can be generated, look finished, pass a client review and still fail people trying to use it. A missing form label is not cosmetic: it is a field some visitors cannot complete. Speed is the same story. Only [48% of websites pass all Core Web Vitals on mobile](https://almanac.httparchive.org/en/2025/performance) , and Portent's study of 27,000+ landing pages found [B2B sites loading in one second convert three times better](https://www.portent.com/blog/analytics/research-site-speed-hurting-everyone.htm) than those loading in five, though the B2B sample was only 20 sites. Looking right and working are separate measurements, and only one of them is being automated.

## Does Google penalise AI-built websites?

No. Google's documented test is whether a page helps people, not how it was made. Its spam policies target pages generated at scale to manipulate rankings, whether by AI, scraping or stitching. A generated page and a hand-built page start level, so something other than production method has to separate them.

Google's spam policies define scaled content abuse as generating many pages "for the primary purpose of manipulating search rankings and not helping users", and list generative AI tools as one route to that outcome alongside scraping and stitching ( [Google Search Central spam policies](https://developers.google.com/search/docs/essentials/spam-policies) ). The mechanism is not what triggers it. The absence of value is.

Which cuts both ways, and the second way is the one worth acting on. If Google is indifferent to how the page was made, then a generated page and a hand-built page start level. Whatever separates them has to be something else.

## The four things that did not get automated

These are the parts that still take a person who knows the business.

1. **Knowing what this specific buyer needs to see before they will act.** A tool can write a services page. It cannot know that your buyers will not enquire until they see a fixed price, or a named consultant, or a client in their own industry.
2. **Having the proof.** Generated copy will happily write "trusted by hundreds of businesses". The case study, the named client, the number you can substantiate: those are assets, and you either have them or you do not. Under Australian Consumer Law a claim about results needs a reasonable basis ( [ACCC on false or misleading claims](https://www.accc.gov.au/business/advertising-and-promotions/false-or-misleading-claims) ).
3. **Making the path actually work.** Submitted forms that arrive. Links that resolve. Phone numbers that are tappable. This is verification, not generation, and it is unglamorous enough that almost nobody does it.
4. **Knowing whether the change worked.** Defining the measure before the change, and reading it after. See [what a conversion rate actually means](https://postcliq.com/blog/what-a-conversion-rate-actually-means) .

None of those four is a design problem. All four are why two sites that look equally good produce different numbers of enquiries.

## How do you know whether your website works?

You know by watching what visitors do and testing the path yourself. Analytics shows where people leave, session recordings show what they tried before leaving, and your own form submission shows whether an enquiry arrives. Together they answer the question design reviews never do.

A free session recording tool is the fastest way to see the gap between how a page looks and how it behaves. Our [Microsoft Clarity guide for marketers](https://postcliq.com/blog/microsoft-clarity-guide) shows how to set one up in an afternoon, and [dead clicks and rage clicks](https://postcliq.com/blog/dead-clicks-and-rage-clicks) explains the two signals that most often mean something is broken. If you are choosing a tool, compare [Microsoft Clarity and Hotjar](https://postcliq.com/blog/microsoft-clarity-vs-hotjar) or the wider list of [Hotjar alternatives](https://postcliq.com/blog/hotjar-alternatives) , and see where they fit among [the best CRO tools in 2026](https://postcliq.com/blog/best-cro-tools) .

Recordings tell you what happened, not what to fix first. That ordering is the job of a [CRO audit](https://postcliq.com/blog/cro-audit) . PostCliq does the reading part: it checks up to 50 public pages for forms and calls to action, content clarity, speed, accessibility, trust signals and search access, and returns a private report with evidence and three quick wins. It does not redesign anything. You can [audit your website free](https://postcliq.com/#try-postcliq) .

## What to do next

Stop asking whether your site looks current. It is the wrong question and it has been for about two years. Ask these instead, this week.

1. **What does a buyer have to see before they enquire, and is it on the page they land on?** Write the list. Then open the page.
2. **Does the path work end to end?** Submit your own form from a personal address. Click every link on your highest traffic page. Try the phone number on a phone.
3. **Can everyone use it?** Run a free automated accessibility check on your home page and your enquiry form, and fix unlabelled fields first.
4. **What would we point at if someone asked why they should choose us?** If the answer is the design, you do not have an answer any more.

The [one hour website check](https://postcliq.com/resources/one-hour-website-check) turns the second question into a list you can tick off. Start there, because it is the one that finds money.

## Frequently asked questions

### What makes a good website?

A good website gets the right visitors to take the next step. In practice that means it confirms the offer in the first screen, shows real proof, has one clear call to action worded as an outcome, works end to end on a phone, loads quickly, can be used by everyone, and is measured against a written definition of success. Visual design supports all of that but is no longer enough on its own.

### What makes a good website design?

Good website design makes the page easy to understand and act on: a clear hierarchy, readable contrast, obvious buttons, and nothing competing with the main action. It is judged by what visitors do, not by how the page looks in a review. With AI builders producing polished layouts in minutes, design quality is now the minimum a buyer expects rather than a reason to choose you.

### Are AI website builders good enough for a business website?

For structure, layout and a presentable first draft, often yes. Wix and Squarespace both market builders that generate a multi-page site from a short brief. What they do not do is verify that your form delivers, your proof is real and your pages answer your specific buyer's questions. Treat the output as a starting point and test the enquiry path yourself.

### Does Google penalise websites made with AI?

Not for being made with AI. Google's spam policies target pages generated at scale mainly to manipulate rankings and not to help users, whether the method is AI, scraping or stitching. A useful AI-assisted page and a useful hand-written page are judged the same way. A thin page is a risk however it was produced.

### How do I know if my website is effective?

Define what counts as a conversion, then track it by page and by traffic source over matched periods. Compare your rate with your own history rather than a borrowed benchmark. Watch a few dozen session recordings to see where visitors hesitate, and submit your own form to confirm enquiries arrive. If the rate is flat while traffic is steady, the site is where to look.

### Test the path, not the palette.

The one hour website check is the verification step that generation does not do for you.

## Looking good is the easy part. The rest is still work.

WP Creative does the website work behind PostCliq. They will tell you when the design is not your problem.

[Talk to WP Creative](https://wpcreative.com.au/)

![Nirmal Gyanwali, founder and chief executive of WP Creative.](https://postcliq.com/authors/nirmal-gyanwali.jpg)

Written by

[Nirmal Gyanwali](https://postcliq.com/blog/author/nirmal-gyanwali)

Founder and CEO, WP Creative · Sydney, Australia

Nirmal Gyanwali has spent more than twenty years building websites, and most of that time watching them fail at the one job the business actually bought them for, which is turning a visitor into an enquiry. He founded WP Creative in Sydney in January 2016.

- [LinkedIn](https://www.linkedin.com/in/nirmalgyanwali/)
- [YouTube](https://www.youtube.com/@nirmalgyanwali)
- [Instagram](https://www.instagram.com/nirmalgyanwali)
- [TikTok](https://www.tiktok.com/@nirmalwpc)

## Keep reading

- [**Trust signals that make B2B buyers enquire**](https://postcliq.com/blog/website-trust-signals)
- [**CRO audit: how to run a conversion audit in an hour**](https://postcliq.com/blog/cro-audit)
- [**Website redesign checklist: the questions to ask before you rebuild your website**](https://postcliq.com/blog/questions-before-you-rebuild-your-website)

[More on Working with agencies and developers](https://postcliq.com/blog/topic/working-with-suppliers) · [All articles](https://postcliq.com/blog)