TLDR; What you need to know:

• This year’s awards were the first where AI was visible in the entries themselves. Judges rewarded AI used with mastery and marked down work that let the machine do the thinking, proving that craft is mastery of the tools, and AI is just one more tool.

• The judging data shows how hard excellence really is: not one of 285 entries scored four or above across all six disciplines, and the average gap between an entry’s strongest and weakest discipline was 2.6 points out of 5.

• An AWIA working group, led by our Judging Chair, is exploring how the criteria should evolve, including automatically generated code and accessibility standards reports for every entry, and it’s open to input from the industry.
  

I’ve spent five years on the AWIA board, but nothing has given me a better view of where this industry is heading than chairing the judging of this year’s Australian Web Awards. And what I saw makes me optimistic, just not in the way you might expect. Here’s a prediction that will annoy some people: AI is about to make great web work more valuable, not less.

This year was the first where AI was unmistakably visible in the entries themselves. When anyone can produce a competent website in an afternoon, competence stops being worth awarding. And craft doesn’t get cheaper just because the tools did. The taste, judgement and mastery behind the best work this year cost what they’ve always cost: years of practice, deliberate decisions, and the discipline to reject the easy default. The winners understood that; the entries that let the machine do the thinking did not. But recognising that difference from the judging table is becoming a discipline of its own, and one we have to master fast. Which leaves AWIA, the judges and me with the real question: how do you design a judging system that fairly rewards the best work, when what the best work looks like is itself being redefined?

Judging 285 Entries With the Same Ruler

Let’s put some real numbers on the table. This year, 81 judges produced 1,710 individual judgements across 285 entries in 32 categories. In the same field: the MCG and Mr Grout, a tradie site built for under $15,000. Redis, a global technology brand, and Maggie, a startup site built by a single developer. Government revenue offices, arts schools, butchers, festivals, IVF research. There is no single yardstick that measures that spread, and pretending otherwise would be unfair to everyone in it.

Our answer has always been structural: we don’t judge categories, we judge disciplines. Every entry, whatever its budget or industry, is assessed across the same six: Accessibility, Content, Design, Development, SEO and User Experience. The category provides the context; the disciplines provide the ruler. Having spent this year close to the criteria and the scoring, I won’t claim the model is perfect; no judging system is. But it’s the fairest way we’ve found to assess work this varied, and we keep refining it.

The proof that those disciplines measure genuinely different things sits in this year’s data. Not a single entry, not one of 285, scored four or above across all six disciplines. The average gap between an entry’s strongest and weakest discipline was 2.6 points on a five-point scale. Excellence, it turns out, doesn’t travel. It’s earned discipline by discipline, specialist by specialist. It’s also why the McFarlane Prize, awarded to the entry that performs best across all six, means what it means: all-round mastery is genuinely rare, and this year Dobre Agency’s work for the Victorian College of the Arts Secondary School earned it.

From the outside, judging can look simple. A panel, some scores, a winner. Under the hood, it’s a different story. Eighty-one judges means eighty-one internal scales, some generous, some brutal; that’s not a flaw in the judges, it’s a fact about humans, and we account for it deliberately: scores are computed and calibrated rather than simply averaged, so where your entry lands doesn’t depend on which judge it lands with. And to say plainly something often misunderstood: as chair, I don’t decide who wins or loses. The judges do, through their scores. My job is to make the process as fair as it can possibly be, and then to stand by what it produces; overriding a result the judges delivered would be unethical. Everything under the hood exists to answer one question: how do we keep this fair at this scale?

The system held this year. But this year, for the first time, the entries started changing underneath it.

What the Entries Told Us About AI

Let me be precise, because precision matters here. AI was not everywhere this year. Of more than a thousand written judge comments, only a dozen or so called out AI explicitly, across eleven entries. This wasn’t a flood. It was something more interesting: the first fingerprints, unmistakable once you saw them, and telling in where they appeared.

They appeared on both sides of the work. On one side, AI as an instrument in skilled hands: an AI-powered site search that judges praised as a smart, modern addition to the experience. A content strategy built on original research that judges noted was exactly what earns visibility in AI systems and LLMs, craft aimed at where audiences are actually going. Used this way, AI didn’t replace the thinking; it extended it.

On the other side, AI as a substitute for judgement. AI-generated hero imagery on a site selling trust, flagged by a judge as working directly against the credibility the copy was trying to build. Blog content described flatly as “shallow and clearly AI generated”. An entry summed up as “a bit too AI”. And one comment I haven’t stopped thinking about, on a site that leaned on AI to save budget: a blunt reminder that in the age of AI, skilled designers aren’t a luxury you automate away; they’re the difference the audience can feel.

The third place AI appeared might be the most significant: inside the judging lens itself. Our SEO judges were already assessing entries for AI Overview readiness and visibility in LLMs, criteria that barely existed two years ago. The ruler is absorbing AI even as the work does.

More than forty judge comments this year flagged work as generic, template-like, or indistinguishable from everything else, with one judge noting a palette so safe it “looks like everyone else”. We are all starting to feel it: an internet converging on the same fonts, the same gradients, the same AI-tinted imagery, the same taste. The tools now make sameness effortless. And that’s precisely what this year’s awards rewarded against: skill in the craft, not quickness with the tools.

Craft Is Mastery of the Tools, and AI Is Just One More Tool

Here’s the position I’ve landed on after this year. The question is not whether AI belongs in award-winning work. Our industry has never been precious about tools: we didn’t disqualify Photoshop, frameworks or CMSs when each of them redrew how the work gets done. The awards have never awarded tools. They award the humans commanding them, and that is a distinction worth defending loudly, because this industry is, and remains, driven by people.

Craft is mastery of the tools. All of them. That includes AI: in the code it generates, the animations it helps implement, the markup it produces. And mastery has a visible signature: code that meets standards, interfaces that work for everyone, markup that’s accessible not by accident but by intent. Prompting a model and shipping whatever comes back is not mastery, any more than installing a default theme ever was. A prompt-and-publish site fails the same way a default-theme site always has: not because the tool is illegitimate, but because no mastery is visible in the result. And there’s a new risk here that didn’t exist in the template era. AI output looks finished. To the novice or the untrained eye, plausible is easy to mistake for good. Accept it at face value and the damage lands twice: audiences quietly lose faith in the brand, and clients lose faith in the agency that let it through. Trust is the one thing on a website you can’t generate.

This is why I’m optimistic rather than defensive. The judges this year didn’t need new rules to tell the difference: AI wielded with intent was rewarded, AI substituting for judgement was named for what it was. The instinct is already right. What it needs now is a system that scales with it.

The honest part is the hard part: that difference gets subtler every year. The obvious tells, the uncanny imagery, the disjointed AI blog posts, are temporary; next year’s defaults will be better dressed. If we want the awards to keep meaning what they’ve meant for twenty-one years, we can’t just trust judges to keep spotting it. We have to build for it.

Raising the Floor So Judges Can Judge the Ceiling

So what do we build? I’m leading a working group within AWIA exploring how the judging criteria should evolve, and I’ll share where my head is at, with the honest caveat that nothing here is settled.

The idea I keep returning to starts from a simple observation: as the tools generate more of the execution, more of the baseline becomes machine-checkable. Whether markup meets web standards, whether code is sound, whether a site meets accessibility requirements: these aren’t matters of taste, they’re measurable. So one option on the table is an automatically generated standards and accessibility report for every entry, produced the same way for all of them, before a human judge ever opens the site.

The data makes the case for it twice over. Accessibility was our lowest-scoring discipline this year, averaging 2.41 out of 5, well below Design, Content and UX. It’s also, historically, our most controversial, and the field carries deep challenges: two accessibility agencies can assess the same component and reach different verdicts, and developers routinely sit between a client who wants the standard met on paper and users with disabilities who need the site to actually work. Compliance becomes the goal; usability becomes the casualty. The industry’s floor needs raising, and nothing raises a floor like knowing it will be measured. Automation won’t settle every judgement call in accessibility, nothing will, but it makes the measurable parts indisputable and strengthens the fairness machinery I described earlier: the objective layer handled identically for every entrant.

But the real prize isn’t the report. It’s what the report frees the judges to do. Every hour a judge doesn’t spend auditing alt tags and heading hierarchies is an hour spent on what no machine can assess: whether the design serves the audience, whether the content earns attention, whether the experience shows intent; whether there’s mastery in the work. There’s a pleasing symmetry in it: the answer to AI in the entries may well include automation in the judging. Machines checking the measurable, so humans can concentrate on the meaningful.

That’s one idea of several, and the working group has a lot of thinking ahead of it. Which is where you come in.

We’re Figuring This Out Together

I won’t pretend we have the answers. Twenty-one years in, the Australian Web Awards are facing the most interesting question in their history, and anyone claiming certainty about what web craft looks like in 2027 is selling something. What I can tell you is where we stand: proud of this year’s winners, who earned it against a field where excellence was genuinely hard to reach. Clear-eyed that telling mastery from machinery gets harder from here. And convinced that the thing worth protecting hasn’t changed: an industry driven by people, and awards that recognise the humans behind the work.

The working group is where that thinking happens next, and it shouldn’t happen in a closed room. If you have opinions about how this industry should judge its best work, and after this post I suspect a few of you do, hit me up on LinkedIn. I’d rather argue about it with you now than explain a decision to you later.

The tools will keep changing. The craft is ours to defend.

Written by: Fredy Lievano, Judging Chair, 2026 Australian Web Awards

The Australian Web Awards are an initiative of the Australian Web Industry Association (AWIA), celebrating the best digital work across Australia for 21 years.