“A pleasure to deal with. Highly responsive and agile in their approach”
“A pleasure to deal with. Highly responsive and agile in their approach”
Most agencies talking about AI in marketing are selling a fantasy. Perfect ad creative generated at the push of a button. Zero waste. Infinite scale.
We’ve been running AI-assisted campaigns across energy retail, automotive, and education for over five years. The reality is more nuanced — and considerably more useful — than the pitch.
This article shares what we’ve actually learned: where AI-assisted creative consistently outperforms, where it falls flat, and the one variable that determines almost everything. We’re publishing it because in-house marketing managers deserve real data, not vendor mythology.
The Hype Society manages performance campaigns across three high-stakes industries: energy retail, automotive, and education. Each is characterised by long sales cycles or high customer acquisition costs, regulatory complexity, and intensely competitive paid media environments.
Over the past six months, we’ve been systematically tracking performance differences between pure human-led creative, AI-drafted creative with light human editing, and AI-drafted creative with deep human strategic input. The results have shaped how we now build every campaign we run.
This isn’t a controlled academic study. It’s real campaign data from real clients — with real revenue on the line. Where we reference specific results, they come from active campaigns. Where we reference directional findings, we’ve seen them play out consistently enough to treat them as reliable signal.
Before the data, a definition. When we say “AI-assisted creative,” we mean a workflow where:
This is not “set and forget.” The AI accelerates volume and variation. The human layer determines what actually works and why. That distinction matters enormously — and it’s the core of what we’ve learned.
1st Energy is a tier 3 energy retailer operating across the East Coast of Australia, South Australia, and Tasmania, competing directly against the major retailers in one of Australia’s most price-sensitive consumer categories.
In energy retail, customer acquisition windows are short. Competitor promotions can emerge overnight. The brand that can respond with relevant, high-converting creative fastest typically wins the switching customer. Before AI-assisted production, the creative cycle for a new campaign took days. That’s too slow.
We built an AI-assisted creative production system that allowed us to generate and test new ad variants within hours of a strategic brief. Human strategists defined the switching triggers (price anxiety, distrust of big retailers, desire for simplicity) and AI tools produced multiple headline and visual concept variants against each trigger. Human editors refined the top candidates. The best went live same-day.
The results across the 1st Energy engagement speak for themselves:
In energy retail, AI-assisted creative didn’t just reduce production cost — it changed the competitive dynamics entirely. The ability to test 10 variants in the time it previously took to produce 2 meant we found winning creative faster, killed losers faster, and scaled winners harder.
The human layer that made it work: strategic definition of the switching triggers. AI cannot tell you that “distrust of Origin Energy” is a more powerful motivator than “save on your bill” for a specific audience segment in regional Victoria. That insight comes from strategic human experience — and it’s what the AI creative is built on.
Pakenham Mazda is a family-owned, multi-generational Mazda dealership in Melbourne’s southeast, competing against franchised dealerships, used car superstores, and the growing tide of online car buying platforms.
Automotive dealerships don’t have a lead volume problem. They have a lead quality problem. Google and Meta can generate enquiries efficiently. The hard question is whether those enquiries convert to test drives, and whether test drives convert to sales. A $45 CPL that generates 200 tyre-kicker enquiries is worse business than a $90 CPL that generates 80 serious buyers.
Rather than optimising for raw lead volume, we restructured the campaign around signals of genuine purchase intent. AI tools were used to generate creative variants that pre-qualified audiences — messaging that would naturally filter out browsers and attract genuine buyers. Human strategists defined what “genuine intent” looked like (search behaviour, geographic proximity, model-specific interest, financing readiness) and that informed both targeting and creative strategy.
The creative approach combined Employee Generated Content (EGC) — authentic dealership content featuring the Pakenham Mazda team — with AI-assisted ad copy variants that emphasised the dealership’s family-owned heritage, local community connection, and Mazda Master Dealer status (achieved four consecutive years: 2022, 2023, 2024, 2025).
In automotive, AI-assisted creative works hardest when it’s built around authentic brand differentiation, not generic “great deals” messaging. The Pakenham Mazda campaign succeeded because AI scaled production of creative rooted in real human insights — family ownership, community trust, Mazda expertise built over 50 years. Generic AI output wouldn’t have achieved this. Human strategic direction made the AI useful.
The lead quality finding: campaigns that used EGC and UGC formats alongside AI-assisted copy consistently generated higher call-through rates than image-only or video-only formats. Authentic content builds the trust that turns a click into a call.
Content pending: this section is awaiting the Vic Uni campaign data (enrolment challenge and competitive context, strategic approach and channel mix, CPL/enquiry/enrolment conversion metrics, creative format comparison, and the insight separating enquiry-ready from enrolment-ready audiences) before publishing.
Across all three industries, the same three variables determined whether AI-assisted creative delivered exceptional results or mediocre ones.
AI creative is only as good as the strategic input it receives. Campaigns that briefed AI tools with precise audience insights — specific switching triggers, specific pain points, specific objections to overcome — produced dramatically better first-round variants than campaigns briefed with generic objectives. “Increase sign-ups” is not a brief. “Reach households in South Australia who are 60-90 days from contract renewal and feel overcharged by their current provider” is a brief. The specificity of human strategic input is the variable that matters most.
AI’s core advantage in creative production is volume and speed. That advantage only compounds if testing infrastructure is equally fast. Campaigns with rapid A/B testing cycles — where losing variants were killed within 48-72 hours and winning variants were scaled immediately — dramatically outperformed campaigns that tested on weekly or fortnightly cycles. The AI gives you more horses to race. Human discipline about killing losers and backing winners determines your results.
Across every industry, creative variants that incorporated authentic proof points — real customer testimonials, real performance data, real community signals — outperformed variants that led with feature or benefit claims alone. In energy, this meant real customer switching stories. In automotive, it meant EGC showing actual Pakenham Mazda staff. In education, it means showing real student outcomes, not aspirational stock imagery. AI can generate compelling copy. It cannot generate authentic proof. That’s the human contribution that no tool replaces.
If you’re an in-house marketing manager in energy, automotive, or education, here’s the direct implication of everything above.
AI-assisted creative will make your campaigns faster, more varied, and — if the strategic input is right — more effective. It will not replace the need for human strategic thinking about your audience, your competitive context, and your authentic differentiation.
The agencies that are positioning AI as a production efficiency play are missing the bigger opportunity. The real competitive advantage is using AI speed to test more hypotheses faster — and using human expertise to identify which hypotheses are worth testing in the first place.
The brands and in-house teams that will win the next three years are those that build genuine AI + human workflows, not those that automate their way to generic output.
If you want to understand what that looks like in practice for your industry, that’s exactly the conversation we build every new client engagement around.
Over the coming months, we’ll be publishing detailed breakdowns of campaign performance across each of our core industry verticals. If you’d like to receive these insights directly, subscribe to the Hype Society Knowledge Hub below.
If you’re ready to discuss what an AI-assisted performance campaign could look like for your business, book a strategy session with our team.
The Hype Society is a Melbourne-based performance creative agency specialising in energy retail, automotive, and education marketing. Multi-award winning, with offices in Melbourne, Brisbane, and Auckland.
At TheHypeSociety, we are specialists in fusing strategy with digital performance outcomes. Our performance marketing and strategy teams will ensure that your campaign has the perfect channel mix to achieve your business objectives.
Immerse your audience in a complete end-to-end digital experience – drive leads and sales, re-engage current customers, and build your community.
