There are some boring but essential adult decisions I can’t seem to muster the required energy to get excited about: replacing a mattress is one of them.
In the days after Christmas, marking nearly two years of sleeping on a second-hand mattress rescued from a friend’s grandparents’ spare bedroom during a hasty move, I realised it was sagging so badly that it was actually starting to create back problems. I finally faced what I considered to be an uninteresting and tedious task of researching, considering, comparing and purchasing a new bed.
Only I didn’t need to do all the research on my own – hardly anyone does any more. Despite being a self-avowed AI hater, I consulted what seems to be becoming everyone’s new best friend: ChatGPT. I’m an able-bodied, healthy-ish journalist renting in Australia’s most expensive city; I don’t need something particularly fancy, plus I want the dopamine hit that comes when you feel you’ve stumbled on a bargain.
What’s the best mattress I can find for around $600? Give me your top three suggestions based on quality and value, and incorporate real product reviews and user comments to inform your answers.
ChatGPT gave its best three suggestions, each of which I looked at. After some brief deliberation, the Newentor Hybrid Mattress was decided upon and paid for – all in under 10 minutes, and for $59 below budget. [Less than two weeks] later, I was beaming as I pulled fresh sheets over my new medium-firm, seven-layer mattress, delighted that an arguably snap decision had turned out so well.
Not everyone gets it right. While nearly nine in 10 (86 per cent) of Australian consumers have used AI for shopping decisions, according to US marketing technology company Klaviyo, some consumers have admitted to returning impulse purchases made during major sales periods over poor product quality, shoddy descriptions or due to finding a lower price later.
Retailers around the world are rushing to leverage and integrate AI in both the consumer-facing and back-end parts of the business, so that we buy and consume faster and easier than ever before.
To help us get the most out of AI, we asked retail, ecommerce and AI experts how we should use ChatGPT or other AI interfaces to shop:
Specific questions get better answers
AI chatbots are only as good as the prompts and the information you give them, so be detailed and descriptive about what you’re looking for.
“AI tools work by matching intent to product attributes such as price, size, features and suitability,” said ecommerce software platform Athos Commerce marketing manager Sherrine McInerney.
“When a shopper is vague, the system fills in the gaps with broad, often unhelpful results. When a shopper is specific, the recommendations improve dramatically.”
AI shouldn’t entirely replace our purchase decision-making, but it’s useful for saving time if it is used like a research shortcut to narrow down the best options and prices, said RCA Advisory director Trent Rigby.
“I personally use it to shortlist products, compare options and understand trade-offs (price versus quality versus reviews). For example, why one product is cheaper, or what you gain or lose by spending a bit more, then I’ll verify the final decision myself,” he said.
“If you already know the product you want, AI can save a lot of time jumping between tabs. It won’t always have the perfect price history, but it’s very good at spotting obvious overpaying.”
Do: Use words like “compare”, “summarise”, “explain the differences”; “minimum 4-star rating”; name your budget and currency.
Don’t: Plug in vague terms like “good vacuum” or “best TV”.
Sample prompt: “I’m looking to buy [product name and model] for [context e.g., family of four, small apartment, budget-conscious]. Compare major retailers in Australia for price, quality and reviews, flag any sales or discounts, and summarise the trade-offs. Based on pricing history over the past 365 days, tell me if this looks like a good time to buy. Recommend the best value option.”
Give context
The more information you provide about your needs and circumstances, the better the results. When shopping, AI consultant and Tribe Gen AI chief executive Kelly Slessor said she treated ChatGPT more as a personal stylist or buyer than as a search box.
“The magic is in describing who you are and how you live, then adding constraints like budget, vibe, practical needs,” she said.
Slessor has found that ChatGPT suggestions don’t always localise well, so adding location and shipping constraints will help you avoid disappointment after landing a great find, only to discover it doesn’t ship to Australia.
Do: Be detailed. When buying household appliances, consider: “Pets? Kids? Small kitchen? Apartment? Budget? What do you actually care about (quiet, durable, easy to clean)? AI shines when it understands your situation,” said McInerney.
Don’t: Forget to use common sense. “Avoid sharing sensitive personal information or images you wouldn’t normally be comfortable putting online,” said Rigby.
Sample: “Create a capsule wardrobe for autumn. I am nearly 50, trying hard to be cool, trendy and maybe a little bit sexy; I present on stages a lot and spend too much time in Zoom meetings.”
“I used this exact prompt to build me a capsule wardrobe and it absolutely nailed the brief,” said Slessor.
Don’t want buyer’s remorse? Say so
A quick decision doesn’t have to be a hasty one if you’ve accounted for pitfalls. This is particularly handy for bigger-ticket purchases like tech, skincare, furniture (or mattresses): prompt ChatGPT to include pros and cons, weaknesses, or negative reviews. “They force the model to think like a critic,” said Slessor.
McInerney pushes ChatGPT to justify itself. “If AI can’t explain why something is better, that’s a red flag,” she said.
“If an AI assistant cannot clearly explain the recommendation, that’s often a sign of weak product information, inconsistent reviews, or marketing noise outweighing substance.”
Do: Ask why. “Why is this product better than the cheaper option? Why does it suit my needs? What are the common complaints?” said McInerney.
Don’t: Forget to iterate. “AI is designed to be conversational. Refine your question if the first answer isn’t useful,” said Rigby.
Sample: “Help me buy anti-ageing cream with the goal of minimising regret. Give me:
- the five most common ‘I wish I’d known…’ pitfalls
- the six key specs that actually matter (not marketing)
- five products that dodge those pitfalls
- a one-sentence ‘who this is for’ for each.”
Try the virtual dressing room
It’s not quite like trying on the real thing, but generative AI can help you visualise what clothes or accessories look like on you, how decor fits in your living room, or plants in your garden, before you buy.
“This one’s a bit more controversial and really depends on who you speak to or how much you trust AI,” said Rigby. But the quality of AI image generation has improved a lot and is improving all the time, he said.
Do: Be cautious about uploading personal or private images. Rigby personally uses only pictures of himself that already exist online, and he turns off settings that permit images to be used for AI training, where possible.
Don’t: Treat the AI images as a guarantee, or upload anything that you wouldn’t want publicised. “My general rule of thumb is the ‘billboard test’; if you wouldn’t be comfortable with that image appearing on a public billboard, don’t upload it,” said Rigby.
Sample: “Here’s a photo of me. Show me what I’d realistically look like wearing the attached image(s). Keep proportions natural and lighting realistic. Show front, side and back views.”
Always ask for a discount
You’d be surprised at how often people will let you have something for a discount or for free if you just ask (charmingly, of course). AI can help you unearth discount codes, and it takes just seconds to ask.
Sample: “I’m about to check out online on [retailer name] for [product]. Are there any current promo or discount codes available for Australian shoppers?”
The Business Briefing newsletter delivers major stories, exclusive coverage and expert opinion. Sign up to get it every weekday morning.
Jessica Yun is a business reporter covering retail and food for The Sydney Morning Herald and The Age.Connect via Twitter or email.
