ChatGPT can write product content, but should you let it? Why expert, credible content beats AI filler, and three things to do before you scale it up.
Yes. But Here's Why That's The Wrong Question.
Let me start with a question: can ChatGPT write product content?
Of course it can. And if you haven't tried it yet, honestly, what are you waiting for? Fire up Claude or ChatGPT, paste in a SKU or product photo, and watch it spit out descriptions that sound perfectly decent. Some of you have probably already done this experiment with surprisingly little information and got surprisingly good results.
But here's where we need to talk. Because the real question isn't whether ChatGPT can write product content. The question is: should you let it?
The T-Shirt That Says It All
I saw a t-shirt recently that made me laugh out loud. It said: "In my defence, I was left unsupervised."
That's ChatGPT in a nutshell.
I don't know about you, but I'm seeing this play out everywhere right now. At NRF, retailers were coming onto our stand, telling me they'd been running experiments with AI product content. "It's done quite well," they'd say. "So we're going to push ahead."
And look, I get it. You absolutely can push ahead. Well done for testing. But here's what I want to challenge: what are you actually trying to do?
Is it just ticking the box of product data completeness? Or is there a specific strategic agenda behind why you want to enrich your product content using AI?
Because those are very different things. And only one of them will survive what's coming next in commerce.
When AI Hallucinations Meet Real Customers
Here's a real example from our own work FoundIt! . We just put a load of wallpaper live on B&Q and found that one of our AI agents flagged a wallpaper as appropriate for kids' bedrooms because, and I quote: "The gentle illustrations of soft green hue are calming and whimsical, fitting the criteria for a nursery."
Yeah. Except it had snakes on it. Cult snakes, no less.
When you leave AI to just "get on with it," you genuinely don't know what it's doing. At NRF, the conversation that kept coming up was: How do I trust what AI has done?
Not even what I wanted it to do in the first place, which assumes you had a good idea. But just the basics: how do I trust what it actually did?
This isn't academic. Someone told me about a customer whose goat died after they bought feed based on AI-generated product information. You are accountable for what you write. Full stop.
The Strategic Shift No One's Making
People are using AI to automate the current job that they do.
You've got someone - let's call him Jeff - who sits there, looks at the box, and types up what he sees. Can you automate Jeff's job? Yes. Yes, you can.
But that's not what product content needs to be anymore.
If you were at NRF, the big takeaway was everywhere - even a senior leader of one of our partners" commented on Sundar's speech. Product content is where it's at. Not the future. Right now.
Because customers aren't just searching anymore. They're having conversations. They're asking detailed questions. They're iterating. Those searches are getting longer, more semantic. It's not "black dress" → scroll → click → buy. It's a journey of discovery.
And if your product content is weak - if all you've done is replace Jeff with ChatGPT - you haven't taken your business into the next iteration of how customers actually shop.
Sure, you can fire Jeff and save a few shekels off the bottom line. But you haven't advanced the cause.
Product Content Isn't "Words and Pictures" Anymore
This is the belief system that's killing most retailers: product content is just words and pictures. When we need some, we'll come get them.
It's the old "colouring in department" mentality applied to merchandising. Here's the Pantone swatch book, there's a good boy.
But content in 2026 is fundamentally different. It's not just descriptive fluff. It's the strategic pillar for being shoppable in the next 12 months.
Think about it: as ranges expand - especially for those embracing marketplace - you're making catalogues almost impossible for shoppers to work with. More range on paper looks great. More wallet share, more competitiveness. But it impacts the relative performance of your whole site when customers struggle to actually shop.
A FoundIt! prospect recently told me their NPS score has been dropping. When they dug into why people scored them low, it came down to customer experience. What was wrong with the experience? "Too confusing. Wasn't usable. Wasn't enjoyable to find something."
Your NPS will hurt. Which means your bottom line will hurt.
What Expert Content Actually Means
Let me give you a real example. I was looking at a laptop bag. Nice brown leather, got pockets and a strap, stylish. Great.
But here's what I actually needed to know: how big a laptop can I fit in there? I've got a MacBook - is it going to go in or not? And is it padded? Will it be snug?
None of this was discernible from the product page. I had to work it out from "39cm × 12cm × 16.4cm." Like I'm going to sit there with a tape measure.
This was a first-party product too. They made it. They knew all this information. But absolutely no effort was made to answer the key questions customers would have when looking at a laptop bag.
Because they thought their job was just to describe it. To fluff it up. Give it the jazz hands.
The Validation Pendulum (And Why Both Extremes Fail)
Here's another pattern I'm seeing: the pendulum has swung wildly on validation rules.
It used to be that products got listed without proper attributes. Remember Boots? They had 800 shampoos, and if you selected "women's", it dropped to 16 results. Only 16 out of 800 shampoos were for women? Obviously not. Someone just didn't tick the box.
So retailers swung the other way. More validation rules. More mandatory fields. More checks.
Now, merchandisers are complaining that there are so many validation rules that they can't get products onboarded at all.
Products are hitting the buffers. Everything's getting stuck.
Both extremes miss the point: you need expertise in what you're selling. Not more rules. Not fewer rules. Expertise.
Three Things You Must Do Right Now
If you're still reading this, you get it. Product content isn't a nice-to-have. It's the foundation of everything that's coming next - from LLM commerce to conversational shopping to actually making money from your expanded ranges.
So here are three things every eCommerce director should be doing this month:
1. Ask Yourself: Can We Be Credible In Everything We Sell?
You need to develop expertise in the products you're selling. Not just have a merch team that "manages" fuel ingredients as one of 400 categories in automotive.
I don't just mean hire people with expertise, but the kind that means your product content can answer customer questions, support buying decisions, and work across every channel where customers might discover you.
If you can't be credible, you need to either build that expertise or seriously rethink your range strategy.
2. Define Your Use Cases (Or Stop Wasting Money)
You've heard content is important. Great. But why do you need better content?
When someone comes in and says, "We've hooked it up to ChatGPT, we're pushing the feed out, it's writing all the content" - how do you know whether that's actually solving your strategic needs?
There's going to be a cost to enriching content properly. What do you want it to do for you that will deliver a return? Be specific. Write it down. Make it measurable.
3. Build Your Measurement Framework Now
How are you going to measure your return on improvement in content?
This isn't a free lunch. It's an investment. So track the value you'll see in:
- Conversion rate improvements as customers find and buy more easily
- Efficiencies in paid search from having better structured data
- Improved organic traffic from LLMs and search engines
- Quality of any AI projects you're building (this is your insurance policy)
- Reduced reliance on discounting to move products
The returns come immediately. Content has always been the fastest, easiest thing you could improve. But now it's also the most critical.
The Real Bottom Line
Want to re-platform your PIM? That's two years of your life you're not getting back, and no one ever got a meaningful improvement in their business from changing their ERP.
You want a big win on a silver platter? Its content.
Expert content. Content you can discuss, query, and interrogate. Content that answers the questions customers actually have. Content that makes your expanded ranges an asset instead of an overwhelming mess.
Content that means you can stop playing the discount game and start winning on customer experience.
The future of commerce is conversational, semantic, and intent-driven. Your product content is either going to make you shoppable in that world, or it's going to leave you invisible while your competitors figure it out.
So yes, ChatGPT can write product content. But without strategy, without expertise, without proper guardrails and measurement?
It's just Jeff with a robot arm. And Jeff was never the answer in the first place.
Warren Cowan is the CEO and founder of FoundIt!, the intent-led commerce platform that helps retailers like B&Q, John Lewis, and Neiman Marcus increase their sitewide revenue by 4-6% by making their full product ranges findable and shoppable.
Want to talk about making your content actually work? Get in touch.