Good product data is the only long term bet
AI readiness tools are popping up everywhere. But what does it actually mean for your Shopify store? And how do you prioritise what to do?
There’s a new buzz tool going around: paste in your store URL, get back an “AI readiness score.” Cloudflare shipped isitagentready.com. Shopify has their own at commerce-readiness.shopify.io which prompted me to write this.
It’s bloody great to have these tools. A free scanner that gives merchants a number, a grade, and a short list of things to fix is a brilliant way to get the conversation started. It demystifies a topic that’s been drowning in acronyms — UCP, MCP, GEO, SEO, AEO.
So this isn’t a hit piece on the free tools. It’s a note on what they can and can’t do, and what I’ve learned after 14 years of working with Shopify product data across stores of every shape and size.
The score is subjective. That’s fine.
Every one of these scanners will give you a different number on the same store. Cloudflare’s tool weights robots.txt, MCP endpoints, and emerging agent protocols heavily. Shopify’s cares about ACP and UCP. Instant’s report zeroes in on ChatGPT Shopping fields. ShopAudit looks at llms.txt and product schema.
Run the same store through all of them and you’ll get five different grades. So which one’s “right”?
None of them. All of them. Doesn’t really matter.
A score is a benchmark. It’s a starting line you draw so you can see whether you’ve moved. If you’re on 68 today, the goal is 75 next week. If you’re on 80, push for 90. The specific number is less important than the direction of travel, and the fact that you now have one thing to focus on instead of a vague sense of dread about “AI stuff.”
Scores are useful. Just don’t worship any one of them. And don’t waste your time putting your site through 5 different tools every day. Focus on action.
Prioritise like a merchant, not a perfectionist
Here’s the bit most of these tools won’t tell you: you don’t need to fix every product.
If you’ve got 5,000 SKUs, getting every single one to 100% is a waste of your life. What actually moves the needle is fixing the products that matter most — your best sellers, your highest-margin lines, your deepest-stocked inventory. The top 50 SKUs in most stores drive the majority of revenue. Get those humming and you’re 80% of the way to the outcome you actually care about, which isn’t a score — it’s agent-driven sales.
And get your generic content (shipping policies, returns info, FAQs) all right, because they help boost your entire store.
So when you read your report, don’t start from the top and work down. Start with your winners and make them bulletproof. Then do your next tier. The long tail can wait.
The best bit? Agent readiness is human readiness
This might be the most underrated thing about this whole shift.
In the early days of SEO, you could game your way to the top. Keyword stuffing, dodgy backlinks, hidden text, meta tag bullshit. The content that ranked wasn’t necessarily the content humans wanted to read — it was the content that had been pumped full of tricks.
Agents don’t fall for that. They’re built on top of the same LLMs your customers are using. When ChatGPT recommends a product, it’s looking for the exact same things a human shopper wants to see: a clear product title, an honest description, real dimensions, accurate availability, proper pricing with a currency, decent imagery.
There is no separate “AI version” of your catalogue to maintain. The work you do to make your products legible to agents is the same work that makes them legible to humans. Clean data is clean data. Good content is good content. For once, the incentives line up.
That means the time you spend on this isn’t a tax on top of the rest of your marketing — it’s a multiplier for all of it.
Where the free tools hit a wall
Every free tool I’ve mentioned does the same thing: it diagnoses.
You get a score. You get a PDF. You get a prioritised list of “12 issues to fix.” And then you’re on your own.
Great, so product 1,247 is missing a material attribute. So are 3,200 others. Now what? You’re a merchant, not a data entry clerk. You’re supposed to be running a business, not spending your evenings writing bullet points about thread count.
The real work of AI readiness isn’t identifying the issues — any tool can do that in 60 seconds. The real work is fixing them, across thousands of products, consistently, in a way that stands up against whatever protocol lands next quarter.
Which brings me to Product Pelican
I built Product Pelican to make it easy to audit your product data and then make improvements quickly with the support of AI.
The whole point of AI right now is that you don’t have to do this stuff manually anymore. An audit tells you what’s broken. Product Pelican takes that signal, applies it across your whole catalogue, and fixes it — product by product, field by field. Titles that read properly. Descriptions with substance. Categories chosen from Shopify’s taxonomy correctly. Metadata that stands up against any scanner you throw at it.
And — here’s the thing that’s shaped every product decision I’ve made — it focuses on one job: making your product data genuinely good.
Not optimised for ACP specifically. Not tuned for UCP. Not gaming MCP. Just good. Because when you zoom out and look at what every one of these scoring tools is actually measuring underneath the surface, they’re all circling the same thing. Is your product data rich, accurate, structured, and honest?
Protocols will come and go. There will be a new acronym next Tuesday. Great product data outlasts all of it.
Also- tools like Product Pelican will need to evolve very quickly, because “best practice” is still being defined. Eg. I’m working on an FAQ generation tool next, and looking at how to automate population of Shopify Category Metafields (they currently feel more like Category Minefields to me!).
On Shopify’s readiness tool specifically
A quick word on the Shopify tool… commerce-readiness.shopify.io is a nice marketing moment for Shopify. It shows off what the platform gives you for free — ACP support, UCP integration, native checkout for agents. TBH it’s just Shopify flexing how good its platform is. Shopify has done real work here.
But the tool is built to show you what’s good about being on Shopify, not to be your complete roadmap to AI readiness. It mostly measures things the platform does for you, not the things you need to do with your catalogue. If you’re on Shopify, you’ve already won a lot of what that scanner measures just by being there. The harder work — the product data work — is still on you.
What to actually do this week
If you want a practical take on all of this:
Run one of the free scanners. Pick any of them. Write down your score. That’s your benchmark.
Don’t panic about the score. Don’t try to fix everything. Pull up your top 20 or 50 best-selling, highest-margin products and look at those specifically. Are the titles clear? Are the descriptions actually descriptive? Do you have real attributes — material, dimensions, weight, use case — or just marketing copy? Is availability accurate? Is pricing clean with currency properly declared?
Fix those first. By hand, with Pelican, with whatever works. Then rerun the scanner next week and watch the number move.
That’s it. That’s the game.
The protocols will keep changing. The acronyms will keep multiplying. The scoring tools will keep getting more sophisticated. But the merchants who win the agent era are going to be the ones who stopped chasing the protocol-of-the-month and just built a catalogue worth recommending.
Good data is the only long-term bet.
Connect with me on X, Linkedin, or check out my Podcast, The Shopify App Show.