Your Shopify Product Titles Are Numbers, and AI Can't Read Them

This is one of the most common problems I find in Shopify catalogs, and one of the easiest to fix. You open the product feed, and where the title should be, you find a number. A style code. A SKU. Four digits and a dash. Google Shopping reads a number. Meta reads a number. The AI shopping assistant reads a number. And AI cannot recommend what it cannot read.

This video is about the feed layer. Not your product page copy. The actual data that leaves your store and gets read by machines. Your title field, your attribute fields, the structured record of each product that Google, Meta, and AI shopping assistants pull from. When that record is a number, you are invisible to the systems that decide what gets recommended.

CHAPTER 1: Open your feed right now

Open your Shopify product feed and look at the titles. The actual title field on each product. Not the pretty name on the page. The field.

For a lot of stores, what you find is a number. A style code. A SKU. Something like four digits, a dash, two more digits. That number is the title. That is what leaves your store and gets read by every machine downstream.

Google Shopping reads that number. Meta reads that number. The AI shopping assistant reads that number. And none of them have any idea what your product is, because you handed them a code instead of a name.

AI cannot recommend what it cannot read. That is the whole problem in one line. A shopper asks for exactly the kind of thing you sell, the machine goes looking, and your product is sitting there labeled with a string of digits that matches nothing a human would ever type.

This is one of the most common things I find when I audit a catalog. And it is one of the most fixable. You are not losing because your product is wrong. You are losing because the machine was handed a number and could not read it.

Let me show you exactly what the machine sees, and then how to fix it.

If we haven’t met already, I am the digital avatar of Steve, the founder of Ecommerce Agency working Digital. While he is helping Shopify brands scale with AI, I am here to keep you up to date with the latest AI strategies for ecommerce. Steve reads every comment, so make sure to drop one below.

So let’s get started.

CHAPTER 2: What the machine actually sees

Here is the gap you have to understand. You and the machine are not looking at the same thing.

You look at a product and you see everything. The photo, the context, the brand, the whole story. Your brain fills in all the blanks instantly. You know that number means your bestselling item, because you have lived with it for years.

The machine does not get any of that. It reads fields. Title field, type field, color field, material field. Structured data. And it builds its entire understanding of your product from what those fields say. Nothing more.

So when the title field says a style number, the machine files your product under a meaningless string. It does not know it is a jacket, or a lamp, or a supplement, or whatever you actually make. It knows you have a thing called a number, priced at some amount, and that is it.

Now a shopper asks for the exact category you sell in. The machine scans its catalog for products that match. It is looking for words. Real words that describe real things. Your product, labeled with a code, matches none of them. So it never gets considered. Not rejected. Just never seen, because there was nothing readable to see.

That is how a perfectly good product goes completely invisible. Not a quality problem. A readability problem. The machine only knows what the fields tell it, and your fields told it a number.

CHAPTER 3: Why this happens

Most of the time it comes from how the products got into Shopify in the first place. They were imported from an old system. An ERP. A wholesale catalog. A spreadsheet from a manufacturer. And in those systems, the style number is the identifier. That is how the back office tracks inventory. So when the data got pulled in, the style number landed in the title field, because that was the primary key in the old world.

Nobody made a bad decision. The operations side needs that number. It is how they pick and pack and reorder. The problem is that the number leaked into the field that the outside world reads, and nobody caught it, because on the website the theme shows a nice name anyway and everything looks fine to a human.

So you get this split. The page looks great. The feed is full of codes. And for years it did not matter much, because the main thing reading your feed was Google Shopping, and you maybe were not leaning hard on that channel.

Now it matters a lot. Because AI shopping assistants read that same feed data, and they are becoming the way people find products. The old mistake that did not cost you much just became the thing standing between you and the fastest-growing channel in commerce. The good news is it is a data problem, and data problems get fixed.

CHAPTER 4: The feed is not the page

I want to draw a hard line here, because people mix these two things up all the time.

This is not about your product page copy. The beautiful description, the lifestyle photos, the story on the page. That is a different layer, and it matters, but it is not what we are talking about today.

Today is about the feed. The structured data. The fields that leave your store and get consumed by machines. Google Shopping pulls the feed. Meta pulls the feed. AI shopping assistants pull structured product data. They are not reading your page like a person reads it. They are reading fields.

You can have gorgeous page copy and still have a feed full of numbers. The two are not connected the way you think. The page is for humans. The feed is for machines. And the machines are the ones deciding what gets recommended now.

So when I tell a store their titles are numbers, sometimes they push back and say the product pages look amazing. They do. That is not the point. The amazing page is invisible to the feed. The feed has its own title field, and if that field is a code, the machines never get to the amazing part.

Fix the feed layer specifically. The title field. The attribute fields. The structured record. That is the layer the machines read, and that is the layer that is broken. Which brings me to what a fixed record actually looks like.

CHAPTER 5: The golden record

Here is the standard to aim for. I call it the golden record. Every product has a complete, human-readable record that a machine can fully understand.

Start with the title. It says what the thing is and who makes it. Brand, product type, the key distinguishing attribute. A real name a human would recognize and a machine can match to a search. Not a code. A name.

Then every attribute field, filled in. Color. Material. Size. Category. Use case. Gender or age range if it applies. The full set. Every field the feed supports, populated with accurate, readable values. So that when a machine reads your product, it reads a complete picture. A jacket, this brand, this material, this color, this size, for this use, in this category.

That is a golden record. Complete and readable, every field doing a job. And here is why it wins. When a shopper describes what they want, the machine matches their words to your fields. The more complete and accurate your fields, the more questions you can match. A thin record matches a few queries. A golden record matches everything the product genuinely fits.

One product with a golden record is good. A whole catalog of golden records is the moat. The machine can read every single thing you sell and match it to every relevant shopper. That is the goal. Now let me walk you through how you get there.

CHAPTER 6: The fix, step by step

Here is the order of operations. It is not complicated. It is just work.

Step one. Audit the feed. Pull your product data and find every product where the title is a number, a code, or a near-useless string. You will probably be surprised how many. Flag them all. That is your worklist.

Step two. Rewrite every title into a complete, readable name. Brand, product type, key attribute. Write it the way a customer would describe the thing out loud. If you sell it, you know how people talk about it. Use their words, not your warehouse codes.

Step three. Fill in every attribute field. Go product by product and complete the record. Color, material, size, category, use case, all of it. Do not leave fields blank. A blank field is a question the machine cannot answer about your product, and every blank is a shopper you cannot match.

Step four. Keep the style number, but move it where it belongs. Put it in the SKU field or an identifier field, so your operations side keeps the number they need, and the title field gets freed up for a real name. Everybody wins. The warehouse keeps its code. The machines get their name.

Do those four and you have gone from invisible to readable across your whole catalog. The products that no machine could see are now products every machine can read and match. That alone changes what shows up when a shopper asks. But there is one more step, because clean data in Shopify only matters if it actually reaches the machines.

CHAPTER 7: Make the clean data flow

Fixing the data in Shopify is step one. Making sure it flows out clean is step two, and people miss it.

Your clean titles and complete attributes have to make it all the way through to the places machines read. Your Google Shopping feed. Your Meta catalog. The structured data the AI surfaces pull from. Check each one. Sometimes the title is fixed in Shopify but an old mapping or a stale sync is still pushing the number downstream. The warehouse of clean data does you no good if the delivery truck is carrying the old junk.

So verify the flow. Look at what actually lands in Google Merchant Center. Look at what Meta ingests. Make sure the readable title and the full attribute set are what shows up there, not the codes you just cleaned out of Shopify. Close the loop end to end.

And then keep it clean. New products come in the same broken way the old ones did, imported with codes in the title field. So build the golden record into how you add products. Make the complete, readable record the standard for every new item, not a cleanup project you do once and let rot. The catalog that stays clean keeps winning. The one that drifts back to codes goes invisible again.

This is the unglamorous plumbing of AI commerce. Clean data, flowing clean, staying clean. It is not exciting. It is just the thing that decides whether the machines can read you at all.

CHAPTER 8: Boring work that pays

I will be honest about what this is. It is boring. It is spreadsheets and fields and checking feeds. Nobody gets into ecommerce because they love populating attribute data.

And it is some of the highest-return work you can do right now. Because the payoff is enormous and the problem is so fixable.

Shopify turned agentic storefronts on by default for eligible US merchants in 2026. That means machines are carrying shoppers all the way to checkout now. And AI shoppers convert at around four times the rate of other traffic. So the channel reading your feed is the highest-converting channel there is. If your feed is full of numbers, you are locked out of that entire channel over a data problem you can fix this week.

Think about the trade. A few days of boring work against being readable in the fastest-growing, highest-converting way customers find products. That is not a hard call. That is the best return on a boring task you will find all year.

And almost nobody is doing it. Everybody is busy with the fun stuff, the brand, the photos, the launches. Meanwhile the feed sits there full of codes, invisible to the machines. There is so much room here precisely because it is boring and most people skip it. That is your opening.

So go open your feed. Look at the titles. If you see numbers, you just found work that pays. Rewrite the titles, fill the attributes, check the flow, keep it clean. Go from invisible to readable, and let the machines finally see what you sell.

If you want us to pull your feed and show you exactly which products are titled with numbers and what the fix looks like, we do a free audit at wrkngdigital.com.

Next video, I get personal. I turned off all my paid ads, and AI started sending me customers. I will walk you through exactly what happened and why it changed how I think about where leads come from.