Coveo paid for an AI that fills in missing product details. Most Canadian stores we checked have not filled in the ones they already know.
None of the 37 Canadian online stores we checked on October 31 states what its product is made of in the product markup that search engines and AI assistants read, and that is the gap Coveo says it bought Anglera to close. The deal is product data enrichment: software that reads spec sheets and supplier files, then fills in the missing details.
Read the footnote before the headline number. Coveo’s October 29 release credits Anglera-enriched pages with a 270% rise in AI-referred visits, and the footnote says that came from about 300 product variants at one apparel retailer over 30 days, “reported by Anglera”. Three words follow it: “Results may vary.”
Key points
What Coveo’s product data enrichment deal buys
Coveo, the Montreal search company listed on the TSX, acquired Anglera on October 29, 2026 and will sell its product data enrichment as an add-on to Coveo Commerce. Anglera’s software pulls attributes such as dimensions, materials and compatibility out of manufacturer PDFs, images and supplier files, then checks them against the retailer’s catalogue rules.
The pitch is aimed at manufacturers, distributors and large retailers, and the logic is simple. An AI shopping assistant cannot recommend a wool coat it cannot tell is wool. The Coveo release also cites a NielsenIQ estimate that missing attributes can cut discoverability by up to 30%. (Terms were not disclosed, so the add-on’s price is the first question for any sales rep.) Coveo is one of the names in our Canadian AI companies hub.
What 37 Canadian product pages told a machine
Most told a machine the name, brand, price and stock, and stopped there. We fetched one product page from each of 37 Canadian brand stores on October 31 and read the product markup in the raw HTML, the version Google recommends for best results in shopping. 31 carried product markup. None listed material.
The gap is not missing knowledge. 20 of the stores sell that product in colour options, and 18 of them leave colour out of the markup. 29 sell it by size, and 26 leave size out. The store knows. The page does not say. Kamik and Vessi were the only two to mark up both colour and size.
Google lists colour, size, material and GTIN as recommended product properties in its merchant listing documentation. It costs nothing to read, and it names every field our check counted. If you already chose which AI crawlers may read your site after our robots.txt check of 31 Canadian sites, product markup is the next file to open.
Five product data fixes, ranked from cheapest to dearest
Start with what your catalogue already holds, then pay people, then pay for product data enrichment software. Each step below costs more than the one before it, and only the last one needs a vendor like Coveo. Most Canadian brand stores in our check have not finished step one.
- Colour and size you already sell by. One template change exposes options the store already stores. Verdict: do this first. It costs a developer hour or an app setting.
- GTIN barcodes. 17 of 37 publish one. If supplier barcodes already sit in your product records, this is the same template job. Verdict: cheap, and it ties your listing to the exact item a buyer asks about.
- Material, typed by a person. Nobody in our check had it. It lives on care labels and spec sheets, so someone has to read them. Verdict: worth paying for in clothing, furniture and anything with a safety or allergy question.
- Weight and dimensions. One store published weight, and none published dimensions. Dimensions often mean a tape measure and a spreadsheet. Verdict: do it for bulky goods and B2B parts, skip it for socks.
- An enrichment platform. Coveo with the Anglera add-on, or a rival. Price not published. Verdict: dearest, and only sensible when the catalogue is too big or too technical for the first four steps.
Ten hours of clerk time against a 270% headline
Anglera’s own test size can be filled in by hand for about $235. Job Bank puts the Canadian median wage for data entry clerks at $23.50 an hour. Assume, as an illustration, two minutes per variant to read a label and type material, colour and size.
| Catalogue | Clerk hours | Cost |
|---|---|---|
| 300 variants | 10 | $235 |
| 1,500 variants | 50 | $1,175 |
| 1,500 at 6 min | 150 | $3,525 |
The assumption that moves the total is minutes per variant. At two minutes, a 1,500-variant clothing store pays about $1,175. If the label is missing and someone has to chase the supplier, six minutes triples it to $3,525. Either figure is a budget line a 20-person Calgary or Halifax retailer can approve this month, using Job Bank’s wage data for its own province.
When a 40,000-part catalogue changes the answer
A distributor with 40,000 parts, our illustration, is a different business. Coveo’s release says Anglera was built for catalogues where products carry hundreds of technical attributes and accuracy matters. At that size, the six-minute row becomes 4,000 hours, and a clerk typing torque ratings from PDFs will make mistakes nobody catches.
Anglera keeps a source reference for each attribute it fills, which a buyer should value: you can trace a wrong spec back to the file it came from. Our check also has limits. We read one page per store and only its markup. Google Merchant Center feeds and Shopify Catalog also send product data to AI shopping tools, and some of these stores may already send material that way.
For a manufacturer or distributor, paying for enrichment can make sense. For most Canadian brand stores in our check, the first two fixes on the list need a developer before they need product data enrichment. Our earlier look at agentic commerce in Canada set out why AI shoppers read data instead of photos, and our guide to Shopify’s AI shopping agents covers the settings side.
FAQ
What is product data enrichment?
It is filling in missing product details, such as material, size, colour, dimensions and barcodes, so search engines and AI shopping assistants can match a product to a question. Tools like Anglera read spec sheets, PDFs and supplier files to do it with AI; a person can do it by hand.
What did Coveo buy?
Coveo, a Montreal company listed on the TSX, acquired Anglera, a San Francisco startup backed by Y Combinator, on October 29, 2026. Anglera’s enrichment will be offered to Coveo Commerce customers as an add-on. Terms were not disclosed.
Does product markup decide whether ChatGPT or Gemini recommends my product?
Product page markup is one channel among several. Google’s merchant listing guide recommends colour, size, material and GTIN properties, and Shopify says stores with agentic storefronts also send product data through Shopify Catalog. Check both.
How much does product data enrichment cost by hand?
At the Canadian median of $23.50 an hour for data entry clerks and an illustrative two minutes per variant, 300 variants take about 10 hours, or $235. A 1,500-variant catalogue costs about $1,175, or $3,525 if each item takes six minutes.
Our line for a Canadian brand store: if your gaps fit inside one clerk’s month, which we put at 160 hours or about $3,760 at the national median, fill them in-house first. Past that, a platform quote deserves a meeting, with the vendor’s footnote read before its headline.
Facts in this column were checked on October 31, 2026 against Coveo’s October 29 release, Google’s merchant listing documentation and Job Bank. Store results come from our own check of 37 product pages that morning. We did not test Anglera or Coveo Commerce.
