Job store-product-attributes-extracted-from-descriptions · revised 12 October 2026
Fill your Shopify store's missing colour, size and material fields from your descriptions
Missing attributes for up to 200 Shopify products from your descriptions, in a metafield-ready import file with a dry-run diff. Each value has its quoted source; unknown stays blank.
You might be seeing
- Collection filters miss products because their colour or size field is empty
- Filling fields by hand takes days
No passwords, keys, card details or admin invites needed to start.
What usually happened
The facts are often in the description text but not in the field that filters and feeds use. Guessing them risks wrong values, and invented barcodes can get listings disapproved. Extraction is checkable because each value either appears in the text or does not.
Who it’s for: Owner or catalogue manager whose filters, shopping feeds or marketplaces need colour, size, material or compatibility fields that the supplier sheet leaves blank.
Usually starts when: Filters show gaps, a feed is disapproved for missing attributes, or you face hundreds of blank fields to fill by hand.
The result: Missing attributes are filled for up to 200 Shopify products from your descriptions and supplier data, each value with its quoted source text, in a metafield-ready import file. Unknown stays blank and is counted. A dry run on a test store changes only the agreed fields.
Check whether this job fits
Answer these without sending files or logins. Nothing is submitted unless you choose to contact us.
What you get
- The filled attribute sheet with quoted source text
- A count of blanks left
- The import file
- The dry-run diff
- The labelled test and its results
Included
- Up to 200 products from a Shopify export, with up to six attribute fields in total, grouped into the category-specific sets you choose (for example colour, size, fit and material for clothing; compatibility for parts)
- A value only where the source text states it, with the quoted words
- Unknown left blank and counted, never guessed from images or other brands
- No barcodes or GTINs invented
- A Shopify import file with the values in metafield-ready columns, dry-run on a test store you own, with a before and after diff
Not included
- Importing into your live store
- Guessing values from images
- Inventing GTINs or barcodes
- Rewriting descriptions
- Category mapping (see the supplier category job)
- More than 200 products, or a list that is not a Shopify export (see the feed-wide attribute job)
How we know it’s done
Agreed with you before work starts. Each check produces evidence you keep.
Of the 25 products you keep sealed until the test run, at least 24 have every field matching your labels (a true rate of at least 82% with 95% confidence). A field you labelled 'unknown' must stay blank, and a field left blank that you labelled with a value counts as a miss.
Evidence: Your labels beside the output for the 25 sealed products.
Every filled value carries quoted source text found in the description or supplier data.
Evidence: The attribute sheet; you check ten rows you pick.
No invented GTIN or barcode appears; one fails the whole test.
Evidence: A search of the output for barcode fields.
Unknowns are left blank and counted; a dry-run import changes only the agreed attribute columns.
Evidence: The blank count and the import diff.
Sign-off. You re-run the acceptance checks with the steps we hand over, read the results and sign off in writing. Payment follows sign-off.
If it fails. If an agreed check fails, you do not pay for this fixed scope. We hand over what we found and agree whether to stop or re-quote; no surprise work.
When it fits, and when we stop
It fits when
- Descriptions or supplier columns state the facts to extract
- You can export products from Shopify as CSV with no customer data
- You hand-label 50 products, 25 of them held back from us until the test run; two people on your side label the same 20
- You own a test store or copy for the dry run
We stop and tell you if
- You want values guessed from images or from other brands' pages
- Two people on your side label the same 20 products and agree on fewer than 17
- The descriptions state almost none of the facts
- You want barcodes or GTINs created
What could go wrong
We change a test store or test account only. Nothing is switched on in your live store or helpdesk by us. Your named person applies the result and can switch it off again with the steps we hand over.
Scroll the table sideways to read it all.
| Risk | How we handle it |
|---|---|
| A wrong value is filled and shows on a filter. | Every value carries its quoted source text, and the held-back labelled set measures the error rate. |
| A made-up barcode causes feed disapproval. | Barcodes and GTINs are never created; any in the output fail the whole test. |
| Values on the page and in the feed disagree afterwards. | The import file touches only the agreed fields, and the diff shows nothing else changed. |
A separate automated reviewer checks the work and the test results before hand-over. Your named person approves the values and imports the file. No human supervisor from us is included unless your agreement names one, and at launch the work is largely automated.
How we deliver
We arrange the work and independent review, then show you the result against the agreed checks. You keep authority over your systems.
- Agree the fields, the labelled set and the approved import columns
- Receive the product export and seal your held-back labels
- Extract each value with its quoted source text; leave unknowns blank
- Compare with your labels and run the dry-run import
- A separate automated reviewer checks blanks and invented values
- Hand over the sheet, import file and results
This is a one-off job, not emergency cover or a subscription. We confirm eligibility, price, start window and delivery date in writing before you accept, and work starts only after the agreed inputs are in place. An enquiry creates no charge or booking.
Need to keep it working?
Discuss filling attributes for each new batch of products.
Ongoing work is separately scoped and quoted: no monitoring, response-time guarantee or automatic subscription is included in this job.
Explore an ongoing managed lane, or mention the responsibility you need in your enquiry.
What you can check
This is a new service. We have not delivered this job for a client yet.
Other ways to get this done
- Fill the fields by hand with a spreadsheet filter, product by product.
- Ask the supplier for a complete data sheet.
Questions
Will it read my product images?
No. Values come from text only. Guessing from images is outside the scope.
What about missing barcodes?
We never create them. Products without one are listed.
How is the result checked?
Against your own labels, half of which we do not see until the test run. The pass mark counts only the sealed half.
Does a pass mean it will be right next month?
No. A pass describes items like the test sample on the day of the test. It is not a promise about future items, and it does not replace your own checking.
How is this different from the feed-wide attribute job?
That job fills any spreadsheet or ERP list of up to 20,000 products and tests on 50 held-back products. This one is for a Shopify export of up to 200 products: it gives you category-specific attribute sets, a metafield-ready import file and a dry-run diff on a test store.
Send an enquiry
Send us
- Which attribute fields you need filled
- Roughly how many products, and a few example descriptions
- Where blanks hurt you (filters, a feed)
- No store login
Later, once you agree
- A product export as CSV through an agreed secure route
- Your 50 hand-labelled products (20 of them labelled by two people), with 25 sealed until the test run
- A test store or import copy
- The name of the person who approves the values
You keep your store, your helpdesk and every login. We work on the files you send through a secure route agreed after you say yes, and hand back files and a report. We never ask for store or helpdesk logins, and nothing goes live until your named person applies it.
Email fallback: open your mail app
If website submission is unavailable, review and send the fallback email yourself. An email fallback is not a website receipt. Or write to hello@syntheticindustry.ai with “store-product-attributes-extracted-from-descriptions” as the subject.