Job uncategorised-products-assigned-to-your-category-tree · revised 12 October 2026
Put my uncategorised products into my category tree, with doubtful ones sent for review
Each product gets one leaf category from your tree of up to 300, or goes to 'review'. Tested on products you labelled; categories not in your tree are never invented.
You might be seeing
- Products sit in a catch-all category and cannot be found
- Menus and filters are empty because nothing is categorised
No passwords, keys, card details or admin invites needed to start.
What usually happened
A model asked to categorise will sometimes make up a category that is not in your tree, or choose a plausible wrong one. Category choices are also partly a matter of taste, so agreement has to be measured against your own labels.
Who it’s for: Merchant with thousands of products that arrived without a category in your tree.
Usually starts when: A catalogue import left products uncategorised and sorting them by hand would take days.
The result: Each product gets one leaf category from your list or the value 'review', never a category that is not in your tree. On products you labelled and we have not seen, the share placed correctly is reported, and doubtful products go to review and count as misses.
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 product file with a category or 'review' on every row
- A check that every category exists in your tree
- A results table on the held-back products
- The review list
- The script
Included
- One product file of up to 20,000 products and a category tree of up to 300 leaf categories you own
- One leaf category or 'review' for each product
- A held-back test on 50 products you label
- Per-category results where there are enough held-back products, and a list of weak categories
- The script and the steps to run it
Not included
- Creating, renaming or merging categories
- Resolving disputes about what a product is
- Importing the result into your store
- Marketplace category rules you do not supply
- Images, or text that is not in English
How we know it’s done
Agreed with you before work starts. Each check produces evidence you keep.
Of the 50 held-back products, at least 45 get the category you labelled; review counts as a miss. 45 of 50 shows the true rate is at least 80% with 95% confidence. The report also shows the score from giving every product the most common category, and the pass counts only if our score beats it by the margin agreed in writing.
Evidence: The results table beside your labels.
Every category in the output exists in your tree. One invented category fails the whole test.
Evidence: The script check.
5 products that fit no category, which you add and we do not see beforehand, land on 'review'.
Evidence: The planted list and the output.
Per-category results are reported for categories with at least 5 held-back products, and any of those with fewer than 60% correct is listed as weak; other categories are listed as too few to judge.
Evidence: The per-category table.
Sign-off. You inspect the results and the review list, and sign off in writing. Payment follows sign-off.
If it fails. If the held-back test is not passed, you do not pay for this fixed scope. You keep the results, the failures and the test set, and we agree whether to stop or re-quote. No surprise work.
When it fits, and when we stop
It fits when
- You own a category tree of up to 300 leaf categories
- You label 100 products: 50 for tuning and 50 held back
- Two people on your side agree on at least 17 of 20 sample products before we start
- Product text is in English and holds no personal data
We stop and tell you if
- The category tree is disputed inside your team
- You want new categories invented
- A marketplace taxonomy applies that you cannot supply
- The two labellers agree on fewer than 17 of 20
What could go wrong
The job reads copies you send and writes files. It changes nothing in your systems, so there is nothing to revert. Our copies of your files are deleted when the job ends.
Scroll the table sideways to read it all.
| Risk | How we handle it |
|---|---|
| A product is placed in a plausible wrong category. | Doubtful products go to 'review'. Held-back accuracy is measured on your labels, and review counts as a miss. |
| A category appears that is not in your tree. | A script checks every category against the tree. One invented category fails the whole test. |
| Your files go to a model provider outside your control. | Before any file is shared we name the provider, its country and its retention terms in writing. Entry jobs run on redacted, synthetic or public copies, or on your own provider account with a key you hold. |
A separate automated reviewer checks the assignment rules and the tree check before hand-over and does not mark the held-back test. A named person on your side labels the test set and signs off. No person on our side 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 tree, the pass mark and the severe errors in writing
- Two of your people label the same 20 products to check they agree
- You label 100 products; 50 stay held back until the test run
- Build the assignment and run it on the held-back products
- Independent review of the results, the review list and the tree check
- Hand over the file and script; delete our copies
This is a one-off job, not emergency cover or a subscription. We confirm eligibility, the price, a start window and a delivery date before you accept, and work starts only after inputs, a secure handoff route and permissions are in place. Model usage up to a cap agreed in writing is included, and an enquiry creates no charge or booking.
Need to keep it working?
Need new products placed as they arrive? Ask about a quoted re-run.
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
- If your supplier feed already carries a category, the supplier category mapping job handles it with a table and an unmapped list.
- Google Merchant Center marks its product category as optional and has its own automatic categorisation, so you may not need this for Google Shopping. support.google.com
- For one column of messy descriptions sorted into up to 30 categories of your own, use our text-column sorting job. This job is for a product file and a tree of up to 300 leaf categories.
Questions
Will it create new categories?
No. Only categories in your tree are used. Anything doubtful goes to 'review'.
How is accuracy checked?
On 50 products you labelled and we did not see. The result describes products like the sample.
Do I need this for Google Shopping?
Google says the category is optional. Check what each of your channels needs first.
Send an enquiry
Send us
- The category tree as a list, and how many leaf categories it has
- How many products are uncategorised
- No real records, personal data or credentials in the first enquiry
Later, once you agree
- A copy of the product file by an agreed secure route
- Your labels for 100 products
- A named person who decides the review list
- A written cap on model usage
You keep your product file, your tree and your store. We work on a copy you send by an agreed secure route, return the file and a script, and delete our copies when the job ends. We never import anything.
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 “uncategorised-products-assigned-to-your-category-tree” as the subject.