Job store-duplicate-products-review-list · revised 12 October 2026
Find products listed twice and variants that should be merged, as a list for you to decide
A review list of one style listed under two titles with split stock, and variants that belong together, from your product export. Near-misses kept apart. Nothing is merged.
You might be seeing
- One style appears under two titles with different stock
- A supplier import created a second copy of products you already had
No passwords, keys, card details or admin invites needed to start.
What usually happened
The same product often arrives under a different title, handle or variant set after an import. Exact matching misses these, while loose matching merges items that only look alike, such as two sizes. A wrong merge loses stock history and order links.
Who it’s for: Owner or catalogue manager of a store that has imported from several suppliers or stores and now has the same product listed more than once.
Usually starts when: Customers see two listings for one item, stock is split between them, or a merge by hand takes too long.
The result: From your product export you get a review list of probable duplicate pairs and variants that may belong together, each with the fields that matched. It is tested on 20 planted pairs, 10 of them yours and kept from us, and 10 near-miss pairs that must not be listed. Nothing is merged or changed.
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 review list
- Near-miss list
- Matching rules, written down
- Planted-pair test results
- A script to re-run it
Included
- Up to 5,000 products from one export, including variants, SKUs, barcodes and image addresses
- Matching on identifiers first, then on normalised titles and image addresses
- A review list of pairs with the matching fields shown and a confidence note
- A separate list of near-misses that look similar but differ in a size, colour or pack count
- A planted-pair test on a copy of 300 of your products, with half the pairs made by you
Not included
- Merging, deleting or archiving products
- Changing stock or orders
- Deciding which listing to keep
- Images compared by sight
- Lists of people or customers
- Real customer data
How we know it’s done
Agreed with you before work starts. Each check produces evidence you keep.
In a 300-product copy with 20 duplicate pairs, 10 planted by us and 10 by your named person and kept from us until the test run, all 20 pairs are on the list (20 of 20 shows a true catch rate of at least 86% with 95% confidence). One missed pair fails the test.
Evidence: The planted list beside the review list.
Of 10 planted near-miss pairs (5 ours and 5 yours, differing in size, colour or pack count), none is listed as a duplicate; one listed fails the whole test.
Evidence: The near-miss list.
Every pair on the list shows the fields that matched.
Evidence: The review list.
The script, run twice on the same export, gives the same list.
Evidence: Both outputs compared.
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
- You can export products and variants as CSV
- Products carry SKUs, barcodes or image addresses, or you accept title-only matching with lower confidence
- A named person decides each pair
- No personal data is in the export
We stop and tell you if
- The export holds customer data that cannot be removed
- You want products merged automatically
- There is no product identifier at all and titles are the only clue, so a list would be mostly guesses
What could go wrong
The job works on copies of files you send and returns files and a report. Nothing in your store or helpdesk is changed by us, so there is nothing to undo. Our copies of your files are deleted when the job ends.
Scroll the table sideways to read it all.
| Risk | How we handle it |
|---|---|
| Two different products are listed as one. | Near-miss pairs are tested, and a listed near-miss fails the whole test. |
| A real duplicate is missed. | The planted-pair test counts misses, and the list shows its confidence. |
| A merge loses stock or order history. | We never merge. The list says so and shows both stock levels where the export has them. |
A separate automated reviewer checks the work and the test results before hand-over. Your named person decides each pair and does any merge. 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 identifiers, the matching rules and the near-miss list
- Receive the product export; your sealed test pairs wait for the run
- Match on identifiers first, then titles and image addresses
- Plant our 10 pairs, 5 near-misses; add your sealed 10 pairs, 5 near-misses
- A separate automated reviewer checks the list and that the near-misses stay apart
- Hand over the lists, rules and script
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 re-running the check after each supplier import.
Ongoing work is separately scoped and quoted: no monitoring, response-time guarantee or automatic subscription is included in this job.
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
- Sort the export by SKU, barcode and title in a spreadsheet and look for repeats.
- The merge-duplicate-rows job merges clear duplicates in one list and logs them; this job only lists pairs for you to decide.
Questions
Will you merge them?
No. You decide each pair and merge in your own admin.
How is this different from the merge-duplicate-rows job?
That one merges clear duplicates in one flat list and logs each merge. This one reads a store product and variant export, merges nothing and keeps look-alike variants apart.
What about variants?
Variants that may belong under one product are listed with the fields that suggest it.
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.
Send an enquiry
Send us
- How many products and variants
- Which identifiers the store holds (SKU, barcode, image)
- How the duplicates came about, if you know
- No store login
Later, once you agree
- The product and variant export through an agreed secure route
- Your 10 duplicate pairs and 5 near-miss pairs for the test, made by you and kept from us until the test run
- The name of the person who decides each pair
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-duplicate-products-review-list” as the subject.