Case study A fragrance retailer Store operations
A retailer's whole catalog, run from one email
The client sells a few hundred fragrance and beauty products in a marketplace store and their own online store. Keeping the online store current took the team more than ten hours a week, and it still fell behind. Now the team sends one email, and a workflow we built and run does the rest.
Before: in a typical week the team spent more than ten hours adding new arrivals, pulling sold-out items, and fixing prices and stock counts by hand. After: the same week holds one task, sending the inventory email.
Illustrative: the kinds of tasks the team did by hand in a typical week, adding up to more than ten hours.
More than ten hours a week, back with the team.
Below is one inventory email, followed from the moment the team sends it to the reply that comes back. The screens show how the real workflow handles it. Product names and prices are examples. The numbers in the margin are from the client's real runs.
Follow the email
-
01 · Received
The team sends one email
The team emails their current inventory sheet to one address. The workflow accepts one sender and one subject line. Every other email is ignored, so nothing else in that inbox can start a run.
- From
- team@client
- sender ok
- To
- inventory@ops
- Subject
- Inventory update
- subject ok
Current inventory attached.
inventory-sheet.xlsx- unknown sender Inventory update ignored
- team@client Quick question ignored
278
items received on the sheet
1 + 1
one sender, one subject line. Everything else is ignored
-
02 · Matched
Every line paired with the store
Every row on the sheet is matched to its product in the online store by barcode, size, and packaging, or marked new when the store has never listed it. The same name in different packaging is a different product. Close calls, such as two listings that share a barcode or a boxed and an unboxed bottle, are settled by the client's matching rules. Nothing is guessed, and nothing is handed off.
Row Barcode Product Size Pack Qty Price Match 2 …22871 Cedar & Salt Eau de Toilette 1.7 oz Boxed 5 $58.00 price, stock 3 …40193 Fig Leaf Eau de Parfum 3.4 oz Boxed 4 $92.00 new 4 …17725 White Tea Body Mist 8 oz Boxed 0 $24.00 sold out 5 …66308 Vetiver Night Eau de Parfum 3.4 oz Tester 2 $70.00 by rule 6 …05514 Amber Dusk Parfum 1 oz Boxed 6 $110.00 no change 273 more rows
Row 5 is a tester that shares its barcode with the boxed bottle. The client's matching rules treat a tester as its own product, so the row is matched to the tester and the boxed bottle is left alone.
278
rows matched by barcode, size, and packaging
-
03 · Written
New arrivals drafted in the store's own format
For stock the store has never listed, the model writes the product page in the format the store already uses. Every page has the same parts, and every page starts as a draft.
- Title
- Fig Leaf Eau de Parfum, 3.4 oz
- Type
- Eau de Parfum
- Notes
- Fig leaf, green cedar, soft musk
- How it wears
- Close to the skin, quiet by the afternoon
- Sizes
- 3.4 oz
- Search title
- Fig Leaf Eau de Parfum 3.4 oz Spray
- Photo alt text
- Fig Leaf Eau de Parfum, 3.4 oz bottle, front view
130
product pages drafted in one run
-
04 · Checked
Every new listing checked
Every new listing has to pass the same checks: the right product type, notes from a reliable source, a brand that matches the title, alt text on the photo, and no competitor or source names. A listing that fails any check is flagged and cannot go live as it is.
- Product type set correctly
- Notes from a reliable source
- Brand matches the title
- Photo alt text present
- No competitor or source names
passed ready to publish
- 130 passed every check
- 47 flagged
flagged sent to the runbook
177
new listings checked on the first run, 130 passed
47
flagged by the checks
-
05 · Runbook
The runbook fixes what was flagged
The 47 flagged listings are worked through the client's runbook, the company's written standards, automatically. Each kind of flag has a written rule. The workflow applies the rule, runs every check again, and only a listing that passes can go live.
Runbook
Written standards for flagged listings
-
Beauty and accessory items 26 flagged
Set the right product type, then write the page in that type's format.
-
Notes missing 16 flagged
Look them up from the brand's own pages, never a retailer.
-
Brand does not match title 5 flagged
Rewrite the title to <Brand> <Name> <Concentration>, the store's naming standard.
-
After any fix
Run every check again.
Fix log
-
Beauty item
White Tea Lip Color
- Type
-
was
Women's Fragrance, now Lip Color
-
Notes missing
Amber Dusk Parfum, 1 oz
- Notes
-
was
None found, now Amber, labdanum, dry woods
-
Brand does not match title
Vetiver Night, 3.4 oz
- Title
-
was
VETIVER NIGHT edp 3.4oz, now Vetiver Night Eau de Parfum
47
flagged out of 177
Fixed
by the runbook, re-checked before going live
-
-
06 · Published
The store updated, then read back
Before every write, the workflow backs up what the store shows. It sets prices and stock counts from the sheet, publishes the drafts that passed, and reads every change back to confirm it landed. The store never sells past zero.
-
read back
Cedar & Salt Eau de Toilette, 1.7 oz
- Price
- was
$64.00, now $58.00 - Stock
- was
2, now 5
-
read back
White Tea Body Mist, 8 oz
- Stock
- was
1, now 0 - Status
- was
In stock, now Sold out
-
read back
Fig Leaf Eau de Parfum, 3.4 oz
- Status
- was
Draft, now Live
132
price and stock changes on the first sync, 0 errors, every one read back
49 → 179
live products in a single day, 130 published on day one
-
-
07 · Reported
A reply with what changed, and how each flag was fixed
The run ends with a reply to the team: a summary of what changed in the store, and the flagged items grouped by category, with what the runbook did for each one.
- From
- inventory@ops
- To
- team@client
- Subject
- Re: Inventory update
47
flagged items, grouped in the reply by category
10+ hrs
of weekly store work, now handled by the workflow
The team still decides what the store sells. The ten hours a week of keeping it current now start with one email.
Have a job your team repeats every week? We will tell you whether it can run on its own, and what it would take.
Get the free assessment Back to AI & automation