Review moderation SLA for fashion and beauty brands
Approve, hide, reply and AI-draft your way through a high-volume review queue. Working SLAs for fashion and beauty SKUs.
Monday opens with 140 new reviews across foundation shades and dress sizes. Half need a photo check. A dozen name a broken pump or a seam. Two ask for a refund in the review text. Without a service level, the queue sits until Friday. The product page then shows last week's proof.
In this article
We make Idukki. Reviews land in the same moderation library as your UGC. Approve, hide or remove them there. Draft replies with AI. Switch on labelled AI summaries. This piece is the ops layer on top.
The queue under the stars
Review moderation as an iceberg
Live stars, photos and replies
What shoppers see on the product page
What the moderation queue holds
What you actually build and maintain
- Same-day bestsellers
Top SKUs: approve media, publish text, draft replies.
- Safety and legal jumps
Allergy, injury, counterfeit, regulatory wording.
- Fit and shade checks
Wrong colour photo, old packaging, size mismatch.
- AI draft replies
Draft, edit, publish. Never auto-send.
- 48-hour long tail
Slow SKUs, imports, thank-you photo asks.
The four actions
| Action | Use when | Owner | SLA default |
|---|---|---|---|
| Approve | The review is on-topic, safe and about the right product | Reviews owner | Same day on bestsellers |
| Hide | Spam, abuse, off-topic, or clear policy breach | Reviews owner | Same day |
| Reply | A question, a complaint, or a useful tip to amplify | CX | 24 hours on bestsellers |
| AI draft → edit → publish | Volume is high and the reply is routine | CX | Draft in minutes, publish after edit |
Hide is for spam and abuse. It is not for three-star fit notes. Keeping genuine low ratings is part of staying inside the fake reviews rules.
Working SLAs by SKU band
These are defaults for a fashion or beauty team of one to three people. Change the hours to match your staffing. Publish the table in your ops doc so marketing cannot surprise CX with a launch.
| SKU band | Approve / hide | First reply | Notes |
|---|---|---|---|
| Top 20 bestsellers | Same business day | 24 hours | Photo check on every media review |
| New launch, first 30 days | Same business day | 12 hours | Shade and size mistakes jump the line |
| Mid catalogue | 48 hours | 48 hours | Batch twice a day |
| Long-tail / archive | 72 hours | As needed | Reply only when tagged to CX |
| Safety or legal flag | Immediate triage | Immediate | Escalation path to legal or QA |
What fashion and beauty add to the queue
- Shade and colour drift. A photo under yellow indoor light can mislead. Ask for a daylight retake, or add a note in the reply.
- Size and fit language. Keep the review. Reply with the size chart link and how other buyers described fit.
- Ingredient and allergy mentions. Escalate. Do not let an AI draft invent a medical claim.
- Before-and-after beauty photos. Check for heavy filters and for claims your brand will not stand behind.
- Packaging changes. Hide or label photos of retired packaging when they confuse the current SKU.
Where AI drafts help, and where they stop
In Idukki, open a review, click Draft with AI, edit the reply and publish it. Drafts are told not to promise refunds or invent facts. You still own every published line.
- Good fit for AI draft: thank-yous, size-chart pointers, restocks, confirmed shipping timing.
- Human only: refunds, allergies, injury, counterfeit, chargebacks, influencer disputes.
- Imported platform reviews: Google, Trustpilot and TripAdvisor get a copyable reply. You post it on that platform yourself.
AI review summaries are a separate switch. They need five or more approved reviews. They stay off until you turn them on. They are labelled as AI on the page. Clear the backlog before you trust a summary.
A daily pass that fits a busy week
One morning pass, one afternoon pass
- 01
Safety sweep
Filter for allergy, injury, counterfeit and legal keywords.
- 02
Bestseller media
Approve or hide photo and video reviews on the top SKUs.
- 03
Text approve
Publish on-topic text reviews. Tag products that missed a match.
- 04
Reply batch
Draft with AI where routine, edit, publish. Escalate the rest.
- 05
SLA check
Count anything older than the band's limit and clear it before close.
Questions store owners ask
Should we auto-publish five-star reviews?
Only if spam filters are strong and media still gets a human look. Fashion and beauty photos go wrong often enough that a short media check pays off.
Can AI post replies without us?
Not in Idukki. You edit and publish every reply. That is the safe default for refund and claim language.
What about reviews that name a competitor?
Keep them if they are genuine. Reply with care. Hide only if they break your written policy for abuse or off-topic spam.
How do we staff a launch week?
Move the launch SKUs into the bestseller band for 30 days. Borrow one CX person for the morning safety sweep.
Where do AI summaries fit the SLA?
After approvals. A summary of unmoderated text will repeat problems you meant to hide.
Sources and related reading
- 1Idukki help: turn on AI review summaries
- 2Idukki: Reviews · Moderation, AI reply drafts, imports and summaries.
- 3Idukki: fake reviews law in 2026
- 4Idukki: product page hub help · Where approved reviews and summaries appear on the PDP.
More from Rohin Aggarwal
- Strategy
Trust badges vs real customer content
A trust badge is a generic claim every store carries; real customer content is specific evidence. Use badges for hygiene, content to win the decision.
- Strategy
A/B testing UGC placement: a framework
Where UGC sits on a page changes the conversion lift. Test one variable at a time, on enough traffic, for a full weekly cycle, and measure conversion, not clicks.
- Strategy
Review velocity: why fresh reviews outperform old ones
A review from last week reads as a current product; a wall of stale stars reads as a stalled brand. Velocity, not volume, keeps social proof working.