They open your store. And somehow, it already knows.
A For-You feed for your storefront. Every session learns what each shopper actually wants and re-ranks the UGC + products in real time. No quizzes. No “rate this item”. No login wall. Just your store — rearranged for every visitor, automatically.
- No preference quizzes
- Anonymous personalisation
- Same model, every channel
Personalised feed for
Robin
Less friction. More discovery.
Already knows
Opens the storefront → relevant UGC + products in the first scroll. No quizzes, no rate-this-item prompts.
Gets better each session
The feed re-ranks on every visit. Customers feel the storefront grow with them, not against them.
No preference forms
Zero friction up-front. Anonymous visitors get a great feed too — no login required for personalization to kick in.
Every customer gets a storefront made for them.
+22% CR median
Across our beta cohort, personalized feeds lift PDP CR by 18-28% in the first 90 days.
Same UGC. More work.
The library you already aggregated re-arranges itself per session — you don’t need to ship more content to see more revenue.
Mobile-first
Optimised for phones where 78% of shopping happens. Story-grade swipe + tap mechanics built in.
What the ranker actually watches.
All anonymous, all session-level. No fingerprinting, no IP-based tracking, no precise geolocation.
View time
Seconds spent on each post + product. The strongest first-session signal.
Tap behaviour
Hotspot clicks, swipe direction, scroll velocity, pinch zooms.
Cart + purchase
Add-to-bag, checkout starts, completed orders feed back to the ranker on the next visit.
Implicit preferences
Inferred from creator follow patterns, hashtag affinity, image embedding similarity.
Filters applied
Size, colour, price — every time a shopper narrows the grid, the ranker learns harder.
Return visits
Cross-session memory: the feed remembers what worked last time without nagging logins.
Four steps. Personalised by second six.
- Step 1
Capture signals
View time, taps, swipes, cart events, filter use — collected anonymously per session. No PII required.
- Step 2
Rank in real time
A session-level embedding model scores every UGC post + tagged product against the shopper’s implicit preferences. Recomputes on every view.
- Step 3
Surface what fits
The For-You feed re-flows the storefront, PDP UGC rail and email blocks. Klaviyo + Meta CAPI get the same ranking signal.
- Step 4
Learn forever
Cross-session memory keyed off device + checkout email lets the next visit start where the last one left off.
Personalisation without surveillance.
Every signal is session-level and anonymous. No IP storage, no device fingerprint, no third-party cookie. The preference vector lives in your Idukki workspace — never shared, never sold.
- GDPR + CCPA + DSA compliant out of the box
- No personally-identifiable data required for personalisation
- Per-shopper preference vector exportable + deletable on request
- Shopify customer-data API compliant
What we store per session
- session_idopaque · rotates every 24hanonymous
- preference_embedding128-dim float vectorderived
- view_seconds[post]aggregated per post + slotevent
- cart_eventscounts only — no SKUsevent
- consent_flagtrue / false / unknowncompliance
- ip / fingerprint✗ never storedblocked
- precise_geo✗ never storedblocked
Alyssa Park
Head of Ecommerce, Northwave Apparel
“Our PDP CR jumped 24% in the first 6 weeks. Every shopper sees a different mix of UGC now — and the AOV moved with it.”
FAQ
How the For-You feed actually works.
Do shoppers need to log in or fill out a preference quiz?+
No. Personalization kicks in within the first 6-10 seconds of the first session, anonymously. If the shopper logs in or hits checkout, the feed carries the preference vector across sessions and devices.
What data do you store on each shopper?+
A session-level embedding vector (no identifiable data) plus aggregated event counts. We don’t store IP, fingerprint or precise location for personalization. Idukki is GDPR + CCPA + DSA compliant out of the box.
How is this different from Shopify product recommendations?+
Shopify ranks products. We rank UGC + products together. Every post in your library competes for the shopper’s attention against every other post, with the matching product tagged on it — so personalization works on creator content, not just catalog SKUs.
Will the feed always show fresh content?+
Yes. Recency, novelty and diversity penalties are baked into the ranker so a shopper never gets the same five videos every session. Diversity bonus also avoids creator over-rotation.
Can I override the ranker for specific campaigns?+
Pin a hashtag campaign, a creator, or a product to the top of the feed for any segment. Pins re-rank everything else underneath while keeping personalization intact for the un-pinned slots.
What happens during cold start (no signals yet)?+
First few seconds use a popularity-prior weighted by your top-performing UGC over the last 7 days. Once the shopper makes 2-3 interactions the personalized ranker takes over — typically within the first 10 seconds of session.
Does it work on the Shop app + storefront + PDP + email all at once?+
Yes. The same ranking signal feeds (a) your storefront feed, (b) PDP UGC carousels, (c) the Shop app surface, (d) Klaviyo dynamic-content blocks, and (e) Meta CAPI personalisation events. One model, one preference vector, every channel.
How is the model trained?+
Per-brand transformer fine-tune on top of a shared base. We start with industry-tagged priors (fashion vs beauty vs home) and adapt to your library + traffic within 24-72 hours of going live. No data leaves your workspace.
The one-size-fits-all era is over.
Your feed should fit them. Tell us your platform + your traffic and we’ll have a personalized feed running on a sandbox of your store within one business day.
Stop renting six tools. Ship one.
Spin up your first widget in 4 minutes. Migrate from Bazaarvoice, EmbedSocial, Tolstoy or Videowise in a day. Your CFO will love us. Your CRO already does.
- No credit card
- Cancel anytime
- SOC 2 + GDPR