Curated mall app for a health-retail brand
A shopping app for a health-retail brand: self-operated quality assurance, curated and daily-plus zo...
View Details →The category is not short of goods; it is short of "dare I buy this" and "will I remember to come back". Once the app existed we could see that these are three different system capabilities, not three screens.
The operating manual lists six features. Architecturally they are three groups: trust (self-operated and assured), repeat (points, notifications, content) and referral (share and commission). Each needs its own data structure; none can be faked with a few extra pages.
| Zone / capability | What it solves | Key rules |
|---|---|---|
| Curated categories | Trust: factory-direct, fair price | Self-operated items are flagged and carry assurance copy on both list and detail; the tree covers daily goods, health food, appliances and books |
| Recommended zone | Decision: decide for the user | Editorial selection plus ranking weight; the home feed and the zone list read the same product data |
| PLUS daily picks | Conversion: a reason to buy today | A daily refreshed discount pool, price next to strike-through price, discount expiring back to normal automatically |
| Points redemption | Repeat: turn a purchase into a balance | Points accrue on payment and buy goods; redemption runs as a points order, accounted separately from cash orders |
| Content channel | Retention: a non-shopping reason to open the app | Wellness and product explainers on a schedule, with product cards inside articles |
| One-tap sharing | Referral: customers bring customers | Share links and QR codes carry a parameter; registration binds the relationship; commission is calculated on completed orders |
| Order notifications | Reassurance: fewer "where is my parcel" tickets | Push on paid, shipped, delivered and reviewable; tracking follows order state |
An easy judgement to get wrong: six features are not six pages, they are six views of the same product. So the model holds one catalogue and one SKU set, distributed by five flags — self-operated, recommended, discounted, points and content. Otherwise every new zone means another copy of the catalogue to maintain.
Most mobile drop-off happens where the user cannot understand something or taps the wrong thing. The detail page therefore answers only four questions: how much, which spec, how many, and what the total is.
The cart does one job: final confirmation before checkout. No recommendations, no promotion stacking — just the item list with spec in small print, quantity, select-all and total, and a checkout button carrying the item count. Edit and delete live behind a corner button, because rare and destructive actions should not sit under the thumb.
The order list answers one question: where is my parcel. For the platform, those same four states are four unobtrusive opportunities to talk — and every state change is a legitimate reason to push.
In the manual, notifications are one line. In reality they are a subscription: placing an order subscribes that order number, and the server pushes a templated message on each state change. Marketing notifications can be switched off in settings, but transaction notifications (paid, shipped, delivered) cannot — they are not interruptions, they are part of fulfilment.
Points and commission both look like "just add a number", but they are two different ledgers: one is what the platform owes the buyer, the other what it owes the sharer. They must be separated in the data model.
Most commerce support cost comes from people not finding an entry. The account tab has one plain goal: anything the user can look up themselves should never become a ticket.
| Area | Contains | Why |
|---|---|---|
| Identity | Avatar, nickname, masked mobile number | The mobile number is the account; shown masked to balance recognition and privacy |
| Assets | Favourites, points balance | Two numbers on the surface, detail one tap deeper: result first, process second |
| Orders | Pending payment / shipment / receipt / review | Each state carries a count badge so nobody has to enter a list and filter |
| Redemption | My points, points mall, my redemptions | Points go from visible to spent in three steps |
| Service | Support, complaints, invite friends | Complaint and invite side by side: one absorbs dissatisfaction, one amplifies satisfaction |
| Settings | Profile, payment password, notification switches | The payment password has its own screen, set separately from sign-in |
Two passwords, two doors. Sign-in supports account password or an SMS code, with recovery through the mobile number; the payment password is a separate secondary password, checked only at payment. The reason is practical: shopping apps keep users signed in for months, and phones get lent, lost or handed to family. The sign-in password protects the account from being opened; the payment password protects the money from being spent. Different risks, different locks.
Favourites are an underrated retention surface: a list the user builds themselves, and the most honest interest signal operations will ever get. We only do two things — one-tap save from the detail page, and a reverse-chronological list showing current price with a drop hint. No "you may also like" noise, because saving something already states the intent.
The manual was written screen by screen. We inverted it into one table — flow, path, responsibility — and handed it to both the client and the support team, who now use it to locate exactly where a user is stuck.
| # | Flow | Path | What it must accomplish |
|---|---|---|---|
| 01 | Sign in | Launch → sign in | Account and password or SMS code; signup and recovery through the mobile number |
| 02 | Curated categories | Home entry / categories tab | Enter a list from a level-1 category: daily goods, health food, appliances, books |
| 03 | Recommended zone | Home entry / category page | Browse editorially selected products |
| 04 | PLUS daily picks | Home entry / category page | See the day's discounts with price and validity on one screen |
| 05 | Points redemption | Home entry / category page | Enter the points-only catalogue |
| 06 | Content channel | Home entry / article strip | Read wellness and product explainers that link straight to products |
| 07 | Product detail | Tap an item in any list | Gallery and price, choose spec and quantity, see the total, add to cart or buy now |
| 08 | Cart | Bottom tab / after add-to-cart | Check items and specs, adjust quantity, select, check out together |
| 09 | Orders | Account → orders | Filter by four states, track logistics, start after-sales, review |
| 10 | Favourites | Account → favourites | See saved items with current price and return to the detail page |
| 11 | Points mall | Account → redemption → points mall | Redeem with points, produce a points order, review it under my redemptions |
| 12 | Payment password | Account → avatar / edit → settings | Set the secondary password that guards the payment step |
The table doubles as acceptance: every flow maps to a reproducible sequence, so the client and we walk it once and the system is delivered. It is also the onboarding script whenever support or operations changes hands.
As agreed with the client, this page shows no client or brand names, no product photography and no screenshots of production screens. De-identified performance figures are available during a scoping call, by comparable scenario.
Bring three things to the first call: how products are sourced (self-operated, distributed or both), how you reward repurchase today (points, coupons, nothing), and whether referral commission is part of the model. We will say first whether the bottleneck is catalogue, incentive or fulfilment, then propose scope and a phased quote. Related: how we build and the cloud-factory app case.
MORE CASES
A shopping app for a health-retail brand: self-operated quality assurance, curated and daily-plus zo...
View Details →Start with our pricing logic, then let's talk requirements — cost structure, technology comparison and acceptance criteria, all laid out in the open.