Turning a fragmented, call-dependent order-tracking journey into a clear, exception-first, self-serve experience — for the dealers who run their business on MyAwaaz.
Owned the order-tracking journey end-to-end — from research synthesis to developer handoff.
Research synthesis Problem framing Solution direction Wireframes UI design Validation Developer handoff
MyAwaaz is Asian Paints' dedicated dealer platform — a centralised place for dealers to manage their orders and sales. Order tracking sits at the centre: it's where a dealer answers the questions that run their day. Yet checking an order on the portal had not become the default.
This work sat within the larger MyAwaaz revamp, aimed at reducing dealer dependence on physical support channels and moving order-status checks onto the portal.
Order tracking was a critical workflow because it sat at the centre of a dealer's day — money tied up, deliveries due, and contractors waiting on answers.
However, checking an order on the portal had not become the dealer's default. The habit was support-assisted, not self-serve.
32K+ complaints and sub-3★ ratings collected over the past 5 years, through quarterly TSE-driven holistic dealer surveys. Order tracking was the single largest source of pain.
The same loop, every time a dealer needed a status — until they abandoned the portal and reached for the phone. That abandonment was the habit the revamp had to break.
1A dealer opens tracking to check a live order spread across many invoices.
2Scans and shifts through multiple rows entries for tracking updates.
3Complication leads to confusion, no meaningful insights received.
4Simpler with a phone call — so the dealer rings the TSE or the driver.
A phone call felt much simpler than tracking via the portal. Manual record-keeping became the norm to reconcile orders with financial documents.
The pull-back to the phone had two possible roots: how the portal was designed, and the complexity of the supply chain it sat on top of. The revamp had to weigh both.
How the screens themselves were built — the failures that made a status hard to read at a glance. Examined later, through the heuristic evaluation.
The shape of the underlying order-and-logistics system the screen sat on top of — a genuine many-to-many reality. Examined first, below.
An Asian Paints purchase order doesn't behave like a single line in a table. One PO fragments into many invoices, dispatched on different trucks, arriving on different days and billed on different cycles — a genuine many-to-many reality.
A single order is split across many invoices; and a single invoice can carry items belonging to more than one order. Each item sits at its own stage — ordered, not-yet-billed, invoiced, dispatched, or delivered — so an order is never simply in one state.
Quantities are invoiced in parts, cancellations issue credit notes, and short or damaged deliveries adjust the figures after the fact — so what was ordered, what was invoiced and what actually arrived each move on their own clock.
It exploded this fragmentation onto the screen: the same order appeared across multiple rows mingled with other orders, each status change spawned a new row, and billing-quantity history repeated with every update.
With the picture scattered this way, ordered, invoiced and delivered quantities rarely lined up — and the dealer could never see an order as a single, reconciled set.
A walkthrough of the legacy order-tracking screens — the heuristic failures that kept dealers off the portal and on the phone, pinned to where they actually happened.
Two pathways, deliberately different: structured field work to watch behaviour, and five years of unfiltered feedback to hear it at scale.
Mumbai · Delhi · Bangalore · Hyderabad · Kolkata. Aged 25 – 50, medium–high tech literacy — real order-tracking tasks on their own accounts.
In-portal feedback accumulated across ~5 years — star ratings out of 5★ and verbatim commentary; 32K+ complaints in the most recent year alone.
Five years of complaints, distilled to the patterns that kept repeating.
80 dealers across five cities, completing real order-tracking tasks.
Behaviour we saw on screen, beyond what dealers told us.
“It just says ‘Dispatched’. Dispatched means what — today, next week? My customer is standing in the shop and I still have to phone the ASM for a real date.”
“One order, ten bills. Half came, half didn’t — and the screen won’t tell me what is still pending.”
“What I ordered, what arrived, and what they billed are three different numbers. I keep my own diary to match them.”
“The same order is written across five rows, mixed in with other orders. I lose track of which line is mine.”
“Nobody warns me a truck is late. I find out only when the contractor calls, angry.”
“An item sat unbilled for days and blocked my credit limit — I couldn’t even place a fresh order.”
“I screenshot the status and send it on WhatsApp, but it looks unprofessional and they ring back anyway.”
“I don’t rely on the app’s time. I call the driver — his number is what actually tells me when it’ll come.”
“My big project orders I track on a paper register and WhatsApp groups, not on MyAwaaz.”
“If the driver is in a hurry, I just sign. If something’s short, I find out later and it’s my loss.”
“When a truck is delayed the contractor won’t wait — I need someone to act on it fast, or I lose the sale.”
“My real question is ‘where is my white emulsion’, not ‘where is order 3821’.”
“For a project I just want one number — how much delivered, how much still pending.”
“LR number means nothing to me. Give me the driver, the area, and the time.”
“If I knew which trucks were landing today, I’d keep the loading boys and the godown space ready.”
“A customer asks, ‘can you get me this by Friday?’ I need to see what’s already on its way to me before I promise.”
Asked to find one order’s status, dealers scrolled the long table; several opened the wrong order’s row first.
To judge a partial delivery, dealers opened several invoice rows and tried to total mentally; most phoned the ASM.
Dealers misread the repeated billing-history rows as new shipments and double-counted quantities.
Asked “is product X arriving today?”, dealers left the portal and called the driver instead.
No observed dealer noticed the delayed / exception state in the table; it carried no visual weight.
For month-end records, dealers saved invoices one by one or kept paper; none used a consolidated export.
Older owners handed the laptop to a younger staff member or son to operate the portal.
Juggles many live project orders herself; tracks deliveries on a parallel diary.
Supplies builders across many sites on tight cash flow; fast-moving power user.
Father runs on relationships and phone; son operates the counter laptop.
Capital stuck past the SLA — and the one stuck order he can’t spot in time.
My credit limit’s maxed and half this money is just sitting there — I can’t place tomorrow’s order until something frees up.
Partial fulfilment is opaque — ordered vs delivered never reconcile.
My crew’s sitting idle at Horizon Tower waiting on 300 litres of Apex Ultima — how much has shipped, and when’s the rest landing?
Incoming stock is invisible — no way to promise a date without chasing truck by truck.
I need 200 litres of Apex Ultima on the Riverside site by Thursday — can you do it?
No day-ahead read of arrivals — “Dispatched” gives no real time.
If two big loads turn up together, we’ve no space and no extra hands — the whole counter jams.
Discovers damage or missing items at delivery — no clear next step; raises a complaint separately or calls the TSE.
Sometimes we find items were damaged on arrival when we put them in the godown. It’s too late by then.
Records scattered across orders — assembling GST invoices by hand eats a day.
Pull me all the bills for what actually arrived this month.
Heuristics, feedback, interviews and observations — clustered into the challenges the new UI was built to solve.
Built — unified order detail, single Items + Invoices ledgers, side drawers, PO health bar.
One PO fragments into many invoices and rows; the whole order can’t be seen as a single set.
Partial fulfilment is opaque — no clear view of delivered vs still-pending items.
No site / PO-level “% delivered vs pending” summary for project orders.
Built — exception-first dashboard, plain-language ETA + delay stamps, arrivals readiness, Contact-TSE.
No single current status; updates are scattered across many rows.
Status is a system state, not an actionable delivery promise / ETA.
Delays, shortages, unbilled items and payment-due are never surfaced proactively.
No professional way to share an order’s status with a contractor.
Built — Sales-Assistant search (SKU / Site / Order), incoming-stock view, three-tier drill-down.
No day-ahead view of incoming arrivals to plan unloading labour and godown space.
Can’t see incoming / in-transit stock by product to commit it to a waiting customer.
Can’t quickly find a specific order, product or site in a long, dense table.
The system is order-ID-centric, while dealers search by product and by site.
Built — cumulative ledger (Ordered = Unbilled + Invoiced …), Cancelled / Credit-Note columns.
Ordered, invoiced and delivered / billed quantities don’t reconcile across views.
Repeated billing history causes double-counting and erodes trust in the numbers.
Built — stuck-order cancellation → credit notes that free credit (Credit Pulse), escalate-to-TSE.
Stuck / unbilled items lock the credit limit, with no easy way to release capital.
When a truck is delayed there is no fast way to escalate before a sale is lost.
Built — unloading verification + SKU count, ePOD ‘sign as unverified’, Past-Orders vault + bulk download.
Receiving is risky — driver pressure forces blind sign-off, with no consolidated count or grievance.
Records for GST / audit and late-found damage are hard to assemble after delivery.
Not grouped by feature clusters — but by the stages of the dealer's order-tracking journey, from the reason to check to closing the books.
Who this serves — the dashboard answers all three personas at once: Rajesh & Aarav (the Veteran Counter) scan today’s arrivals and ready the counter before a truck shows up; Vikram (the Volume Contractor) needs exceptions pushed at him, so a stuck or slipping order finds him before the sale is lost; Priya (the Project Player) keeps a pulse on her credit and how her projects are tracking.
A single row of cards, one per truck arriving today — scanned like a morning to-do list. Four versions, each watched in use before the next.
Driver and vehicle details moved a click away; categories now lead — answering what’s coming, when.
Answer five questions, in order:
And never feel like a list.
Truck number & driver matter too — but at the gate, for matching paperwork, not while planning the day. Telling gate-info from planning-info is the insight the redesign turns on.
Where each CTA leads — the card as a hub
One card, four doors: click it for the truck’s contents, Track for the live map, Help to reach the TSE, and Start Unloading once it has arrived.
The operational view of what is to arrive over the next 2 business days — to help preparedness and forecast issues.
Why the strikethrough, not one number? Only the new time hides that it slipped (erodes trust); only the delay leaves the dealer doing maths; showing both answers when, by how much, and why at a glance.
The macro pipeline — one card per active PO, answering at a glance: where is this order’s money, and do I need to act?
Scanning a dozen orders, the dealer needs each card to say do I act? — not to describe everything. The biggest pile of money isn’t always the urgent one.
Full pipeline bar — every number, a four-colour strip + legend.
Pipeline let dealer gauge an order by tracking the monetary distribution. Honest, transparent & complete.
Shows the biggest pile (descriptive) and common status
A grid of 12 pipeline cards is clutter. Showing only the biggest pile works along with a one-word status for gauging at a glance.
Shows what to act on (prescriptive), based on the status
Two un-reconciled signals was disorienting. Instead, show what needs action based on status, & not just describe the largest pile.
This section allows the dealer to access the past orders and displays urgent alerts on pending tasks, bottom-line information like Credit Limit and Site completion status, acting as a guide/reference for them to interpret the 3 main zones of the dashboard contextually.
Who this serves — Rajesh & Aarav want to find one specific order or product in seconds instead of scrolling and reaching for the phone; Priya needs to see what’s still on its way — across all her orders — to commit incoming stock to a waiting customer.
Everything you’re still waiting on, from every order, in one place — sorted by arrival, each line showing the product, how much, and where it’s reached. Click any product for its full story.
Four ways to stack the same products: by status scattered tomorrow’s run into piles you’d re-add; by window echoed the tabs; a flat list hid whose order was whose. By order won — each product under its order, clear at a glance.
Discoverability isn’t loudness. Riding the dealer’s existing scan beat a strip that shouted for space.
Always present in the side rail.
A loud band below the metrics.
“Items List ›” rides each zone heading.
Who this serves — Priya: several live project orders, each arriving as fragments, tracked in a parallel diary because ordered, delivered and billed never matched — the reconciled funnel is what retires the diary. Rajesh: opens it to read a single plain status and a real ETA instead of scattered rows. Vikram: reads the same page money-first — unbilled value and credit notes, stage by stage.
An order is a list of products, so the instinct is one view, by product. But one order is split across many invoices and trucks — and an invoice can carry more than one order’s items (the order–invoice tangle is laid out earlier). Three calls came before any layout could be drawn.
With the structure set, one question was left — placement. Where does the order summary sit, where do the two views go, and where do the urgent cases show up? Three full, clickable layouts were built and judged by how a dealer would actually read the page. Each sketch below shows block placement only, not the content inside.
Summary and table stay in view together; the dealer can filter without losing sight of the table.
Whatever needs action is the first thing the eye lands on.
The whole order, read top to bottom, in one honest frame.
C shipped. The order-detail page is a screen the dealer opens many times a day to scan, not a dashboard they sit and study — so the everyday read was optimised over the exception.
C’s cost is real: the attention-worthy cases are one click away, not zero — the dealer has to pull a filter to isolate them. We took that deal for three reasons.
Net — C trades one filter click (a small, recoverable cost) for a full-width, uncluttered table every single time (a large, everyday gain). On a screen built to scan, the everyday state wins — and the ribbon’s per-stage flags make sure the exception is never a surprise, only un-isolated.
One toggle, no disorientation — both ribbons read money by stage, left to right, so the second view is familiar the first time it’s seen.
“Where’s this invoice or credit note?” — jump straight to it instead of scrolling every invoice row.
“Which invoices carry the most value — or the most issues?” — rank any stage to bring the biggest exposure to the top.
“what needs me, and when?” Time filters (Today / Tomorrow / This Week) order the trucks by arrival; Delayed and Issues pull trouble to the top.
“Where is this one SKU?” — jump straight to a line instead of scrolling 119 item rows.
“Which products carry the most value — or the most damage?” — rank any stage to bring the biggest exposure to the top.
“which products need me?” Delayed, Damaged and Short-Shipped isolate the broken ones; Category narrows by paint line.
“I want to act on these specific lines together, not one by one.” — row checkboxes with a running Selected count.
“I’ve picked some lines — what do they add up to, and what can I do with them?” — a floating bar rises with a running tally (count · total items · total value) and the actions that fit the picked lines.
From the Invoices view — pull or share any invoice’s PDF, its ePOD or its credit note from the row, or bundle a selection into one download.
From the Invoices view — pull or share any invoice’s PDF, its ePOD or its credit note from the row, or bundle a selection into one download.
Pull or share one product’s paperwork — the invoices and credit notes tied to it — without leaving its row.
Pull or share one product’s paperwork — the invoices and credit notes tied to it — without leaving its row.
Confirm receipt, log shortages, raise a problem on arrival.
“What actually came off the truck?” — tick each SKU received, short or damaged as it’s unloaded, before signing.
Follow a truck live — driver, position, revised ETA.
“Something’s wrong the screen can’t fix — who do I call?” — one click to the dealer’s TSE, in context.
“Cancel just these items, by quantity — not the whole order.” — a per-SKU dial-in modal scoped to the ticked lines, confirmed with an OTP.
“Call the whole thing off.” — the heaviest action, kept in the kebab and nudged toward partial-cancel first.
Who this serves — Vikram, the Volume Contractor on tight cash flow: capital stuck in unbilled items freezes his credit limit, and a delay he hears about too late costs him the sale. Cancelling to free credit, and escalating fast, are his two levers.
You can’t cancel what’s already moving. Once units are on the road or delivered they’re out of reach — they go through Returns.
A GST tax invoice can’t be deleted. You can only reverse the items on it with a credit note, never erase the invoice itself.
One System. Two paths. Cancel the entire order or a selected few with the same experience. Scroll the map sideways to travel the full flow.
An invoice can’t legally be edited or cancelled — cancelled items are reversed with a credit note instead.
Fully-reversed wears Settled when all items in that invoice are cancelled; partial cancellation wears Adjusted.
Nets the row to zero, with the CN Issued flag and the count of items cancelled.
Invoices already on the way or delivered can’t be cancelled at all.
Next to the SKU name for easy identification and feedback.
Next to the updated numbers for zero confusion, so that the numbers always add up.
Every cancellation that touches billed stock issues a credit note. Those notes aren’t tucked away in one place — apart from the Statement of Accounts section, inside order tracking they surface wherever the dealer is already looking: the Credit Notes column on each invoice row, the CN Issued tab in the Invoices Drawer, the Issues & CNs tab inside a product’s drawer, the Download menu (credit notes as PDF), and the Activity Log for bulk reconciliation. This is done for contextual information and easy reconciliation — and the cancellation confirmation shows the count straight away too.
On each invoice row.
In the Invoices Drawer.
Inside a product’s drawer.
Credit notes as PDF.
In the activity log for bulk reconciliation.
Units past the Transit Wall could be returned later, but a return almost never happens; when something is wrong it’s caught at the door (damage / shortage → credit note). Returns stay routed through the TSE, off-platform.
Who this serves — the Veteran Counter at the dock. When the screen is unreadable, this pair defaults to a call — the very behaviour the revamp must replace — so the counting flow is built for the proxy reality, where whoever is free has to work it amid the noise.
The dealer opens its card to start receiving.
Confirm, flag damaged, or flag short — any mix, mid-count.
Sign & Verify, or Sign as Unverified to clear the truck and keep a window.
The receipt is created and saved with the order.
All of it while the crew unloads, the driver hovers to leave, and a contractor may be on the phone.
Each row is one product. The dealer confirms it all arrived, flags some short, or flags some damaged — short & damage can land on the same row, any order. Fit all four into one row without it becoming a form.
Every row starts with a filled counter — reduce it by the count of short / damage, so counting stays quick.
No discrepancies from human error between the number of short shippings and damages.
When Aarav is busy, 55+ owner Rajesh works it — phone in hand, noisy floor, interrupted. So the row must resist mis-clicks and read its state at a glance, ending the costly blind sign-off.
In practice, it gets out of the dealer’s way. Nine times in ten nothing is wrong — one click to Verify and the row is done. On a short count, lowering the number flips Verify to Flag Short; a quiet icon-only button opens the damage pane only when it’s needed.
The quiet brilliance: damage or shortage can come first, in any order. The system choreographs the + / − counters behind the scenes — disabling whatever would break the arithmetic — so the totals never drift into disarray, without ever demanding the dealer’s full attention.
Activates only once every item is verified. Nothing is filed until the dealer signs; one ePOD per invoice, credit notes queue automatically.
The driver leaves now, and a 4-hour window lets the dealer raise damage / shortage later from Past Orders — the credit note auto-applies.
In short — receiving used to be a signature under pressure, where any shortage quietly became the dealer’s loss. This makes it a count the dealer controls: honest by default, short & damage in one gesture, consistent by construction, protected either way.
Who this serves — Vikram, who needs clean records he can export for audit, and Priya, who kept a manual diary because ordered, delivered and billed never reconciled — both now pull the final math and the paperwork from one read-only page.
Settled orders and their documents live with the order itself, inside Past Orders. There's no separate place to file or hunt for paperwork — every invoice, ePOD and credit note sits with the order it belongs to.
Everything lives in a dedicated Past Orders archive — no second place to learn or keep in sync.
Once an order settles, its detail page becomes a read-only record — the same Invoices and Items tabs the dealer already knows, but frozen, with a settlement summary on top and the order’s full history a click away.
Totals the order’s exceptions by type, legible before a single row is read.
Jump straight to claims that still need chasing.
A Verified / Unverified pill and the credit-note amount on the line.
“Download All CNs” turns the log into the order’s credit-note bundle.
In short — the settled order stops being a scavenger hunt. Records, credit notes and the final math now sit with the order, so the dealer can pull any document, trust the numbers, and settle a dispute from one read-only page.
Dealers were asked to complete a few key tasks they normally did off MyAwaaz — one of them: tracking an order and reconciling it. Scores out of 5.
Turning a fragmented, phone-first tracking habit into something dealers would trust was never tidy — scrapped layouts, dead-ends, and plenty of “back to the drawing board” days. It was a classical design process — research, synthesis, iteration, judgment — run with AI as an accelerant; the lessons that stuck live where the two meet.
A return is still routed entirely through the TSE today; a real in-product returns flow would close the last off-platform loop.
The post-delivery damage report exists only as a stub; building it out completes the verification-to-credit chain.
Surface a slip before it’s reported, not after, and draft the escalation — extending AI from the design process into the product itself.
