BWS · Omniverse Fulfilment · Research & Design · Feb 2025

When the alarm doesn't alarm, orders get missed.

BWS store teams were missing online orders — delivery drivers arriving before the notification. I went into stores, interviewed seven team members, and found a system of three disconnected tools that no one fully trusted. The fix wasn't just about notifications. It was about getting the right tool in the right hands.

Client
BWS · Endeavour Group
My Role
Product Designer
Engagement
Research & Design
Opportunity
Research → design → increase RF adoption → fewer errors → better NPS
The problem
The opportunity
Staff missing orders
6/7
Of interviewees had missed at least one order due to notification failure.
Want sound alerts
7/7
Every team member wanted audible alerts. None of the three tools reliably provided them.
Notification cap
47
RF NotifyMe silences all alerts after 47 — often hit in a single busy shift.
RF usage pre-JBOMS
73%
Of orders picked via RF before JBOMS. Errors were lower. That's the target to get back to.
RF usage post-JBOMS
29%
JBOMS decommissioned — chance to rebuild RF as the default picking tool.
Context

Three tools. None of them talking to each other.

BWS store teams fulfil online orders from three platforms — the POS system at the counter, an RF handheld device running a NotifyMe app, and a desktop-based Electronic Confirmation of Fulfilment (ECF) tool. Each sends notifications independently. None synchronise with the others. The result: a driver arrives, no one knew the order was waiting.

The problem was identified in IPD research as far back as early 2024. Workshops were held. A May 2024 ideation session flagged the need for a single audible notification through the store speaker. But the in-store reality hadn't been validated by anyone who'd actually stood in a BWS and watched a team member fulfil an order during a Friday evening rush. That's what this research was for.

Feb – Mar 2024
IPD Research
Order notifications identified as a high-priority problem. No in-store validation yet.
May 2024
Ideation Workshop
Stakeholder session including in-store team members. Key theme: a single noise through the store speaker.
June 2024
Order Notifications Workshop
Stakeholder kick-off. Problem scoped. In-store research commissioned.
Feb 2025
In-Store Research — where I came in
7 semi-structured interviews across high and low volume BWS stores. Mix of team members and multi-site managers. 30–60 minutes each, in-store.
In progress
Design & Recommendations
Designs to increase RF adoption and improve notification reliability across all three platforms.
Findings

What seven conversations in seven stores taught us.

The interviews confirmed the known issues — and revealed something more important. The reason staff weren't using the most accurate tool (the RF device) wasn't laziness or training failure. It was rational avoidance of a tool they'd learned not to trust. Fix the trust problem, and RF adoption follows. RF adoption follows, and picking errors drop. Picking errors drop, and NPS improves.

High Priority High Impact
The RF device — most accurate, least trusted.
The RF is the most reliable tool for picking accuracy. But 3 of 7 staff didn't know it had sound at all. Of those who did, only 2 knew that sound cuts out after 47 notifications. Staff had experienced missed orders after trusting a device they didn't understand — and stopped using it.
F.01
NotifyMe app stops audible notifications after 47+ items — builds up quickly with system alerts mixed in.
F.02
NotifyMe notifications don't link to the PickUp app, so are never actioned — further inflating the 47+ count.
F.03
Volume settings changed by one staff member affect the whole device — next shift inherits silenced alerts.
F.04
Training happens shift-by-shift, person-to-person. No baseline awareness of device capabilities.
High Priority High Impact
The POS — most used, most flawed.
6 of 7 staff rely on POS notifications. But POS only shows notifications when a staff member logs in to serve a customer. Between customers, new orders go unseen. When they do appear, they stack — 30 to 40 pop-ups that must be tapped through one by one, mid-transaction.
F.05
Notifications only visible once logged in — invisible between transactions.
F.06
No bulk clearance. Each notification must be manually dismissed. Slows down serving a customer.
F.07
No visual distinction between a new order, an SLA reminder, and a system alert.
F.08
No notification when an order is cancelled — SOH must be manually adjusted, often forgotten.
High Priority Medium Impact
ECF — fastest for management, no notification at all.
71% of staff prefer ECF for order management. It's faster and handles multiple channels. But it has zero notification capability — no sound, no visual alert, no auto-refresh. BWS online orders appear below third-party orders, making them easy to miss. And cancelled orders require manual SOH adjustment.
F.09
No notifications — relies entirely on manual checking and keeping the browser open.
F.10
BWS online orders deprioritised in the UI — appear below third-party orders with no filter.
F.11
No customer names on UCON orders without opening the full order detail.
F.12
No product images — picking errors more likely for similar-looking bottles on large orders.
NPS Impact Strategic
The real cost: errors reach customers.
Staff using ECF acknowledged that picking errors happen — particularly on large orders with similar products. The RF device's scanning step prevents this, but its trust deficit means staff avoid it. Every picking error that reaches a customer is a potential NPS drop, a refund, and a reason not to order again. This is the business case for improving RF adoption, not just fixing notifications.
F.13
ECF picking without scanning relies on staff knowledge — fails on large orders or unfamiliar products.
F.14
RF adoption dropped from 73% → 29% after JBOMS introduction. Decommissioning JBOMS creates an opportunity to rebuild RF as the default.
F.15
Picking errors, refund rates, and NPS impact of increased ECF usage have not been formally measured since JBOMS removal.
"

The alarm 'notify me' doesn't alarm.

BWS Team Member · Store visit
"

The notifications only come up when the POS is being activated — so if you haven't served anyone, you won't see them.

BWS Team Member · Store visit
"

Sometimes when it's a larger order, I see myself coming back and forth a bit. If there was a picture of the item, it would be a lot quicker.

BWS Team Member · Store visit
System Strengths Limitations
POS
  • Most-used notification method
  • Always-on at counter
  • Only visible when logged in
  • Manual clearance, one at a time
  • No differentiation between alert types
  • No cancellation notification
RF Device
  • Most accurate for picking
  • Barcode scan prevents errors
  • Good for new staff learning products
  • Sound silences at 47+ notifications
  • Crashes frequently
  • Settings change per-shift — no defaults
  • Often only 1 device per store
ECF
  • Fastest order management
  • Handles multiple channels
  • No notifications of any kind
  • BWS orders deprioritised in UI
  • No product images or customer names on UCON
  • Cancelled order SOH is manual
Designs

Making the right tool the easy choice.

The design response has two goals: fix the immediate notification failures, and rebuild staff confidence in the RF device. If the RF becomes reliable and easy to use, adoption increases. If adoption increases, scanning replaces manual picking. If scanning replaces manual picking, errors drop — and NPS follows.

Design 01

RF device — notification redesign

Auto-clearing notifications to prevent the 47+ silence cap. NotifyMe linked directly to the PickUp app. Default volume settings that restore on restart — no more silent devices from the previous shift.

📱
RF Device · NotifyMe Redesign
Notification flow · Auto-clear · Deep-link
Design in progress
🔔
RF Device · Default Settings
Volume default · Session reset · PickUp link
Design in progress
Design 02

POS — notification overhaul

Bulk clearance of notifications. Visual distinction between new orders, SLA reminders, and system alerts. Always-visible order count — not hidden behind login. Automated cancellation alerts with SOH adjustment prompt.

🖥️
POS · Bulk Clear
Notification stack · Bulk action
Design in progress
🔴
POS · Alert Hierarchy
New order vs reminder vs system
Design in progress
📦
POS · Cancellation Alert
Auto SOH adjustment prompt
Design in progress
Design 03

ECF — accuracy upgrades

Product images on picking screens to reduce same-bottle errors. Customer names visible on UCON orders without opening the detail view. BWS online orders surfaced at the top with filter options. Sound notification on order arrival.

🖼️
ECF · Product Images
Picking view · Image per line item
Design in progress
📋
ECF · Order Priority + Sound
BWS order prominence · Audio alert
Design in progress
Outcome

Research delivered. Designs in progress.

The research report and prioritised recommendations were delivered in February 2025. Twelve recommendations across two tracks — order notifications and order management — tiered by priority and effort. The design work is currently in progress and will be updated here when complete.

Participants
7
In-store interviews across high and low volume BWS locations, mix of seniority.
Findings
15
Documented across three platforms — RF device, POS, and ECF.
Recommendations
12
Prioritised across order notifications and order management, tiered high/medium/low.
Design areas
3
RF device, POS, and ECF — each with targeted improvements. Designs in progress.
In-store research — 7 interviews, 15 findings
Feb 2025
Delivered
12 prioritised recommendations — notification & management
Feb 2025
Delivered
RF device — NotifyMe redesign & default settings
High priority
Designing
POS — bulk clearance, alert hierarchy, cancellation flow
High priority
Designing
ECF — product images, order priority, sound alert
Med priority
Designing
Pilot testing — store trial of RF improvements
Next
Planned
Measure: RF adoption rate, picking errors, NPS impact
Post-pilot
Planned
Reflections

What I'd keep. What I'd change.

— What I'd do again
Going in-store, not just interviewing over video. Seeing a team member try to serve a customer while 35 POS pop-ups stacked up was worth more than any survey. The physical constraint of the counter, the single RF device for the whole store, the positioning of the ECF screen — none of that shows up in a remote session.
Connecting notification failure to NPS and refunds. The brief was about notifications. The finding was about tool trust. The business case is about NPS and picking accuracy. Reframing the scope early got the right stakeholders in the room.
Separating order notifications from order management. They overlap but have different owners, different timelines, and different technical constraints. Bundling them would have meant slower progress on both.
— What I'd do differently
Quantify the error rate before starting. We knew picking errors were happening. We didn't have a baseline rate to measure against. Going in with even a rough refund-rate figure would have sharpened the business case considerably.
Visit more stores before forming hypotheses. Seven stores gave us clear patterns. But high-volume stores in dense metro areas and small regional BWS locations behave very differently. A wider sample in round one would have given us more confidence in the RF recommendation.
Involve engineering earlier on the NotifyMe cap. The 47-notification limit turned out to be a third-party app constraint, not a device setting. That changed the technical approach significantly — and we found out late.