When Blooms Arrive Late
How late deliveries drain a florist's margin
Data Stories LabInteractive dashboard2,409 orders · Jan–Jun 2026
Live model on 2,409 real orders. Every view recomputes.
What happened
–
orders on time vs standard
–
refunded in the window
–
of refund money on late orders
–
on time on busiest days (>50/day)
On-time delivery holds, then dips in peak season
Weekly on-time rate against the delivery standard
Source: 2,409 orders, Jan–Jun 2026. Illustrative dataset.
Refund cost spikes with the peak months
Refund money (RM) by month
Source: order-level refund records. Illustrative.
Same-day is the fragile promise
On-time rate by delivery service
Source: on-time flag vs zone standard. Illustrative.
Most orders run through Klang Valley
Order share by delivery zone
Source: order counts by zone. Illustrative.
When it slips, it slips by hours not days
Orders by lateness severity
Source: lateness vs requested slot. Illustrative.
Why it happened · ours or the courier's
–
late orders in view
–
of delay hours are our handling
–
avg courier transit hours
–
avg dispatch slack (hrs)
In the city the delay is ours; outstation it is transit
Share of late-order delay by zone: our handling vs courier
Source: internal vs courier delay hours. Illustrative.
The outstation carriers own their delay
Share of late-order delay by courier: our handling vs transit
Source: internal vs courier delay hours. Illustrative.
In-house riders lead on reliability
On-time rate by courier
Source: on-time flag by courier. Illustrative.
Outstation lanes run later than the city
On-time rate by delivery zone
Source: on-time flag by zone. Illustrative.
Where we cut our own timing too fine
Orders by dispatch slack: time budget left when the order left the shop
Negative slack means dispatched after the slot was already at risk. Illustrative.
How it happened · the throughput ceiling
–
on time up to 35 orders/day
–
on time above 35 orders/day
–
busiest day volume
–
of orders fall on peak days (>35)
On-time delivery falls off a cliff past ~35 orders a day
Each dot is a daily volume level; on-time holds up to ~35 orders a day, then collapses
Source: on-time rate grouped by daily order volume. Illustrative.
The breakpoint in four bands
On-time rate by daily volume band
Source: on-time flag by volume band. Illustrative.
Volume climbs into the seasonal peaks
Average daily order volume by month
Source: daily order volume by month. Illustrative.
Same-day breaks first under load
On-time rate by volume band, same-day vs scheduled
Source: on-time flag by service and volume band. Illustrative.
What to do · stop the leak
–
refund cases in view
–
refund money at stake
–
refund money on outstation lanes
–
biggest single fix (RM)
Past six hours late, refunds jump
Refund rate by how late the order was
Source: refund flag by lateness bucket. Illustrative.
Outstation carriers are the refund engine
Delay-driven refund rate by courier
Source: delay-driven refunds by courier. Illustrative.
The occasion decides how badly a delay hurts
Refund rate on orders more than 6 hours late, by occasion
Source: refund flag among orders 6h+ late. Illustrative.
A few conditions carry most of the loss
Share of orders vs share of refund money
Source: order and refund shares. Illustrative.
Refund-leak fix list
Every refund assigned to one lever, ranked by money at stake
| # | Lever to fix | Cases | Refund RM |
|---|
Work it top down; stop when capacity runs out. Illustrative.