PaDi UMKM - Decreasing Average Time from Delivery Process
+46% GMV from service transactions in 2024 +14% number of service transactions in 2024
Jul 30, 2022

Background
PaDi umkm is a B2B marketplace based in Indonesia. The platform was initiated by the Ministry of State-owned Enterprises & the President of Indonesia to help small businesses during the COVID-19 pandemic. The idea was to move some of the spending budget of the SEO (state-owned enterprise) from the big companies/vendor to the small business entities.
Our primary buyer are users from BUMN (SOE). These SOE users have unique behaviour during shipment process such us.
Choosing Private courier for shipment process
3rd party courier is considered as a service that need different budget allocation from their regulations
Most of them don’t provide that budget allocations
They tend to choose a final price that cover the product price & the shipping costs

Most of the seller chose private courier as their main delivery method
Objectives & Metrics
Starting Q1 2023 we started the monetization process, so we collected revenue from transaction fees. And we focused on these two objectives.
Our north star metrics is GMV & Revenue
To increase the number of GMV and get the Revenue as soon as possible, we need these two derived metrics.
📈 Increasing Conversion Rates between funnels
⏱️ Decreasing Average times between funnels
In this case, we need to decrease average process time during shipment phase
Decreasing average time from seller processing order → delivery
Decreasing average time from delivery → order received by buyer
Understanding the Problem
To better understand the problem and how it affects our metrics and objectives, I tried to collect more information about the condition of Average time between status in the Shipment process.
Stakeholder Interview & Data Checking
From the Squad leader and Product Manager, we knew that the average time from seller processed order → ready to delivery was really bad. It shows +5000 hours for the process.

Previous Design

Our previous design when seller want to deliver the package using private courier
The previous design when seller would deliver the packaged, we only asked AWB number, but the problem is that this AWB number is not quite related to Private courier
Most of Seller would put random number like phone number
Some seller who use 3rd party courier would put the AWB number from the courier
User Insights

From these data we understand that
Most of buyer chose to make transaction with seller/vendor in the same area or city, this fact is proved from our previous research about buyer preferences with seller location
The private courier was used for transaction in the same city that can be delivered using seller own transportation fleer like cars, motorcycle, & truck
So, if this private courier were used for transaction in the same city and in a day delivery, why would it average process time takes too long?
It is because
The delivery can’t be tracked, so we don’t know whether the package is well received by the buyer or not
We only depend on buyer confirmation on PaDi umkm platform
It made impact on business metrics such us conversion rates & average time
Problem Definition

Most of the issue is occured in delivery phase and absolutely it had impact on CVR & AVT metrics
Ideation

The goal is how we can apply Service Level Agreement (SLA) for private courier during the delivery phase?
By applying a SLA, we could manage the time for delivery process, and the number of transactions won’t be stucked in that stage.
Idea 1 : Shipment evidence as SLA trigger
Competitor Analysis

Gojek instant courier feature
This delivery evidence feature is quite common for instant delivery like go-send & grab-express. By using this evidence as trigger for us activate the SLA, we can change our dependency from buyer confirmation to our new SLA.
New User Journey

Design
We added new status between
“Diproses” -> “Dikirim” -> “Diterima”
changed to
“Diproses” -> “Siap Dikirim” -> “Dalam Pengiriman” -> “Diterima”
And asked private courier to upload evidence as new trigger for SLA

Result
In 2024 , after the implementation of this idea, average time during processing & delivering time decrease up to more than 90%
The conversion rate during “diproses” status also increase. Previously, so many order stuck not only in “dikirim” status, but also “dirposes”. This is due to seller often deliver the package without any confirmation

After we released the feature in Q3 2023, the average time from “Diproses” -> “Siap dikirim” decrease significantly from ±6000 hours to ±40 hours (this Quarter)
it shows that our initiatives gave impact on AVT metrics, that also helped to achieve GMV & Revenue sooner

Idea 2 : Give seller acces to change the courier
Competitor Analysis

SLA is quite common in ecommerce & marketplace rules of transactions, it was used to make something that uncontrolled (user action) into something that under our control.
New User Journey

Design

by allowing seller to change from private courier to conventional courier, we made from untrackable shipment to trackable ones.
Because of this we can use SLA as a rule, if the buyers don’t confirm the acceptance of product within 2x24 hours after system said it was delivered, system will automatically changed it to “received”
Result
We made our last adjustment (Design & Dev) in the first month of Q3 2024, and it shows a significant result.Our OKR is incrasing the number of shipping subs transaction from 113 to 220 during Q3 2024. But, from the 2nd month it shows double result from our initial target
Source Fulfillment looker

Other than number of transaction that is increased, shipping subs transaction also support as additional new revenue stream
It might not as big as take rate for seller service fee, but not only it helps to increase revenue , it also increase conversion rates as well

Conclusion
Focusing on move variable that we can’t control to the area of control, so everything will be more managable. By applying SLA (Service Level Agreement) to shipment process, it helps we to have more control on average time of the process.