A different mental model
The form looked and behaved very differently from the familiar monolithic version.
inDrive · iOS & Android, 2024Delivery
Delivery
order form
Redesigning inDrive’s courier order form to improve conversion and adapt to regional requirements.
Explore the case ↓
Context
Conversion dropped after the redesign
inDrive moved from a monolithic to a microservice architecture, rolling out a new design alongside the codebase update. For courier delivery, the changes went further: new UI components, a revised UX, and new market-standard features created a very different experience.
After launch, V20 (View to Order) conversion fell by an average of 16% across all cities.
−16%
Average drop across all cities
The starting point
The new form changed a familiar flow
The new form had to support a growing service. But the change disrupted the way people were used to placing an order.
Hypotheses
Too much to take in
The long, single-page layout made ordering feel heavy and demanding.
The price came too late
Users expected to check the price after setting point B, to compare services or decide whether the price was acceptable. Repricing was now at the end.
Constraints
Why we couldn’t roll back
Rolling back was impossible: the old form belonged to the phased-out monolithic architecture and lacked essential new features.
A main reason for the drop
Price moved to the bottom of step 2
Users could no longer check the price immediately after setting point B. Moving the price block to the second step was one of the main reasons conversion dropped.
The design constraint
Price depended on earlier fields
We couldn’t simply move it to the first step: both the final and recommended prices depended on information entered in the preceding fields.
Three design approaches
Three ways to make ordering easier
I organized a brainstorming session with the courier design team. Each designer explored a concept, then we refined the ideas for UX testing.
The brief: familiar, visually light, quick to complete, and clear about value and price — while still collecting the information needed to match and fulfil an order.
USP focused
Start from familiarity
A replica of the legacy form, exploring an earlier price field and making the service’s benefits visible.
Cognitively unloaded
Make one page feel lighter
A simpler single-page form, with secondary details moved out of the main flow to reduce cognitive load.
Matching based
Guide the order step by step
A more radical, staged flow that integrates the information needed for matching and successful delivery.
Quantitative UX research · Kazakhstan
Comparing three forms with real users
I organized a quantitative UX study to compare the three solutions. Real inDrive users received a push notification inviting them to complete a task in a prototype.
The task“Imagine you want to send flowers to a loved one using inDrive’s courier service. You need to deliver flowers from Dostyk Avenue, 82 to Abay Avenue, 2. You’d like both pickup and drop-off to be at the entrance. What would you do in this case?”
What we measured
Successful orders
The share of users who started the prototype and successfully created an order.
Cross-service journeys
Users who started entering points A and B for a city ride, then switched to Courier.
Missing-details errors
Users who tried to create an order without completing order details and encountered an error.
Prototype journeys & heatmaps
The original test artifacts show where users interacted with the forms.
UX test results
Concept 02 was selected
The second option won by a small margin, balancing a reasonable completion speed with a high rate of successful order creation.
The third option had the fastest completion time. This was partly due to the intentional absence of an error when users skipped the Order Details field.
We took Concept 02 into an A/B test.
| Metric | Short form (3) | 2 step form (2) | Monolith like (1) | Current NO |
|---|---|---|---|---|
| Avg time to complete | 65 seconds | 80 seconds | 157 seconds | 85 seconds |
| Avg time to overall | 43 seconds | 101 seconds | 72 seconds | 53 seconds |
| Total participants | 257 | 264 | 172 | 178 |
| Completed | 56 (22%) | 70 (26%) | 49 (29%) | 33 (19%) |
| Skipped | 32 (13%) | 26 (10%) | 21 (17%) | 28 (16%) |
| Dropped-off | 169 (65%) | 180 (68%) | 102 (60%) | 117 (66%) |
| Task complexity | 70% | 73% | 41% | 64% |
A/B results
+7%
V20 uplift versus the control group
from product changes
A/B results varied by country
The new form significantly compensated for the conversion drop. But results varied greatly between countries, and design was only part of the explanation.
Microservices bugs mattered
A technical deep dive revealed that part of the conversion drop came from bugs introduced during the transition to microservices.
Some orders could not be delivered
Separating supply types made certain order categories undeliverable in some cities. This helped explain why the A/B results differed so much across countries.
Regional differences
One fixed form
could not fit every market
Order conversion varied by region. Analytics showed different patterns in required fields — such as the recipient’s phone number and package description — and in the order people completed them.
A universal form with the same mandatory fields everywhere could not provide the flexibility needed to optimize conversion in each market.
Order Form Builder
A form configured for each market
We developed an Order Form Builder based on the tested Concept 02, with refined spacing and field organization for consistency.
Required or optional
Customize which fields a user needs to complete.
Field visibility
Show the fields relevant to each market.
Field order
Adjust the sequence to regional completion patterns.
Final results
+2.3%
V20 increase after launching the builder
Conversion recovered
Through a series of improvements, including the new order form and the builder, we fully mitigated the conversion drop and achieved a 2.3% increase in V20 after the builder launched.
We also gained a clearer understanding of regional courier behaviors and form completion patterns. The journey to a courier order often begins outside the courier section itself.