Self Inspection : Revolutionizing the Car selling experience for millions
UX DESIGN | RESEARCH | PROTOTYPING
My Role and Responsibilities
-
Analyzing data
-
Conceptualizing new ideas
-
Wireframing
-
Prototyping
About the Project
-
Project Duration : Jan'21 - Present
-
Team : Mrunal Dhaygude, Shalab Vaishnav, Nirmal tandel
Tools used
-
Figma
-
Sketch
-
Miro
About OLX Autos
What is OLX Autos? What do we do?
OLX Autos is one of the 30+ companies owned by the OLX Group and has presence in several countries like India. Mexico, Chile, Argentina, Columbia, Indonesia etc.
OLX Autos is an online Car trading platform. User can sell their Car, buy one from the car listings from the platform or get their purchase financed by OLX. Selling, Buying and Finance are the three verticals OLX Autos deals with. They have Android, iOS as well as Web platforms.
This Project fell under the selling vertical, we later explored its implementation in Finance as well.
Project Overview
Why did we need a Self Inspection capability?
Everyday thousands of users come on the OLX platform to sell their Car. Currently, for every user OLX first collects information about the user's car online and then if the user proceeds to book an inspection, an offline inspection of the Car takes place. For the inspection an expert from OLX inspects the car either by going to the user's home or by calling the user to the store.
In this entire process, OLX incurs operational cost and the offline process increase the time taken for a sale. Offline processes also increases dependency on factors like availability of an inspector.
Solution
To solve this problem we designed a online car inspection capability. Using this the user can provide us details about their car that will help us estimate the car's condition. These details include basic car details, car condition details as well as car images. Our system has the capability to analyse user inputs to study the condition of the car. Users can also take images of the car using our smart camera. The smart camera guides the user to take the right images of the car and also detects damages, cropped images, blurred images etc. Using all this data the data science model is able to generate a price quote for the car.
The below video gives a thorough representation of the product :
A demo video of self inspection
Final Impact
Exponential increase in lead ~ 5x
Sellers were now able to put their cars to sell to OLX directly from their homes which led to an enormous generation of leads.
Increase in procurement of cars ~ 2x
Sellers were now able to put their cars to sell to OLX directly from their homes which led to an enormous generation of leads.
What to expect
1.
Research
2.
Ideation
3.
Improvisations based on research
4.
Phase 2
Research
The research plan was broken down into two phases :
1. Pre launch research
2. Post launch research ( we will know more about it as we go further )
The first phase consisted of understanding what could be termed as self inspection.
For this we did the following things :
Interacting with Inspection Engineers
In order to understand what are the parameters basis which we could decide a car's price we interacted with inspection engineers who determine prices on field.
Understanding the user persona of sellers
To under who we are designing for we took up the task of analysing the characterstics of the sellers that often visited OLX by studying existing data
Competitive analysis
Analysed the current solutions present in the problem space
Research Insights
1. Insights from Inspection Engineer interactions
We met with the inspection engineers to understand what are the car parameters necessary to give a close to final offer.
Basis the information shared by the Inspection engineers we need the following kind of data from the user for self inspection
Car Basic Details
Working condition details
Car Photos
2. Insights about User Persona
To understand our user in two major markets : India and LATAM we conducted a research activity with the market teams and also analysed the user data to come up with these prominent personas.
.png)
.png)
.png)
3. Insights from competitive analysis
Two get an idea of how other platforms which provide a similar capability were performing self inspection, we did a secondary research analysing other platforms. All the platforms analysed were not direct competitors, some were different industry apps like insurance apps.
We found the following to be the two broad ways in which Self Inspection were approached.
1. Form based approach
-
Fill-up form (Type/Typeless)
-
Uploading Images (Via gallery or camera)
-
Uploading Video (Via gallery or camera)
-
Combo - Form+ Images+Videos
-
AR using camera
2. Assistive approach
-
Assisted Interfaces
-
Conversational UI
-
Camera UI
-
Augmented Reality based Help
-
Assisted form filing (with call center)
Ideation
The idea is to build a experience that enables users to inspect his/her car from the comfort of their homes. Based on the data point requirements, user personas and ideas from the competitive analysis we started our ideation process leading up to the final solution.
The steps involved in the ideation process were :
STEP 1 : Brainstorming concepts to build on
STEP 2 : Finalising the concept based on feasibility and easy of use
STEP 3 : Finalising the workflow
Step 1 : Concepts
We started with brainstorming ideas based on the research and the requirements and came up with concept cards without constraining to any limitations.
Here are few of the concepts from the list :
Step 2 : Finalising the concept
After having listed down the pool of possibilities we sat down down with the team to understand the technical constraints and the timeline of launch at hand.
Considering all the factors here's what we finalised on :
Overall approach :
Form Filling + Images
Ways to Capture basic Details :
1. Manual form filling
2. Scanning the Car Registration certificate
3. Using a car detail fetching API based on just the Car number
Ways to Working Condition details :
1. Manual form filling
2. Uploading specific Car photos
Step 3 : Finalising the workflow
Having decided the approach and the ways of going about self inspection, the final step before going towards the final designs was the workflow.
Ways of entering data in the sections
Final Self Inspection workflow
Final Designs
Different ways of filling Basic Car details
1. Uploading a picture of registration certificate
The basic details of the car are fetched by scanning the image
2. Fetching details from the Car number
The basic details of the car are fetched by an API using the Car number
3. Manually filling Car details
Car's Working condition details
Uploading Car images
1. Uploading images from Gallery
2. Taking Car images using our smart Camera
Users can take images of the Car using the stencil on the screen for assistance.
The stencils guide the users in taking the right images.
The AI model analyses the uploaded photo to make sure that the photo is not cropped, blurred or is taken from too far.
An example of error analysis
Some research insights about the camera flow
_%20Report.png)
In order to help users take pictures in constrained spaces while still giving us a complete view of the car, we introduced angled images.
Taking an image at an angle takes lesser space but also captures the entire car
PHASE 2: DIRECT AUCTION
Current post self inspection process :
Self Inspection
Offer communication
Home Inspection
(If offer accepted)
Car is put for auction on the dealers app
Final price communicated based on auction
1
2
3
4
5
What do we plan to do here?
While self inspection automated a huge part of the inspection process, still in order to put the Car for an auction we need to perform a physical inspection. The physical inspection gives the dealers more data points and trust about the validity of the details in order to be able to bid on the Car.
We took a step ahead and worked towards making Self inspection even more detailed and made the entire auction process digital as well. Now, the only time there would be physical intervention is at the very end for picking up the Car!
Process
Understanding from Dealers about the data points required to be able to bid on a car
Modifying the Self Inspection experience to accommodate dealer requirements
Final Designs with next data points and capability to track auction progress and bids
Insights from research with Dealers
From the discussions with the dealers we wanted to understand what is the minimum information about a car they would need in order to be able to bid on the car. basis that we coul dmodify the self inspection experience to accommodate the same.
Additional information needed by the dealers :
Additional Car basic details
Loan details
Warranty details
Additional Car condition details
Gearbox condition
Engine Oil condition
Additional Car photos
All four tyres with threads
Car Dashboard
Seats
Engine
Final Designs
1. Additional Car images required for Dealers
Earlier while conducting self inspection, the user only had to upload 6 images including images of front view, right view, left view, back view, odometer and seats.
In this case, for dealers to be able to bid on a car they need certain specific images of the car like images of the tyres, dashboard, back seats, engine etc.
Car Tyre
Dashboard
Car Seats
Car Engine
In order to help users take these specific images we have new stencils and also small attachments on the screen for reference
2. Starting the auction after collecting all the details
Users receive a quote based on the details provided about the car. This is an approximate price quote range and not the final one.
Here, The users can now choose to put their car in auction.
After starting the auction the users receive real time updates about the auction status
Next steps
1. The flows we saw in the project are currently live. The direct auction flow is being run as a pilot experience. Next, we would be examining the user data to see the results.
2. Optimizing the flow using the data insights.