Overview

There are a huge number of metrics Auto Trader reports back to our dealer customers. Metrics such as number of times an advert page has been viewed change over time and change depending on how the experience of Auto Trader changes. The one constant is the number of direct and indirect connections Auto Trader facilitates between consumers and our dealer customers.

Those direct primary connections; calls, emails, chats, texts, provide dealer customers a chance to further the purchase process with a consumer. These can also include finance quotes and part exchange valuations facilitated on the Auto Trader platform and passed over to the dealer. As the automotive world moves towards a more online transactions, we need to start thinking about how these interactions between consumers and customers can become more and more transactional. Whilst increasing the number of users visiting both the site and the advert pages is beyond our control of this project, the primary focus is to encourage a larger share of users visiting an advert page to make one of these direct connections.

Recent Auto Trader research has indicated that around 65% of ‘walk-in’ visitors to a dealers showroom have used Auto Trader to research the seller prior to their visit. Those indirect actions, clicks to a dealers website, map views, directions, are a way to demonstrate value and influence to our dealer customers. Whilst making the primary direct actions our priority it is important to make sure we are still demonstrating value with these more indirect connections.

Problem statement

This is an ongoing project with goals evolving as we learn more and more. Our current focus and OKRs are as follows;

Generate more, valuable leads for our customers from our current consumer base. We will achieve this primarily by encouraging a larger share of consumers to send leads, but also by making it easier for consumers to send multiple leads.

We generate more primary leads (calls, emails, chats & texts)
The share of our audience that sends a lead increases
Consumers send multiple leads

Key Result 1

Increase number of hard leads (normalised by stock and users)
1. Initiated - Measured to point of conversion on AT (email lead submission, click to call/text/chat) - A/B tests.
2. Engaged - Measured the actual conversation starting (email lead submission, IVOX call data, CAO chat/text) - not A/B tests.

Key Result 2

Increase FPA view to hard lead conversion

Health metrics

Soft leads (Website click, map clicks, directions).
Leads per session (monitor this as it’s not something we have full influence over).
Finance leads increase in line with KR1.
No pushback from influential dealers around reduction in website clicks and profile clicks.

Users and audience

Consumers on site who either wish to purchase a car or have a question for a seller.

Roles and responsibilities

Lead product designer for this project working within a team consisting of a product manager, tech lead, a team of developers, front end developer and QA. This project included close collaboration with the research function within Auto Trader and the agency Endless Gain.

Scope and constraints

In Auto Trader there are always many moving parts and this project is no different. With a team also working on the advert page but concentrating their efforts around pricing and confidence, we had to wait until they had a stable version of the page before we could start our work in there. This meant we could focus on the on the end of the connection first and look at optimising the contact form.

Our process

All good projects start with research and this is no exception. Endless Gain were engaged to provide an unbiased overview of consumer lead journeys and this was combined with our on site data through GA, Looker and ContentSquare. With an informed view of current usage and problems we took a step back to look at a consumers decision process on an advert page. These decisions were distilled down to the following:

Im considering this car:

User can add the car to a favourites list to return to another time. From their list they can easily return to the advert and also compare that car against others within their list.

I have a question:

User wants to know a little bit more about the car. They either want an instant answer or are happy to wait. The methods of contact can also reflect a users confidence level in their conversations with a seller and their technical knowledge.

I want to buy this car:

User has decided this is the car for them. They may have already have finance in place or want to arrange finance with the seller. User has the ability to tweak those finance terms and include a part exchange as part of the process.

This car is not for me:

User has no interest in this car and has discarded it from any consideration.

With these decision options in mind and with a view of how other industries facilitate the connections between buyers are sellers, a number of assumptions were formed to frame our thinking.

The ‘why’ not the ‘how’

Focus on lead generation moves from the method of contact to ‘why’ the user wants to contact the seller. Do they have a question about the car or are they ready to proceed with the purchase.

Price and actions align

The affordability is the most important element to progression. Align the pricing and the contact options closely to increase conversion to lead. Also these elements should be visible most of the time.

A process not a page

When a user has made a purchase decision, finance and part exchange form part of a ‘checkout’ process rather than happening on page. Mirror common checkout patterns and processes.

Conversational in tone

As part of contact with the seller both tone and interaction are conversational. A friendlier and more helpful tone will increase conversion and make the change to a ‘process’ easier.

Availability

A user is more likely to generate a lead if they know if the seller is available or if not available provide estimated response times or a call back / message back service or an alternative option.

With goals and assumptions set, the team were engaged in a collaborative ideation session which had proved successful within the new car project. With a stack of ideas we are ready to tackle the first part of the project

Form optimisation

With another team engaged in work on the advert page, our project started at the end of users connection journey, the enquiry form. A series of rapid A/B experiments were pushed out in an effort to both increase conversion rate and efficiency of the form.

To begin with though, the tech the form was built upon was old and buggy and in bad need of an update. So the form in its current state were rebuilt in new tech to provide us with a solid base to iterate from.

The following tests were run;

Remove main navigation from form - flat, test accepted
Switch to floating label input fields - 0.5% increase in conversion - test accepted
Non sticky button on form - -3% decrease in conversion - control accepted

The next test we felt would be the silver bullet, through many user research sessions and ContentSquare analysis, the ever popular GDPR consent check box came back time and time again as a barrier to conversion. So was with great anticipation, after negotiating with our security and policies team, that we are able to test the form with a consent statement rather than a check box.

Remove GDPR checkbox - 1% increase in conversion

I think, underwhelmed summed up our reaction to that result. Sometime no matter what the research or data tells you, you can never truly know until you test something in the wild.

Undaunted we cracked on with our experiments.

Replace checkboxes with pre-populated message - Flat, test accepted
Side by side name fields - 1% decrease on mobile, 1% increase on desktop - test accepted on desktop, control accepted on mobile
Captcha update (replacing the small captcha square which was interfering with and overlapping the sticky button) - 1% decease on mobile, 2% increase on desktop. Overall conversion was flat so control accepted

Having run through a number of tests, we had made little ground in terms of conversion, but we had made good strides in terms of;

A form built on the new code base
Around 200px reduction in the height of the form
Around 25% reduction in terms of page weight
Around 25% reduction in load time of the form

With the other team now at a point of stability with their work, we could now look at the initiation of leads on the advert page.

FPA contact optimisation

Beginning with mobile, we set about experimenting and testing with the way a user makes those direct connections with dealer. Our first port of call was the sticky contact options on mobile. With a series of rapid fire A/B tests we looked to optimise the ‘how’ of the section and also the visual execution and affordance of those options.

The following tests were run with the current execution acting as control until a test is accepted:

Pill shaped red call button plus icon and copy for email, text, chat - Test failed, email, text and chat saw uplifts but these did not cover the drop in calls
Block shaped red call button plus icon and copy for email, text, chat - Test failed, email, text and chat saw uplifts but these did not cover the drop in calls
Block shaped green call button plus icon and copy for email, text, chat - Test failed, email and chat saw uplifts but these did not cover the drop in calls and also texts

We then moved away from our theory of presenting the user with all the possible contact options and moved to a two option configuration, with the options being what we felt, through consultations with dealers, that were the two most important, call and email

Block shaped green call button plus icon and copy for email - Test failed, but by a minimal amount compared to previous tests

From there we refined our two option configuration to something more mobile focussed, call and text. This gave users a direct interaction and a delayed interaction with a seller.

Block shaped green call button plus icon and copy for text - Test passed with a 2% increase across initiated leads

This version was accepted as the base and we moved on to tackle the other contact options within the page and the main contact options on the desktop experience. The dealer contact module within the advert page was lacking in visual hierarchy and an overload of options. We set about fixing the hierarchy with clear defined options. A primary call to action, on mobile this is ‘call’ and on desktop this was ‘email’. A secondary set of actions comprising the rest of the direct options, text, chat and email / call dependent on device. Under that was presented the soft lead options comprising of website, location and directions actions.

As of writing there are 2 versions of this test currently live:

Test 1: all soft options displayed: Currently 3% increase in initiated leads across desktop but a 0.5% decrease on mobile. The test has seen a small decrease in the number of initiated soft lead actions.
Test 2: directions action has been removed: Currently 2% increase in initiated leads across desktop but a 1% decrease on mobile. The test has seen a larger decrease in the number of initiated soft lead actions compared to test 1.

With some confidence around the configuration, it is time to make some more fundamental changes to the advert page and the contact system. Looping back to one of the initial assumptions, around the alignment of price and calls to action, the direct contact actions are moving out of any dealer related module and sitting with the pricing information. The contact then becomes more about the car and less about the seller. Moving to a more recognisable e-commerce pattern at the top of an advert page consisting of the ‘What’ (What is the car?), the “How Much? (Can I afford it and is it a good deal?) and the ‘Next’ (How can I buy this car / contact the dealer?). Aligned to this test, is a new iteration and behaviour of the mobile sticky contact options using the design system pill buttons in a two button configuration and a delay in the appearance of the contact bar until after the user has scrolled past the in page options.

Contact button and pricing alignment - current brand styles

Contact button and pricing alignment - future brand styles

A system for leads

In the background of this optimisation work is a rethinking of how Auto Trader moves to a more e-commerce check-out experience. Weaving in our part exchange, finance and insurance products into a systemic flow for all onward journeys post advert page decision. The work for this is only just starting, investigating patterns to house these checkout style experiences, with the hope of aligning around a north star of a fully online car buying process.

View prototype

Outcomes and lessons

With this project far from over, it is hard to quantify outcomes just yet, it is merely the start of a journey for the future of Auto Trader. There have been some valuable and hard lessons learnt, from collaboration with other teams and their objectives, to the non success of tests you were sure were definite winners.

Project update:

The new contact seller module has proved very effective in driving hard leads (phone calls, email, chats and texts) to retailers. On mobile, total hard leads are up 5% and performing even better on desktop, up 13%. The challenge for this project has always been to balance any increase in hard leads with any changes to the softer lead metrics such as map and website clicks. At the moment the reduction in those metrics is not at an acceptable level.

Leads, Leads, Leads