Give the UX team superpowers by enriching qualitative analysis tools

Business

Rightmove is the UK's number one property listing site, with over one billion hits per day.

Requirements

Rightmove.co.uk allows users to request a property valuation via a form from local agents.

However, although this form generates very substantial revenues, its submission rate seemed relatively low.

Rightmove has implemented Hotjar, which allows a large number of sessions to be replayed in the form of videos, in order to identify potential problems. However, the site counts 30 million sessions per day, which poses the challenge of knowing which visits are interesting to view.

Objectives

Answering the following two questions would generate millions of euros a year in additional revenue for Rightmove:

• Where exactly do users abandon the form?

• What prevents visitors from submitting a request for a property value estimate?

Proposed solution

• Add more granular events to each step of the form to identify quantitatively which steps are missing.

• Add new events to detect any errors in the form due to bugs or incorrect user input.

• Push user events and attributes from the data layer to Hotjar to tag automatically all users and their session videos. For example date_picker_failure, property_valuation_view or even user type : customer

Implementation

Project duration

The project lasted 2 weeks, from writing the specifications, to implementation by the team of developers, to setting up the custom Js tags in Google Tag Manager and analysis results.

Results

The first funnel analysis reveals two major drop-off points:

1. The first step (address submission)

2. The last stage (contact details)

Until now, the Product team was convinced that there was a gap between what users expected when they arrived on the page and what actually happened: a long form to fill in to pass on their information to a real estate agent.

However, our integration between Hotjar and the data layer events enabled us to filter the video recordings that seemed most relevant to view from the hundreds of thousands of recordings available.This enabled us to quickly identify that this was in fact an interaction problem:

1. The "Next" button was hidden by the "Cookie Consent" banner, especially on mobile.

2. The date field didn't work for some browsers.

3. The drop-down list for home addresses was outdated, making it difficult for many users to find theirs.

4. Some users had no agent results proposed for their address.

Further information

Some activation ideas based on the events collected :

1. Send errors detected in Zendesk as no_results, address_info_failure etc. to trigger instant chat assistance.

2. Send events to CRM to trigger a support procedure for identified customers who have been involved in the process without having finalized it.