Lovehoney - Data Science and Advanced Analytics
For online adult toy and lingerie retailer Lovehoney, content is everything. For each new product launch, they hire video presenters, send out samples for review, retouch photos, produce buying guides and create promotions. These initiatives come at a considerable cost to the business.
So, Lovehoney came to Forward3D with a simple question: “What is it that we do that actually matters to visitors?” Forward3D’s data science and analytics team set out to understand the impact and extent to which Lovehoney’s content influences a customer's likelihood to purchase. The team created a plan to leverage the wealth of analytical data accumulated by Lovehoney to identify the type of content that directly impacts sales.
Forward3D have been working with Lovehoney, the UK's most popular online retailer for adult toys and lingerie for nine years across content, paid search, display and data science.
Content plays a substantial part in Lovehoney’s business and these efforts need to be meticulously planned and executed, as a ‘how to’ video could be the make or break factor for a customer when deciding whether a product is right for them.
Producing content pieces come at a considerable cost to the business in terms of time and resources, and without a thorough understanding of how effective they are, can be wasted efforts.
Consequently, Lovehoney tasked Forward3D with determining the extent to which their content efforts impact on sales and ultimately, benefit their bottom line. Forward3D needed to help Lovehoney make informed decisions using historical data, in order to recognise when and where to invest in promotional videos and buying guides. By doing this, Lovehoney were able to understand content that was producing a measurable impact on conversion.
Innovation and Creativity
Forward3D specialise in utilising data to our advantage. The team targeted existing users already on the website, who had an interest in the products but may not have necessarily converted into a paying customer. They could then collect and interpret historic data at a highly granular level, enabling them to understand how users interacted with the website and how they were impacted when on a certain customer path. To achieve this, firstly, site interaction data was segmented and collected to include every product view, visitor on the site and multiple sessions. Stitching this information together allowed us to determine which products and media would impact customer buying decisions and where this fell in the customer journey. By doing this, the team were able to learn how to best optimise the site navigation and product page content to ensure visitors are presented with the most relevant materials, leading to higher conversion rates. For ease of use, a custom Google Data Studio dashboard was also built to add flexibility and user-friendly display for Lovehoney so they can use it without our team to provide analysis for any questions or scenarios they had.
Execution / Strategy
Working closely with Lovehoney, Forward3D split the product page into content areas that have a cost to produce, such as videos, reviews and secondary images. They began capturing page interaction data at an extraordinarily granular level, which was then linked to site analytics log data.
This meant capturing every interaction, not just clicks but where the element was visible, for every product view, for every visitor to the site, with multiple sessions linked together to understand the impact to the user, over a period of time and not just for that visit. From this, the team could answer questions about an individual product but how a product view would ultimately impact what that user would buy. This allowed them to determine what products were popular, but also their presence in a user’s journey to aid conversion.
Systematic Data Collection
Data collection and processing was the biggest challenge as the vast amount of data meant we exceeded data import limits imposed by the Google Analytics API. To circumvent this issue, we created new Google Analytics views specifically for these events. Additionally, the vast amount of data output was crashing both Google Data Studio and database servers. By optimising our code, splitting output across multiple tables and aggregating data by month we optimised for both speed and durability.
Initially we implemented machine learning to identify how different paths through the website affected conversion rate before moving on to identifying if the initial item ‘Added to Basket’ affected the conversion rate. Finally, we settled on product page interactions as it was the most useful and actionable for Lovehoney.
The final part of our project was settling on a visualisation method for the vast amount of output data (almost eight thousand products). In the end, we created a custom Google Data Studio dashboard solution as it offered both flexibility for Lovehoney and the ability to display and break down data in a digestible way. This enabled us to leverage data to optimise product page layout by utilising our in-house tech stack and output a user friendly, insightful dashboard.
The impact of the analysis we provided on historic customer data, allowed Lovehoney to make strategic judgements on which product categories needed additional content investment to aid conversion and boost their bottom line. Importantly, the analysis has allowed Lovehoney to refine their content strategy by helping the business understand the key drivers that were eroding profit in order to sustain marketing, finance and merchandising efforts.
These data driven insights gave Lovehoney an understanding of:
The conversion impact of video (a near 50% increase in conversion).Forward3D drove budget efficiencies by identifying cost savings on unprofitable content efforts, allowing this to be reinvested incrementally in other areas of the business. Review aggregation being more impactful than reviews themselves.The development of a priority list of categories that require video (such as lingerie products).The creation of a priority of products and categories that require re-evaluation by Lovehoney’s content team.
“Forward3D have always been forthcoming in showing that they can apply cutting edge data science techniques to our marketing campaigns, such as causal impact studies, sub-channel attribution models by user types, and profitability analysis.
However, applying all that expertise to our site itself has been beyond illuminating and a real impact on our bottom line. This is cutting-edge conversion work, outside the realms of what is possible purely client-side, and Forward3D should be heavily congratulated for it.”
Matthew Curry, Head of Ecommerce, Lovehoney Group