Creating products users love through data

By Pedro Cerqueira, Product Owner.

This post was originally published on our F-Tech Blog. Come check it out here :-)

As a platform business that operates a marketplace, we are in the business of shaping demand.

Some users know exactly what they want when they use our technology products, but there are others that need a certain degree of persuasion before converting.

We need to design and build technology products that bring value to our users and catch their eye, hearts and minds.
We need to appeal to our users’ emotions and most basic human needs: physiological, safety & security, esteem, social and self-actualisation.
It’s up to us to: design and build products which leverage how the human brain thinks in both fast and slow modes [1], possibly making split-second decisions without conscious effort or awareness [2]. Human responses are shaped by heuristics, and cognitive biases inherited from our ancestors, and we’re now building online products to cater to these primal instincts
[3]. Our products must be contagious [4] and hook [5] our users by building on triggered actions that are both rewarding and habit-forming [6]. It’s up to us to create products users love! [7]

How can we know if we are doing a good job of creating products users love?

Or more specifically, which data points should we be leveraging to understand our product performance better? Looking at the quantitative sales numbers is one indicator to understand ‘how much’ users might love our digital products. The problem is that just looking at sales won’t allow us to understand ‘what’ our users are doing and ‘why’ they love the products we create. This analysis won’t allow us to understand how much more loveable our products could be, and it won’t allow us to better prepare for a possible disruptive future [8], in which our users eventually fall in love with and leave us for a competitor. Quantitative data alone might, in fact, lead us in the wrong direction making us believe that users really like the experience provided by our digital products when, in reality, they only love the items we sell.

That’s why we also need to pursue the qualitative if we truly want to understand if, and why, we are doing a good job. Surveys, questionnaires and lab usability testing can prove to be valuable tools in enriching our view of our users: if we ask the right questions, maybe we’ll get the answers we’re looking for. We can also look at the sentiment and voice of the customer analysis [9] to try and understand how our digital products are landing.

However, for these tools to work effectively, we are counting on users’ conscious and active collaboration, and this isn’t always guaranteed.
Amongst others, humans:

• Don’t always tell us exactly what they think/did (they omit/lie)
• Don’t tell us anything (they don’t care, don’t understand, don’t know how to express what they want, don’t want to)
• Tell us what they think they thought/did (are not always rational and sometimes tend to rationalise and make narratives to justify their choices and actions [10], they’re predictably irrational [11] and their memory is often flawed [10])

Even though this analysis might yield some useful insights, in itself it isn’t sufficient to understand how exactly users are engaging with our products and ensure we’re focused on optimisation in the right areas.

Looking at how users are engaging with our products to increase our understanding of it and why they love the products we create:

Research company CEO Paco Underhill wrote a book named “Why We Buy: The Science of Shopping” [12] based on hard data gleaned from thousands of hours meticulously observing customers’ behaviour in shopping malls, department stores, and supermarkets across America during the 90s. They tracked every aisle customers visited, shelves and items looked at, items picked up, put back in place or added to the cart and actually bought.
Amongst other interesting discoveries, his team described the “butt-brush” factor to a New York store president: women shoppers being far less likely to purchase if they are brushed from behind by a person, a display table, or a piece of merchandise while examining retail goods. After this, the store’s president immediately took measures into his hands to avoid jamming narrow aisles full of merchandise and sales increased substantially for items that were previously located in racks in these circumstances.
As in the example above, this intensive job of tracking and analysing data allowed the team to understand customer behaviour better and to help stores change their physical environment to adapt to the consumers’ minds, and ultimately increase sales.

Luckily, with e-commerce, we can tackle this problem in a much more efficient way. Modern tracking capabilities and Data Models can help us understand how each micro and macro conversion within an omnichannel (online and offline) user journey contributes to different business goals.

From a macro conversion perspective, e-commerce businesses might have products focused on order completion while others in assisting it (e.g. by increasing brand/item awareness, by increasing engagement, by lead generation, by increasing reputation).

From a micro-conversion perspective, products might allow for navigation, interaction or engagement based conversions (e.g. CTR from search, item viewed or picked up, entering the checkout funnel, adding an item to the cart, time on site upped a specific threshold) each possibly leading up to a macro conversion of its own.

In the end, analysts must stitch together data from sessions across different stages, channels, and devices to reconstruct and understand each specific user session, mainly by focusing on which pages/screens and actions took place.

A theory put into practice at Farfetch

At Farfetch, we are focused on connecting and understanding the full user journey across our different products, to better achieve our goal: “One Platform, One Single Journey, One Connected Experience changing the way people buy or sell luxury fashion across all physical and digital channels”.

Essential to achieving this is our commitment to session data and truly understand the micro (e.g. CTR from search in iOS Farfetch mobile app, CTR from recommended items in Android Farfetch mobile app, add to wishlist in Farfetch website) and macro conversions (e.g. order placed in Farfetch website), grouped by the different analytical dimensions (e.g. date, time, user, screen/page viewed, action performed, characteristics of the screen/page/action) that make the most sense per product and across products (e.g. Farfetch website, iOS Farfetch mobile app), throughout our users’ journey.

To enrich this analysis and achieve an even more complete view over the user journey and behaviour we can tie session data to:

• The marketing channels and specific campaigns driving traffic
• User and customer segmentation by parameters such as visit recency or frequency, total spend, browsing behaviour, loyalty tier, market, etc.
• User experience testing permutations
• Post-purchase survey comments and Net Promoter Score, Customer Service communications (e.g. “I wasn’t able to finalise the checkout process successfully on the WeChat Store”)

Let’s imagine the following scenario: there’s a new trend in which a significant proportion of our top female customers are visiting our physical retail store, Browns East, to try on clothes in the Connected Mirror and then buying them online through the Farfetch iOS app. With session data, we have full visibility of this user journey and a better chance at understanding this behaviour, ultimately informing our product roadmaps and delivering an enhanced user experience.

As a practical example, let’s say this generic behaviour is becoming widespread within our iOS active user base, but, for some reason, when the users get to the app don’t convert as often as would be expected. Let’s say we also have access to iOS users’ survey data which seems to indicate that these users are having second thoughts about how they looked and felt when trying on the clothes in-store.

What if we add to our roadmap the design of a feature which allows our users to take really stylish photos of the outfits tried in-store via the Connected Mirror and automatically sync them to their accounts to be seen in the iOS Farfetch mobile app to test whether having the photos closer to the conversion point might help enhance the user experience?

No doubt, the complexity of possible data combinations turns “creating products users love” into a really challenging job, but if we can equip our Analysts, Product Owners, Designers and Engineers with a rich Data Lake, robust User Behaviour Models and user-friendly Business Intelligence tools, we can truly work towards building the future of fashion.

References:

[1] Thinking, Fast and Slow, book by Daniel Kahneman
[2] Blink: The Power Of Thinking Without Thinking, book by Malcolm Gladwell
[3] Leaders Eat Last: Why Some Teams Pull Together and Others Don’t, book by Simon Sinek
[4] Contagious: Why Things Catch On, book by Jonah Berger
[5] Hooked: How to Build Habit-Forming Products, book by Nir Eyal
[6] The Power of Habit, book by Charles Duhigg
[7] INSPIRED: How to Create Tech Products Customers Love, book by Marty Cagan
[8] The innovator’s dilemma, book by Clayton M. Christensen
[9] Web Analytics 2.0: The Art of Online Accountability and Science of Customer Centricity, book by Avinash Kaushik
[10] The Art of Thinking Clearly, book by Rolf Dobelli
[11] Predictably Irrational, The Hidden Forces That Shape Our Decisions, book by Dan Ariely
[12] Why We Buy: The Science of Shopping, book by Paco Underhill