Building a record-breaking Black Friday

Helping customers find a deal without the digital equivalent of a scrum

Product Designer · Discovery to launch · Worked with UX research, product, data and trading

The short version

Black Friday is the biggest shopping day of the year at The Very Group, and the website experience hadn’t kept up. Customers landed in a cluttered event layer, and colleagues had to put it together manually. With a small product team, I took the redesign from discovery to launch in two months, backed by five user studies and on-site experiments. All five hypotheses were accepted, and Black Friday 2020 became Very’s best ever festive trading period.

The problem

For years, Black Friday had worked the same way. For seven days, the site switched on a separate layer of categories, deals and rich content, so thousands of product lines could be pushed at customers while trading teams dropped new deals on the fly. Great for trading. Less great for everyone else:

• Customers faced a different IA and look just for the event, and it got messy fast.
• Colleagues in content were dealing with an increasingly manual process that would only get worse as we scaled.

Growing the event every year shouldn’t mean compromising the experience. Customers need breathing room to find the deals they want, in a place they recognise, and not the online version of shoppers squabbling over 85 inch TVs. We also had to bring our understandably nervous trading stakeholders along for the ride.

Discovery: finding our focus

We had years of research findings and our own hypotheses, so we kicked off with a discovery session with the research lead, product manager, BA and data analyst. It came down to three steps:

1. Lay our cards on the table: gather everything we knew about customer problems, and why each one mattered
2. Reframe: turn each problem into a “how might we” question, and iterate until everyone agreed on the wording
3. Narrow down: with tight milestones and limited resource, we had to be ruthless about where to play

Four customer missions came out of it:

1. Narrowing down: quick, easy filtering of deals, however customers prefer to find things
2. Product page lands: encourage customers who land on a product page to explore other relevant deals
3. Ways to pay: tell customers clearly and consistently how they can pay
4. Basket building: encourage relevant add-ons, or trading up

To see where we really stood, I ran a competitor benchmark across six research themes, scoring us out of ten against Amazon, Argos, John Lewis and Debenhams.

Amazon led on almost every measure, especially overall experience — while we lagged well behind, largely down to product discovery, deal range and the ability to compare.

I then mapped the end-to-end journey across all four missions to show the pain points and data insights in one place. It also made it much easier to explain to stakeholders where we wanted to focus.

Ideation

Before getting stuck in, I spoke to stakeholders about their ideas. They weren’t yet weighed down by technical constraints, which made for some refreshing thinking. I also put together a benchmark of competitor events, based on the retailers our customers already rate highly.

For each mission I sketched several ideas, then refined the strongest digitally alongside the stakeholder ideas. Every concept had to work as part of one end-to-end journey, so I created user flows and pieced the wireframes together to check it all held up.

We then presented the concepts to product and technical stakeholders, debated them, and voted. The winners either went into prototypes for user testing or were built for on-site experimentation. They struck a good balance between a clean, creative experience and what our platform could actually do.

Validation: five tests

The research lead and I planned the validation as a mix of user studies and on-site experiments. We split the experiments across two sales events before Black Friday, because Very isn’t in sale season all year round and these were our only chances to measure real-world impact. Each feature was tested in isolation, so nothing skewed the results.

Event page vs gallery

We replaced the bespoke event page with a familiar gallery, with filters and sorting, so customers could narrow down faster. It would also save the business from building a custom page every year.

What customers told us: they want to filter a long list of deals down to a shorter one, and when they’re unsure they fall back on search and departments. Result: +1.3% progression to product, +6.0% visitor conversion, and a 40% refine interaction rate versus 6% on the event page.

Content cards

Discreet content cards inside the gallery let us share brand and credit messages with the flood of new customers, without a takeover page.

What customers told us: mostly deal shoppers saw them, so they needed to appear on filtered galleries too, and show a customer benefit and not just a business one. Result: +1.9% visitor demand and +1.7% average order value.

Filter and toggle

A deals facet and toggle let customers narrow to event deals during a normal shopping journey.

What customers told us: they clearly wanted to see only event deals, but the UI needed to be more prominent. Result: +1.5% progression to product, and heavy usage, at around 100 interactions per 200 impressions. We iterated based on the study findings.

Top picks

Top Pick banners at the top of gallery pages let trading push deals at the right moment, without disrupting the shopping journey.

What customers told us: they liked time-saving recommendations but wanted to know why an item was picked. Some scrolled straight past, and some thought it looked like an ad. Result: +51.5% progression to Top Pick products, +51.1% visits to Top Pick products, +2.1% overall progression to product, and higher overall visitor conversion.

View more deals like this

A link on product pages showed similar deals if the current item wasn’t quite right. It also helped new customers who land straight on a product page see that more deals exist.

What customers told us: half interacted with it, though some missed it at first. It was a useful shortcut, but the similar deals needed to be more closely relevant. Result: +1.4% progression to another product, 1.0% lower bounce rate, and higher visitor conversion.

Launch: our most radical trading update in five years

With validation done, we had an end-to-end experience we were confident in. I showed it off across the business, from trading directors to technology forums, to inform people, get feedback and answer questions. Having the raw data behind our decisions did a lot to ease nerves before the biggest trading period of the year.

Results

Black Friday 2020 was a huge success for The Very Group, in a year like no other. These are business-wide numbers, not just the effect of our changes, but the January 2021 press release called out the experience changes as part of the story:

Read the press release

What I took away

• Get stakeholders closer, earlier. Working closely with trading instead of in a silo made a real difference. Next time I’d go a step further and get them into user testing, experiment playbacks and discovery workshops.
• Data calms nerves. Showing the evidence behind our decisions was the best way to bring nervous stakeholders along.
• It’s never finished. Black Friday comes around every year, so we started a post-mortem and a team retro straight away, and took what we learned back into discovery for 2021. As the technology strategy improves, we can be bolder and learn faster.