Summary

A cautionary story for product people on how to move fast to optimize conversion with data-informed, lean product development to sell gift cards, while consumer appetite is affected by a global pandemic.

I hope to inspire your teams to optimise your product and design craft as the world continues to struggle with COVID-19 into 2021 and (perhaps) beyond.

The background

2020 has been a tough year for us at Treatwell. The hair and beauty industry has been significantly affected by the pandemic, with many of our salon partners having to close their doors in lockdowns.

To help, we prioritised the work to integrate the services of a new 3rd party gifting and warehousing solution to enable us to sell physical gift cards online. Aside from their lucrative appeal, gift cards are a great way to create a credit line for our partners and help retain their clients. Plus our customers love our cards and we couldn’t think of many better times where gifting would be so top of mind as now. The timing couldn’t be better!

Our product already supported gift cards, both physical and digital, through our (legacy) gift card page. The new provider enables us to scale our offering to better handle peak times (like Christmas) with cards made from eco-friendly material.

While in that space, we also identified and fixed a bug with the gift card checkout journey on mobile, which was taking users to our desktop checkout. A no brainer to improve conversion (and thus sales).

At the request of our Head of Customer Support (CX), we announced the feature with a “coming soon” note, hoping to prevent the usual customer support contacts asking how to purchase physical gift cards. We hoped this would also create a bit of excitement and buzz, without negatively impacting digital sales.

The problem

Shortly after making these changes, we noticed a problem.

Both YoY conversion and sessions were, coincidentally, 25% down. Fewer sessions didn’t surprise, as we invested less in performance marketing, but the drop in conversion did, especially given the lower sessions.

We discovered that Mobile Web in the UK was the main culprit, with progression to checkout down 50% YoY. Desktop and our other 10 non-UK markets seemed ok, although progression was down everywhere. The checkout page conversion was also ok.

My team was in the middle of delivering another important major feature, so we had limited development capacity to devote to this, but given the opportunity, we decided it was worthwhile to spend a bit of time to try to fix this.

The solution

After digging deeper into the numbers (with help from Erin, our product analyst), we decided to set ourselves the ambitious target of improving progression to checkout on mobile by 2x, ideally without causing more contacts to CX. Historically, we knew gift card traffic and sales continue to grow steadily and spike around Christmas Eve so we had (some) time to improve things for those last-minute shoppers!

But we did not have the time, nor the traffic, to A/B test our changes so we did the next best thing: get user insights to identify the possible problems by enabling a Hotjar heatmap and poll on the gift card page.

Inspired by our research (both qualitative and quantitative), we progressed to some lightweight cross-functional ideation (involving a few other teams like marketing and the wider product team) to come up with a list of hypotheses. We debated, scored and prioritised them with RICE (my own personal favourite prioritisation method) applying extra weight to low effort changes, for obvious reasons, before designing and implementing the changes.

We considered around 20 ideas, but in the end, we went with our five best bets:

• Revert the earlier mobile checkout bug fix (n/a)
• Push digital printing (3)
• Un-announce “By Post” with some new copy (5)
• Make the card expiry more prominent (7)
• Some cosmetic UI changes (1,2,4)

The result

In the end, we didn’t manage to deliver the integration and enable physical gift cards sales in time for Christmas but our changes did seem to make a positive impact (although without A/B testing, we can’t be confident it was due to our changes, as opposed to improving consumer appetite, for example), with progression improving by 66% from 8.5% to 12.8%, in the lead up to Christmas.

YoY mobile conversion in non-UK markets jumped 23% up but the UK remained down. Ultimately we feel that negative consumer sentiment in the UK (evident from various signals) was a route cause, but equally plausible, was advertising physical card sales that never came. With this, we had inadvertently given customers a reason not to commit purchasing digital cards instead.

Nevertheless, we moved fast, chose smart hypotheses backed by data, reacted well to market changes and ultimately made a positive impact.

In the next section, I’d like to share 3 of our main lessons from this initiative. This is the juicy bit so if you got this far, read on! I hope these learnings can provide you and your team with some cautionary, inspiring advice to improve your product and design craft as the world continues to struggle with COVID-19 into 2021 and (perhaps) beyond.

So. What did we learn?

1. Move fast, but build the right things and things right.

Give yourself an out, beware of hacky solutions. Use a feature flag to enable features. We didn’t when first announcing “By Post” coming soon, so reverting our change ended up too expensive to remove.

Stick to the right process, even when pressed for time. Don’t compromise and choose objective, unbiased criteria to critique and prioritise your best product bets. In retrospect, our RICE scoring was subjective, so in the future, I plan to at least use my team’s collective scores to avoid bias.

Don’t over-rely on copy. Copy matters (design starts with copy is a product principle of ours) and it’s the ultimate quick win / lean experiment, but recognise it’s limitations. Users tend not to read copy, and we overestimated the impact of un-announcing “By Post”, rather than removing the option altogether to push digital sales.

2. Follow your (team’s) product spidey sense and intuition.

What’s better than your own intuition? Your team’s collective one.

Confronted with the lack of development capacity, the prospect of increasing CX contacts, and other reasons I mentioned above, I considered but didn’t follow the suggestion of my Product Designer (Jon) to prevent making users think by removing “By post” altogether.

Product is all about making hard choices, often with limited information. I stand by my decision, but it’s a good reminder to stay open-minded to other viewpoints.

3. Pay attention to the right data. Solid data trumps intuition.

We spotted that the mobile checkout bug fix caused issues (likely due to our slower mobile checkout app) and solved it relatively quickly. This appeared to immediately help YoY conversion, particularly in the UK, despite the worse UI. This is a great reminder of the impact of application performance on users’ success!

In absence of solid, abundant data, compromise wisely. Data builds confidence, but ensure you use broad, reliable sources of data and be wary of low volumes. This of course is a well-known limitation of qualitative research.