Product Strategy: How to Reduce Online Returns at Matches

We all know that online shopping offers limited engagement with the products. Buyers cannot touch the fabric of the clothes and try them on physically. Hence, return and exchange becomes more likely in fashion eCommerce.

Similar to other online fashion sites, Matchesfashion is currently facing a 30% returns of all orders, which has a negative impact on the profitability due to additional operational costs such as shipping, restocking, and repackaging. As the UK and the US being biggest markets, they saw the highest return rates among all 170 countries.

So what are the reasons for Matchesfashion’s customers who spent lots of time from wanting a new clothes, comparing, considering, and finally making purchase, end up return the products after receiving them?

Here are some key reasons I identified: the first one is necessary returns caused by poor product quality or wrong shipment, which can be resolved by improving quality assurance or shipping process. The second one is unnecessary returns that encouraged by lenient return policies, including buy to try, buy to pad and buy to borrow.Buy to try is probably the most common cause of high return rate. Having difficulties in finding the right size online, in UK, 61% of shoppers use their home as fitting room by ordering multiple sizes of the same item and only keep the one fits. This echoes the stat that 60% of the returns in MatchesFashion were due to sizing issues.

Another reason for unnecessary returns is where people buy more items that they intent to return so as to get a free delivery. The last one is where people buy an item, wear once for a certain occasion and return to store, which may be solved by putting a label outside the clothes.

To narrow down the scope, I chose the UK market as it has the highest return rate. According to Statista survey in UK, female aged 25 to 34 is the main luxury fashion ecommerce buyer. To be more specific, I define the target segment as british female, age 25 to 34, single and with high income.

Then, I searched for UK fashion ecommerce surveys and conducted 3 customer interviews to verify my assumptions on sizing issues. The results shows that: in the UK, women clothing has highest returns among all categories due to fitting issues. 52% of multi-size orders are returned, which means that customers tend to buy two sizes of the same item and return one of them. 

The interviews indicate the same pattern as interviewees all mentioned not knowing which size to choose simply based on the fit description and product measurement when they tried to buy clothes like tops and coats. And two of them admitted having ordered multi-sizes of the same clothes due to the ease of return.

With the market research results, we can determine the goal of this presentation, which is to lower return rate by helping customer find the right size. 

Firstly, to narrow down the scope, I chose the UK market as it has the highest return rate. According to Statista survey in UK, female aged 25 to 34 is the main luxury fashion ecommerce buyer. To be more specific, I define the target segment as british female, age 25 to 34, single and with high income.

Then, I searched for UK fashion ecommerce surveys and conducted 3 customer interviews to verify my assumptions on sizing issues. The results shows that: in the UK, women clothing has highest returns among all categories due to fitting issues. 52% of multi-size orders are returned, which means that customers tend to buy two sizes of the same item and return one of them. 

The interviews indicate the same pattern as interviewees all mentioned not knowing which size to choose simply based on the fit description and product measurement when they tried to buy clothes like tops and coats. And two of them admitted having ordered multi-sizes of the same clothes due to the ease of return.

With the market research results, we can set the product goal to lower return rate by helping customer find the right size. 

Now, having the sizing problem in mind, I use a customer journey map to examine at which stage did the problem occur, and how did user respond to it.

Let’s take a look at Annie, a 27 year-old UK resident and a regular online fashion shopper. Last month, she saw a passerby wearing a trench coat, which triggers her want for a trench coat. So she started to ask friends for recommendations and search online.

After comparing several different coats, she found the most ideal one. Just when she was about to place an order, she is getting confused and uncertain about which size to choose. Based on the product measurement, she should choose S, but the model who is taller than her is wearing S too. Thanks to the free and easy return process, Annie decided to buy both S and XS and return the one doesn’t fit. Although feeling a bit annoyed, the lax return policy eased her concerns about sizing and encouraged her to make a purchase.

From Annie’s customer journey, we can see that the lack of size & fit information at consideration phase coupled with the free return led to multi-size buying behavior and increased returns of all orders. To reduce the ‘buy to try’ purchase, we can review the product page and think about how to improve the fitting experience.

I also compared Matchesfashion’s current sizing solutions on product page in the clothing category with two key direct competitors. Net a Porter and Farfetch. The first one is free return, where matchesfashion and net a porter offer longer period. The second part is the model fitting photos, video and size and fit, where Matchesfashion presented most detailed information to help user decide on the size.

The fit description is roughly the same. In terms of size guide, Matchesfashion also stands out by having 11 conversions, details measurement and measuring guide, but Net a Porter offers more instant customer communication channel through live chat feature.Farfetch, on the other hand, uses machine learning technology and creates Fit Predictor that looks at the designers and sizes saved by user and past purchases, mapping with preferences from other similar shoppers to auto-recommend the right size. 

Overall, although Matchesfashion’s sizing solution is arguably the most complete and detailed among the three players, there are rooms for considering new technology as Farfetch did in assisting users to find the most suitable size.

After reviewing customer needs and competitor’s offering, I come up with 7 possible solutions to help customers find the right size more efficiently and reduce multi-size ordering.

1) switch model is where user can change the product photos to a selected model of similar body shape across the website once landing

2) multiple models’ fitting report is where we show a chart of about 5 to 10 models’ size and fit for different sizes of the clothes on product page, which is commonly seen in Asian ecommerce site and mentioned by one of the interviewee

3) customer reviews is used by online ecommerce like ASOS and SHEIN, which I suppose Matchesfashion may have thought about but there might be some reasons for not moving forward. My concerns for this solution is how to encourage user to leave useful reviews for as many product as possible, how to monitor the review quality, and whether it would be a plus for Matchesfashion’s high-end brand image.

4) personalized size recommendation is where user get a size recommendation by comparing body measurement input and purchase history to other similar user. However, it’s costly to develop machine learning technology and we may need to consider if the cost of recommendation is lower than cost of return because of sizing.

The rest is Net a Porter’s live chat, Farfetch’s fit predictor, and the rising AR/VR virtual fitting technology. But they’ve been ruled out due to low value and feasibility.

To prioritize these ideas, I use RICE scoring model to evaluate the reach, impact, confidence and effort. My first step is to evaluate how many people each feature will reach in a month. Referencing from Similar Web, I assume that Matchesfashion’s monthly visits is 4 million and 18% of them are from the UK, making it 720 thousands. As the bounce rate is 50%, my assumption is that 360 thousands of visitors would land on product page and 25% of them will fill in body measurement or brand and size for personalized size recommendation.

It is worth mentioning that customer reviews’ reach is 180 thousands because I assume that 50% of the less popular products may have no reviews. The second step is where I evaluate their impact on user goal by demoing the solutions to customer and gather quick feedback. The third step is where I estimate my confidence level based on how many data I have to back up this decision. Lastly, I consulted with an engineer to decide the effort may needed for each idea and calculate the RICE score. 

I find switch model to be the best solution not only because it has the highest score in solving sizing issue, but also because it differentiate Matchesfashion from key competitors, encourage models to advocate matchesfashion on social media for more followers, which potentially increase brand visibility, and motivate users to purchase more items presented by their favorite model.

Finally, as implementing the Switch Model feature requires website revamp and lots of efforts for models’ fitting and photoshoot, I will consider doing more user survey and usability testing to validate the needs and plan phased rollout by starting with 3 models that covers the top 10%  best selling clothing in phase 1 and review the total clicks of model fitting report as well as return rate of sizing issue to determine whether to increase to the entire user base.

For each release, I would mainly collaborate with size & fit, technology, marketing and customer care team from pre-launch to post launch period.

Roughly speaking, at pre-launch period, I will work closely with size and fit team to define model pool and get sample data for the tech team to run some test and start development. I will also involve marketing team and Customer care team during kick-off meeting to plan awareness campaigns and get ready for questions customer may have. On the day of launch, I will start tracking early KPIs and user feedbacks. After launch, I will review the performance, continue to monitor user voice and propose the next steps that will further improve customers’ sizing experience and lower the return rate.