Hello Molly digital strategist Guangyong Chuang shares the findings from a recent brand survey the online retailer sent out to its email database.
Chuang explains Hello Molly has always kept close tabs on customer sentiment and buying patterns which previously led to the creation of its Dear Emilia and Hello Molly Swimwear ranges. This new survey, however, was structured in four parts: identity, awareness, perception and consideration.
Who did the survey go out to?
The survey has so far only gone out to our email database. However, we have plans to use it on a sample audience within the target demographic outside of the Hello Molly customer base. This not only allows us to fine tune how we interact with our current customers, but also allows us to perform gap analysis and figure out what ground to cover to reach new customers.
How many respondents were there?
Around 1,200.
What were the results/some of the feedback?
Near the top of the list of what we really wanted to know was purchase frequencies. Not just what we could see, but how often a customer actually purchases new things from whatever store. This is data only credit card companies would have. We found that it's mostly around once a month. People come to us to update their wardrobe or find something for a special occasion. They also navigate our site with the release date of products, price, and style at the top of their minds.
What changes have you made since receiving the survey results?
The most intensive change we're looking to make is related to increasing discoverability of product on our website, since we have so many. Despite the fact that we have many filters, we found that there was a gap for certain types of buying behaviours. For example, browsing may be difficult for some users who are looking for a specific festival piece even though we're well known for stocking many festival worthy styles. So we're opening up a slew of new categories and attributes as well as new rules for recommendations that change depending on how they're browsing the site. This should facilitate discovery a bit better.
We're also using the results to polish the way we speak to our customers through our channels based on their style preferences and buying patterns. There will be a few design tweaks and changes to the brand to really highlight what our customers come back to us for.
How frequently do you survey customers?
We try to make an effort of doing this at least once a year in a formal way as we've done here, but we continuously have a system sending out review requests after every purchase to give us feedback at the product level. Through this, we quantify what the sizing, quality, and service was like for each product, and use sentiment analysis and natural language processing to analyse any written feedback.
Are you able to provide examples of how customer sentiment and buying patterns led you to creating Dear Emilia and Hello Molly Swimwear?
Many of our customers will talk about what they wore a specific piece to and how it made them feel. So, in a way, customers rely on us to make sure they're the best dressed at whatever event they attend. This also means what styles we stock revolve around key events for them, and we're able to see that manifest through search queries and general interest. Our Dear Emilia range has a very formal element to it, which is perfect for their most important events. We wanted to create this line as a distinction from the typical style you might find on Hello Molly and meet this demand for them.
As for swim, it's a bit simpler. We pride ourselves on being a Sydney-based, Aussie brand. Beach and swim is so ingrained in the culture that it was a very logical next step. After seeing swim interest building on our site, we wanted to curate a collection that we could be proud of.
How else are you using the data from the survey?
We've also taken another approach to the data with hierarchical clustering and unsupervised segmentation to build a dendrogram of segments. In simple terms, it's essentially letting the computer decide how to segment people according to their answers, then presenting it in a visual format so we can figure out how many groups makes sense. Then, we can analyse these groups and build personas. It's yielded some very interesting subgroups of people so far.