Avoir Fashion
Avoir Fashion overcame a crucial obstacle in their road to business success, thanks to their collaboration with the Data Innovation Accelerator at Cardiff University.
Avoir, meaning ‘to have’ in French, summarises husband and wife, Kathy and Amir’s business idea. They told us “our website provides links to legitimate stores that sell luxury goods to automatically show shoppers the retailer offering the lowest price for a particular luxury fashion item”.
The concept originated from Kathy’s disappointment following buying a pair of designer shoes (which she’d wanted in a different colour) only to stumble across them (both cheaper and in the colour she’d originally wanted!) just days later. Feeling this was unfair, Kathy’s idea for the fashion comparison website was born.
Set up in 2017, Cardiff-based Avoir Fashion originally employed staff but now consists of just Kathy and Amir. They told us ‘not coming from a tech background has been hard. We had a very clear idea of what we wanted for our business but didn’t have the data skills to make it a reality’.
The barrier to Avoir Fashion’s success was both simple and complex and lay solely with their data capabilities – how can you compare products in the absence of a common denominator?
Accurate product comparison is fundamental to their business. But early on, through manual comparisons using spreadsheets, Kathy found that textual product descriptions varied hugely across different retailers, e.g. a colour could be described as ‘cream’ or ‘nude’ and an item could be described as ‘sneakers’ or ‘trainers’. This is truer in the fashion industry than anywhere else and can be understood when we look comparatively at say electronics, where items are easily identified via a product code.
It was the recommendation by a good friend with a connection to Cardiff University that led them to the DIA in March 2019. A meeting was set up, the problem discussed, and immediately the DIA knew this was a business they could make a real difference to.
It was Kathy and Amir who realised that the common denominator between products was really the image itself, but just how to create and harness image-recognition technology lay beyond their skill set. The DIA were eager to explore the challenge and confident that a solution could be found.
After an initial data ‘health check’, the DIA provided a comprehensive report confirming that the data could indeed be ‘normalised’ so that Avoir Fashion could successfully compare fashion items.
By using a combination of natural language processing and image recognition, Cardiff University Data Scientists were able to provide a clean data set (a ‘fashion dictionary’ if you like), creating categories for their algorithm to analyse, resulting in the accurate matching of high-end fashion items.
The collaboration ran for 9.5 months and, as in most research projects, the team faced a number of challenges. The determination that the husband and wife duo showed in the early days of their business idea, coupled with the data science and research skills of the DIA team, helped drive the collaborative to its successful conclusion.
Speaking about the relationship with the DIA, they told us “they were very nice people, very approachable with no barriers. We could be open and honest about what we wanted and how we felt. It was amazing to sit down with such clever people and be given this opportunity. It was such a positive experience and has given us the tools to continue our journey. We would recommend working with the DIA to anyone, especially start-ups like us”.
So, what’s next for Avoir Fashion? Amir told us “before working with the DIA we didn’t actually think about having a data scientist and now we realise that 2-3 are probably needed!”
Thanks to the code produced during the collaboration, Kathy and Amir will be able to develop the product comparison algorithm and fine tune it to launch a high-end, fully functioning comparison site with the plan to launch in January 2021.
Amir told us “this has opened the door for future collaborations with professors at the University, working on state-of-the-art image recognition technology, and even the possibility for us to employ Cardiff University computer science graduates. We have gone through a very difficult time, facing lots of setbacks, spending time, money, and effort, but now we feel confident for the path ahead and have the tools to make Avoir Fashion a reality, and we thank for DIA for their invaluable contribution”.