One of CobbleWeb’s core values is radical transparency. We practise it in every step of our relationship with our audience, whether it is you reading this blog post or our interactions with long-term clients. So, despite our proven ability to significantly increase success rates for marketplace entrepreneurs, we have to admit that there are projects that fail.

That said, learning is as much key to our own evolution as it is to our clients’ projects. It is therefore valuable to share what we have learned from failed projects. Such a scenario, which offers some crucial lessons for would-be marketplace entrepreneurs, presented itself recently.

Affordable Art Fair (AAF), a hugely successful organiser of global art fairs, approached us in 2020 to help them expand their footprint digitally via an online marketplace. They presented us with an impressive set of goals: upgrade their existing platform architecture, improve the user experience across the board, and integrate several new features, such as advanced marketing tools.

Affordable Art Fair website

We were confronted by a number of challenges

The new platform architecture had to be much faster and scalable than the existing (very slow) Magento plug-and-play setup. 

Affordable Art Fair version 2.0 was built with PHP framework, Symfony, and JavaScript library, React. While Symfony provides backend scalability and extensibility, React is known for its fast rendering of frontend assets.

However, with an image-heavy site like Affordable Art Fair you need some extra tricks up your sleeve. Site speed was increased to a blazingly-fast 2 second average by using AWS cloud infrastructure, lazy loading of images, optimised API calls, and RabbitMQ message-queueing software. 

Legacy data, including 50,000 product SKUs and 13,000 existing users were successfully migrated to the new platform without compromising the integrity of data sets, such as order histories or wish lists. 

We used Sentry, an application monitoring system that supports a wide range of front- and backend languages and frameworks, such as PHP, Node.js, React, Angular, and JavaScript, to discover and prioritise errors in real-time. This allowed us to quickly spot any errors related to the migration of data from the old Magento installation to the new custom platform.

Product discovery had to be made intuitive for novice art collectors. 

We created profile-page functionality for galleries which expanded the buyer’s journey of discovery from artwork to gallery and vice versa. Visii, an AI-powered visual search application, was added to increase the relevance of search results and product recommendations.

A smart category feature allowed the AAF team to easily add product categories based on niche audiences, such as Valentine’s Day or equestrian art. Buyers could then filter their navigation in a more granular manner.

Buyers could now add individual product listings to their favourites and follow specific artists or galleries. This further refined the product recommendations they received while browsing.

The AAF marketing team required data that could help them grow and engage the platform’s target audience. 

We tracked specific user events, such as which product categories were viewed or which artists were added to wish lists, by building a data pipeline with Snowplow (collect & model data), Amazon Kinesis (capture & analyse streaming data) and Google Merchant Center.

AAF used the data to create more targeted Google, Facebook and Pinterest ads (based on different buyer personas such as Art Lover or Art Investor). This increased engagement 2 – 3 times.

Snowplow data pipeline

Since AAF wanted the marketplace to mirror their global physical footprint, we had to consider complex payment flows that involved multiple currencies, languages and shipping options.

We developed a custom shopping cart feature that could handle complex transactions that involve multiple currencies, discount rules, automated shipping estimates, variable stock levels and reminder emails.

Flexible shipping options allowed sellers to use local shipping providers instead of the default UPS options. Potential buyers were given shipping estimates when they browsed product pages to reduce cart abandonment. Shipping fees were further reduced by grouping products by seller location or packaging type. 

Custom shopping cart feature for Affordable Art online marketplace

Other features that had to be developed from scratch:

  • A promotion system for sellers that allowed them to manage a comprehensive list of variables and conditions such as price threshold, product type, number of items, and validity dates. 
  • A system was developed to flag duplicate product listings. 
  • A curation feature allowed popular galleries and artists to be displayed on appropriate site pages. 
  • Stripe Connect was used to build a variable marketplace fee structure (percentage or fixed amount which could be set globally or per seller). 
  • The new marketing and transactional communication systems incorporated multiple data sources, triggers, templates and segmentation tools.

Launch of the new platform, first results, and a goodbye

The first version of the new platform was launched in December 2020 after a six month development period. Another six months later the first tangible results came in, and it looked like we were on the right track. Turnover was up 50% year-on-year and customer spend increased by a pleasing 65%.

The growth potential was definitely there. Trends like increased ecommerce spending due to the Covid19 lockdowns, work-from-home, the rise of blockchain technology like NFTs, and VR/AR technology have all boosted the online sales of artworks. Leading art auction houses like Christie’s and Sotheby’s are investing millions in their online platforms. 

According to respected niche market insurer Hiscox, “Online art sales rocketed during the pandemic, from 4.8% in 2019 to 64% in 2020. Further strong sales growth of 72% for the first half of 2021 reported by online art sales platforms gave us projected online sales for 2021 of $13.5 billion.”

Unfortunately, this is also where the AAF team’s vision started diverging radically from our proven process. AAF wanted to fast-track their features roadmap with a global focus, while we advised them to first establish sustainable product-market fit by identifying and growing the most profitable user segments, geographical locations, and product categories. This incremental approach has worked very well for our other clients, such as FanPass

Eventually the decision was made to hand over the project to AAF’s new business development consultant who wanted to use his preferred software development team on the project. Less than six months later AAF shut down their marketplace due to lack of traction and ballooning costs. 

Notice of closure

So where did it all go wrong? And could this seeming failure have been avoided? Let’s start by looking at how online marketplace entrepreneurs should view failure.

The meaning of failure in the marketplace development process

‘Fail’ tends to be a word loaded with negative connotations. However, we’ve all heard of the term ‘failing forward’ coined by leadership guru, John C. Maxwell. In his bestselling book, 

Failing Forward: Turning Mistakes into Stepping Stones for Success, he identifies some important ways that successful people and companies approach failure:

  • Take responsibility for mistakes (don’t play the blame game)
  • Have a clear perspective (see failure as part of a learning process, rather than the end of the road)
  • Set realistic expectations (if all your goals consist of moonshots, you are setting yourself up for demoralisation)
  • Focus on strengths (don’t waste valuable time and energy on what you are not good at)
  • Embrace change (know when to pivot; most successful entrepreneurs experiment with different business ideas before finding a winning formula)

Maxwell’s strategy has been proven scientifically. Researchers at Northwestern University analysed a large data set of venture capital startup investments, spanning almost 50 years. They concluded that those who succeeded were the ones that built learning from failure into their business model. 

Another key finding was that entities that failed faster, had better chances of success. This slots in nicely with the Lean methodology CobbleWeb applies to software development, which emphasises continuous improvement in the speed and quality of deliveries.  

“You have to figure out what worked and what didn’t, and then focus on what needs to be improved instead of thrashing around and changing everything,”

– Dashun Wang, Professor of Management and Organizations at Northwestern University’s Kellogg School of Management

The bottomline, how your business responds to failure will determine its future.

5 mistakes that killed the AAF marketplace’s chances of success

#1 – Cognitive dissonance

Experts in one discipline often think they are experts in everything else. AAF’s business development consultant may have seen success in other ventures, but his lack of marketplace expertise created a loss of focus which placed the project in jeopardy.

#2 – Constantly replacing point people 

Successful marketplace projects require good communication lines between the project owners and their development partners. When there’s lots of personnel changes it creates a lack of continuity and communication breakdown.

 #3 – Not sticking to a proven process

AAF ignored the advice from their development experts with regards to following an iterative process in order to identify profitable market segments and establish product-market fit. 

Why it is important to invest in a process, rather than a product, is discussed in-depth in our blog post: Marketplace development: budget vs cost (they’re not the same).

#4 – Focusing on go-to-market strategies instead of product-market fit

AAF spent far too many resources on a shotgun global marketing strategy, instead of carefully choosing a highly engaged market, in a country such as the UK, to test their assumptions and grow profitable segments first. In order to add real value, an online marketplace extension to a successful bricks-and-mortar business should be complementary, rather than simply replicate the physical model.

#5 – KISS

Adding far too many bells and whistles and layers of complexity to the first version of their marketplace made it difficult to test core assumptions about user behaviour. It also took focus and resources away from establishing product-market fit.

We must make it clear that we are not picking on AAF. Online marketplaces are complex digital platforms that require many moving parts to run in sync for them to be successful. Many things can go wrong.  Sometimes it simply comes down to no market for the product. By illustration, another client, Moto+, had a brilliant idea to establish a rental marketplace for motorcycles similar to Uber. After a promising start, their platform was shut down due to lack of traction.

Therefore we insist on minimising our clients’ risk with a proven iterative process and do not simply churn out a static product. That’s why Minimum Viable Platforms are so valuable . . . with the caveat that they are properly implemented.