Skip to content

Interview with Andre Brown, CEO of Locayta: Predictions for 2013

December 20, 2012

Bullish on 2013

Bullish on 2013 – watercolor by Charles Plaisted


As 2012 comes to a close, I have reached out to top technologists working on personalization solutions and technologies. I want to hear where they think the industry is headed, and how 2012 has advanced the art and science of personalization.

André Brown is co-founder and CEO of Locayta, and a 20-year veteran of the European and US software technology industry. Locayta is an ecommerce software-as-a-service provider to US and European retailers sector. Its services include product recommendations, site search and visual merchandising.

Question: How has the recommender systems market changed in 2012?
Clients are asking for the capability to apply business rules to their Product Recommendations, such as being able to apply filters to recommendations (e.g. only recommend from the same category or related-accessory category; or only recommend more expensive items).

Magento has gained enormous traction in the ecommerce market. We have seen our client-base change character from a mixture of in-house platforms and established platforms such as Venda, ATG etc, to most of the in-house platforms migrating to Magento and some of the smaller established platform clients also moving to Magento. Given that Magento is an open architecture, it’s made it very easy for Locayta to deploy a Magento extension that includes our product recommendation technology and has significantly reduced the entry cost for a retailer to implement our product recommendation technology.
In the UK we have seen retailers take a more scientific approach to deploying product recommendations – so moving away from blanket coverage of recommendations on almost every page, to a more considered approach to where on the site recommendations will actually add value. The practical implication of this, is developing a culture of testing and being willing to be agile in making changes and not holding onto the “old way of doing things.”

The move to mobile access is continuing. In the UK, a recent study (Sept’13) by the Direct Marketing Association (DMA) has predicted mobile web traffic will reach 20 per cent by Christmas. A study published by mobiThinking (June’12) shows that 22 per cent of UK mobile web users only ever use a mobile device to browse the internet. This means that whatever product recommendations are deployed, need to be device-agnostic and work equally well on mobile sites/devices.

Question: What are your proudest accomplishments for 2012?
Locayta has grown significantly, bringing a new sales team on board and signing 30+ new clients. These clients are primarily a mix of large and small UK retailers.
On the technology front, we extended Locayta Freestyle Merchandising® functionality including improvements to all of the search, recommendation and merchandising content and released a full function Magento extension.

Question: What part will recommender systems play in enabling an increasingly personalized customer experience?
Product recommendations will become more important in delivering a personalised customer experience – however, it is not as simple as just turning on a new product recommendation system. In my view, there are 5 big challenges that all such systems face:

1. Being relevant in real-time. The whole purpose of product recommendations systems is to provide relevant recommendations. This means that as stock-levels, product availability and prices change on a site, then the available recommendations should also change. However, the dilemma for many providers is that the maths involved in generating millions of recommendations is computationally very intensive and so that may force some of them to make compromises about how real-time their recommendations actually are.

2. Avoiding the impersonalisation trap. In concept, product recommendations are great as you benefit from the experience of previous users. However, there is an in-built assumption that all users are buying for themselves at all times. On that basis, it makes sense to show “people who bought X also bought Y.” However, this approach starts to break-down when users are buying for themselves and for other people – e.g. imagine a school-mum buying for herself, her husband and her daughters/sons. The kind of blind recommendations you would get from such a user, would produce out of context recommendations – e.g. a girls gymslip being recommended for a man’s work shirt. I think this trap is especially problematic for recommendation systems that are essentially “black-box” algorithms.

3. Avoiding the “self-fulfilling prophecy” trap. If product recommendations on a retailer’s site are successful, then they will increase the sales for those products being recommendation. However, this can become a self-fulfilling prophecy if only products that get recommended become popular – i.e. you are in danger of cannibalising your own online sales by skewing buying behaviour to focus too heavily on only those products that are being recommended. In addition, you may now also create a problem for those products that are new to the site not having any behavioural data to drive the product recommendation for those new products.

4. De-empowering /de-skilling the merchandisers. Many online retailers will have staff who have come from a bricks & mortar store environment and are trained merchandisers. The danger here, is that by implementing an automated product recommendation system, you will de-empower and de-skill the merchandisers, who are main the people in the business who actually know how to sell product.

5. Delivering what the business actually wants. The ultimate test for all recommendation systems, is are they delivering what the business actually wants? For those that are more black box in their approach, they may deliver a quick short-term gain, but our experience with retailers suggest that eventually all retailers will want a lot more control over product recommendations and not rely on an automated system.

Question: What does the future hold for recommender systems and personalization?
Product recommendation systems will become more popular in the future, but I believe the ones that will be successful, will be those that address the 5 challenges that I have outlined above, and strive to empower the merchandisers by giving them merchandising tools that enable them to implement their real-world merchandising skills/experience in the online world.

Advertisements

From → Personalization

One Comment
  1. Thank you for the good writeup. It in fact
    was a amusement account it. Look advanced to more added agreeable from you!
    By the way, how could we communicate?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: