New ecommerce trends: interview with Patrick Smarzynski
The future looks promising for online sellers, especially if they want to sky-rocket their sales on international markets. The advances in technology and automated solutions will help them achieve that.
- What trends can we expect in eCommerce and cross-border trade?
- Will machines help your business achieve better results than humans?
- What’s the future of eCommerce technology?
We asked Patrick Smarzynski, CEO at Webinterpret, who has worked with eCommerce technology for 15 years.
Artificial Intelligence and machine learning meet eCommerce reality
Superfast connectivity and cloud computing are driving Artificial Intelligence (AI): the simulation of intelligent behavior in machines. As a result, more and more complex tasks are being performed by machines with less human supervision.
But it doesn’t mean machines should be left to their own devices. After all, it is the human brain that is behind how machines work and learn. Technology combined with the human touch can produce amazing results. This is the direction that technology will or should be taking.
This is what Patrick will focus on in his interview. Read on: The ecommerce future will be exciting indeed…
Patrick Smarzynski, CEO at Webinterpret
Interview with Patrick Smarzynski
Technology is advancing at an unstoppable pace. Where, do you think, is eCommerce technology going?
Patrick: To start with, as of today software requires a lot of human intervention. However, Artificial Intelligence (AI) is likely to change this in the near future. With AI becoming smarter, a lot of tasks will be performed by machines.
A subset of AI is machine learning: using algorithms to break down data, learn from it and make predictions. This way, using large amounts of data and algorithms, machines can be trained to perform specific tasks. One technique for implementing machine learning is deep learning that basically tries to reproduce how the human brain learns.
At Webinterpret we’re focused on machine learning. We use a big data analysis to successfully localize eCommerce catalogs at scale. On a monthly basis, we automatically analyze between 20 to 25 million eCommerce items.
What’s our goal? We’re striving for a fully automated solution that sellers will simply love.
We’re striving for a fully automated solution that sellers will simply love.
What are the current automated solutions for global eCommerce sellers?
Patrick: Currently retailers have to do a lot of manual work, matching product categories, sizes, attributes and so on. However, AI can carry out most of this product tagging on a larger scale. A simple raw description will be enough for AI to accurately map the right category, extract attributes, convert sizes and localize the whole content, putting global multi-channel selling in motion.
Webinterpret wants to make all necessary processes fully automated. This way a raw catalog can be imported to be fully structured, localized and listed on multiple channels. This will happen with minimum human supervision. In other words, we can leave fairly complex tasks to machines and let sellers focus on growing their businesses.
It sounds pretty simple, but are there any challenges when it comes to the fully automated solution?
Patrick: We know how to handle the process, but of course there are a few challenges for eCommerce. Think about this: each and every channel has its own rules. For instance, a title on eBay cannot exceed 80 characters. On Google Shopping in turn, sellers can submit titles containing up to 150 characters and 70 will be displayed.
There are other complexities to be taken into account. Take the fashion category and size conversion as an example. Different brands can have different conversion tables and sizes can also differ between sellers and models. This can be very confusing for both humans and machines.
Similarly, the car parts category has its own challenges when it comes to car parts compatibility. There may be cross-country differences as to the features of car parts. Additionally, the same brand can have different names, e.g. UK’s Vauxhall is known as Opel in Germany.
Of course, you must make sure to list both Vauxhall and Opel as compatible brands. Hence, Webinterpret offers full support for the car parts compatibility list feature, which has resulted in 40% better sales for our customers.
The examples I’ve mentioned show that automation isn’t a straightforward task. We need to make sure that specific rules and exceptions are recognized by machines. This is what our eCommerce experts and developers do on a daily basis.
What quality can we expect comparing humans vs. machines?
Patrick: The fact is… we must recognize the importance of both. It is humans that have the knowledge and do the thinking, but it is machines that are fast and more consistent. To make the most of both in global eCommerce, humans can teach machines how to localize/extract content. Then machines will consistently perform what they’ve learned.
Let me give you an example of a practical application. A few years ago we were affected by critical errors made by human translators, e.g. when converting sizes from one country to another. Since we taught the machine to convert sizes based on brands, the amount of critical errors has reduced significantly.
This turned out to be a progressive move. It proved that combining human expertise and machine automation delivers the best results. Human expertise is key to define the objectives of the task and machines will always perform repetitive tasks in a more consistent way.
This combined approach minimizes the number of errors and maximizes efficiency. Additionally, I’d like to highlight the value added from the HR perspective: repetitive work tends to lower human satisfaction so why not leave it to machines.
combining human expertise and machine automation delivers the best results. Human expertise is key to define the objectives of the task and machines will always perform repetitive tasks in a more consistent way.
Can you give us more examples of how machines can help humans overcome eCommerce related challenges?
Patrick: We’ve had to cope with some inconsistencies when it comes to managing our catalogs. For example, some products are available in one country, but may not be in another … for no particular reason. Some products may have a lot of attributes, some may not, which limits sales potential.
How to solve this? If a machine is properly configured, there will always be the same number of attributes and all products will be available in all relevant markets. This way, you will maximize your sales potential.
We put a lot of focus on accuracy. When we classify a product into a given category, extract information from unstructured sources, e.g. item descriptions, or extract sizes, accuracy levels of such actions are close to 100%. This means that a big part of the job can already be automated and the remaining part can be checked by humans.
Still, the future isn’t about replacing humans with machines. It’s about combining both to deliver the best results. Our eCommerce environment is a perfect example: human expertise together with continuous machine learning has enabled us to accurately generate product listings. At the same time, consistent quality has had a positive impact on buyer satisfaction.
Automation is a big benefit for retailers who can spend less time managing their catalog and more time on their core business to deliver the best products and great buyer experience.
the future isn’t about replacing humans with machines.
Any other trends you’d like to mention?
Patrick: Multichannel sales for online retailers. This means that the same retailer can offer their products via their online store and via the buyer’s favorite marketplaces. The number of interesting channels is growing, including the online giants such as Google, Facebook or Twitter. Such channels offer more and more eCommerce features.
Giving the buyer the possibility to shop directly on Facebook or a Google site equals even more sales potential. However, it’s very important to ensure great buyer experience. So on the one hand, sellers can have more social reach potential and sales opportunities, but on the other, they need to keep up with buyer expectations and the features of the new channels.
What would you recommend to online retailers, planning to grow their business in the future?
Patrick: Firstly, focus on multiple channels but … stay sensible. Each channel requires a lot of management. Each channel has its own rules and constraints. So don’t try to juggle everything at once. Even the biggest retailers we’re working with a focus on 3 to 4 main channels.
Secondly, adapt your strategy to each channel. For instance, on eBay, your conversion rate is key to visibility in top search results. Hence, use different techniques to maximize your conversion rate, for example by taking care of promotions, listings’ quality, sales history and so on.
On Amazon, in turn, the BuyBox is critical. So you must automate your repricing and deliver top customer service to get as much BuyBox share as possible. You can also deliver your own unique products or bundles to avoid fierce competition.
Thirdly, your own online store is key to developing your brand. Traffic is the lifeblood of your online store so you must master different marketing channels, such as Google Advertising, to drive plenty of relevant traffic to your website.
Finally, go cross-border. It has never been so easy and affordable to sell worldwide as with today’s technology without making too much effort. Automated processes supervised by humans are an optimal solution for most global online sellers. Test multiple markets and invest in the most profitable ones.
Are you expecting the coming years to be good for online sellers?
Patrick: Yes, I’m pretty confident that in the coming years, even more, opportunities will come up for online sellers. I wish them much success with running their eCommerce businesses.
I hope to see more online stores sell worldwide. Today most online sellers are unable to maximize their potential trading mainly on their domestic market. But never before has it been easier to sell products online on a global scale.
Cross-border trade is in full swing so this is what I wish all online sellers: unprecedented profits on domestic AND international markets!
Interested in selling your products on international markets?
Helping about 30,000 professional retailers to boost their international sales on different channels, Webinterpret has acquired unique knowledge and translated it into a unique technology. Feel free to contact us if we can help you with your expansion strategy!
Further reading
Sell more, optimising your international shipping: interview with Adrien Salvat
Strategies for growing your eCommerce sales: interview with Mark Ellis
Facts most online sellers didn’t know about eBay: interview with A. Figas
The future of international sales on Amazon: interview with M. Wejtko
Customer experience is key to long-term business success: interview with Dennis Otto
Related Posts:
Ecommerce Newsletter
By clicking the “Subscribe now” button, you agree to receive our monthly e-mail newsletter and regular marketing and commercial communications by email from Webinterpret regarding marketing trends and our digital marketing services. You confirm that you have read and agreed to Webinterpret’s Terms of Service and Privacy Policy.