Do we teach the machine, or is it the machines that train us?

The next generation in voice-based search.

What will you discover from these few pages?

• How AVA (Autonomous Voice Assistant) will change the way customers shop (and the way you sell)
• Learn how to harness the power of voice search
• As we might think privacy trends are ending the era of hyper-personalization but yet our research leads to change these fears into a competitive advantage.

Are you leading or following the eCommerce crowd?

It has become cliche to talk about the continued explosive expansion of eCommerce. We’ve become used to numbers that illustrate amazing year-on-year growth. There is literally no one left who doesn’t believe that the online space is the future of retail. 

That’s why you’ve already invested in your online platform and will continue to do so. That’s why you’ve optimized every detail in the sales funnel. That’s why you have a team dedicated to the maintenance of your entire online operation. That’s why you keep up with trends, technology, and anything that will give you an edge in the most competitive marketplace on the planet. 

But are you really standing out from the crowd or just keeping up with it?

See below examples of mom day and mother's day on crowded eCommerce sites from PL and GB. It's obvious the intent is identical and results vary significantly. Empik is the largest entertainment marketplace in CEE.

Another example.

Voice search is the new growth frontier in eCommerce

Voice search is a remedy in the declining era of cookie-based personalization. The privacy revolution is here and the era of personal data scraping is ending.

Mozilla Firefox and Safari have already rejected cookies by default. Google Chrome's privacy sandbox is just around the corner and will also change methods of personalization in marketing. This has consequences in the shift from behavioral-based to search-based personalization. If you plan to benefit from hyper-personalization in the future, you should rely on search queries instead of a behavioral, cookie-based understanding of customer expectations

There is another reason to consider adding the voice-search and conversational-search to your priorities and to start seriously working on it. It is worth considering being a leader once specific tech adoption is close enough to start working on it but also far enough to be in the avant-garde. To win being before others.

Until now, search systems trained humans how to look for products. We have been taught how to distill the need in our heads into a succinct keyword so that information retrieval algorithms can match something relevant. Incredible recent progress in Natural Language Processing means we’re at the dawn of a new era for search engines. Even though we have only begun scratching human-level communication in terms of shared context or ambiguity the effects are already astonishing.

Piotr Stachowicz, edrone CTO

The numbers tell the story

• 20% of mobile queries are currently done with voice search, even with their current limitations
• 72% of people who use voice search devices claim they have become part of their daily routine
• By 2022, voice search for shopping needs will have turned into a $40 billion industry.
• Google says that 70% of searches on Google Assistant use natural language—shouldn’t your ecommerce be ready to handle the same customer queries?
• Searching with your voice is 3.7x faster than typing according to Bing.
• 52% of people who search by voice are interested in receiving info about deals, sales, and promotions from brands.
• 63% of smartphone users are more likely to purchase from companies whose mobile sites or apps offer them relevant recommendations on products they may be interested in, according to Google

Autonomous Voice Assistant - AVA

The future of voice search technology

Turning natural language into the best, most precise recommendations possible and giving you an edge in customer engagement. 

What if your eCommerce had a voice?

Voice search is the last key field of online shopping that has yet to be perfected. edrone is leading the way in research and development in this field. 

Current search options are often ignored for a simple reason—they don’t work. The strange reality is that customers prefer to type out their search even though it’s less convenient. Smartphone users can speak into their phones to get any piece of information they want but can’t do it to find something in an online store?

We’re going to fix that. 

Here’s why current voice search technology disappoints most users:

• Subjective descriptions - What does “nice” or “inexpensive” mean, exactly?
• Use of categories - When users want to look at “kitchen appliances” or “lipstick”, where do you start?
• Search by feature - Customers often want to choose from among various items with a common feature, but usually cannot.
• “Solution” searches - Why can’t we just ask for recommendations for a particular need or problem?
• Current sales and promotions - This is among the most popular categories on clickable menus, so where is the voice search equivalent?

Voice search can match actual customer habits in a way that conventional search cannot. 

Some top eCommerce sites are already putting in some effort. But keep in mind the more specific and large is the product stock the more issues need to be solved with AI.

baymard.com/research

Effective voice search assistants will promote an easier, hands-free experience for shoppers

Voice Search is mobile – and part of your everyday life [...] It’s about the conversation. [...] The choice of words used in the question provides more context about user intent, which in turn can provide advertisers with more insight into where the consumer is within the purchase funnel. Are they simply researching, or ready to purchase?

Campaign, Christi Olson, Global Media SEM Team Lead @ Google

This is not just about providing the best recommendations for each user. A truly effective voice search option will keep shoppers on your site and make shopping as easy as possible, relying only on voice commands. Your site will suddenly become more welcoming, easy to navigate, and feature a voice search option with a level of performance unmatched by competitors. 

AVA will not only increase the precision of search results, but will make them far more accessible than any current option.

The print screen above is an example of a natural language search result - Elasticsearch implementation VS AVA search.

Product attributes extracted automatically while building product graphs and linguistic corpora.

The research has begun

edrone has built a dedicated team of scientists and post Samsung or Allegro engineers focused on pushing the boundaries of voice search capabilities. Greetings from our twenty-strong AVA team. Meet our colleagues with whom you will work.

AVA remote seminary

• Prof. Grzegorz J. Nalepa, Ph.D. – Research & Development Lead – President of the Polish Artificial Intelligence Society. With over 20 years of experience in AI research, he has participated in the implementation of several research projects, including cooperation with industry (NCRD, MCP) and international (CHIST-ERA). He is a laureate of the Scientific Award of Tygodnik POLITYKA for the best scientist in the field of technical sciences (2012), the award of the Committee of Computer Science of the Polish Academy of Sciences for the best monograph in computer science (2018), and numerous scientific awards of the Rector of AGH University of Science and Technology.
• Szymon ??ski, Ph.D. – Lead Artificial Intelligence Specialist – Previously Allegro R&D Senior Research Engineer, earlier R&D engineer and manager in Samsung's NLP division. Besides a PhD in mathematical physics, he has a habilitation in biology: he was scientifically involved in both black holes and signal activity of the human brain. Author of over 20 scientific publications.
• Anna Gut – Senior Software Engineer – An experienced Python programmer, with extensive experience in software engineering for ML projects. Former Machine Learning Team Leader at 10Clouds. Co-author of publications in the Machine Learning field. 
• Leszek Bukowski, Ph.D. Artificial Intelligence Specialist – He worked as Lead Data Scientist at Emplocity, where he did research on the chatbots field. He has experience with the implementation of NCR&D backed projects. A former lecturer at PJATK. Co-authored several papers in statistics and machine learning. Even before that, he did a PhD in philosophy, in which he dealt with, among others, deontic logic.
• Arkadiusz Flinik – Senior AI Specialist, – CTO of Lekta.ai, a company creating technology for building voice assistants used by the largest Polish companies. Graduate of computer science at the University of Wroc?aw and has over 10 years of experience in IT and new technologies. 
• Piotr Stachowicz – Software Engineer Lead/CTO – has been associated with the IT and new technologies industry for 15 years. He gained his experience abroad, mainly in Ireland. Piotr has earned his experience in leading teams of programmers, building high scalability and reliability IT systems. Responsible for organizing the work of programmers, and raising the level of code and quality of work.

Understanding client queries is about understanding their language

We’ve already discovered crucial insights that will pave the way forward:

• Conversation and high-quality product search are essentially the same thing. Developing a conversational client assistant is about developing a hi-end search engine, and vice versa.
• Clients search for the same products differently. No tag or annotation can reflect a product’s nature better than an understanding of what the purpose of the search query is. To deliver high-quality search results is to develop a semantics-based search engine. 

Even if we are talking about such a popular attribute as "size", the task is challenging. A shoe size is one thing, and a t-shirt size is another. AI will not know which type of "size" to use, especially when the product is defined by several different sizes. In addition, there are also features restricted to a certain category of products… whose nomenclature in different stores is often different, and so are the customers’ queries.

Join us on our mission to perfect voice search with AVA

• Free of charge
• No integration necessary
• Privacy neutral

Getting started with AVA

• Discovery call - We want to understand your situation and expectations. Learn your insight and the way you approach search on your website
• Data snapshot - Every industry is different. We want to bow over your clients, and analyze the way search for products, regarding our knowledge end experience.
• Data analysis - Based on insights we draw from previous steps we deliver valuable insights in the form of a detailed report, we will share and discuss with you. Regardless of our fit, your time will not be wasted. 

If we’re a good fit for working together to advance AVA, we’re interested in:

• Research commitment - We are dealing R&D stuff, there is no easy answers, but they provide a huge value. It’s worth it!
• Free-of-charge cooperation model - we want to help each other. Only resource that costs you here is time. 
• Customer journey insights - To successfully analyze search engine you need to take care about detailed customer journeys. The more you know about people - your customers - the better.

Join us on the path to a new age in eCommerce

edrone is a leading e-CRM platform for the comprehensive support of ecommerce operations. From the simple elements of effective sales campaigns through full marketing automation support, edrone is a pioneering force in online retail and is dedicated to turning your site into a sales machine. 

AVA is the latest in a long line of innovations that has placed edrone among the most trusted names in eCommerce sales and marketing support. 

Partnerships that we form early in AVA’s development will be the very first to benefit from the next major breakthrough in eCommerce.

Let’s talk about taking this path together.

Thank you!

Credits

Thank you, John Adkins and Marcin Lewek for helping with this article!