Is language the medium of the (retail) future? – Part 2

Will consumers be more likely to search for offers per voice in the future? Or won’t that be relevant in the shopping context?

Basically, speech input offers convenient possibilities that can improve the usability of online shops and should therefore be taken into account by retailers. Precisely because networked speakers such as Amazon’s Dot or Google Home and the vast number of newly developed devices that integrate Amazon Alexa, Apple HomePad or Google Home are getting consumers used to interacting with devices by voice.

Retailers often not only use the sales channels/marketplaces provided by Google and Amazon, but also operate their own online shops. Therefore, a fundamental strategic question must first be answered as to whether retailers should rely on the speech analysis capabilities of previously strong gatekeepers (e. g. Amazon, Google, Baidu, Alibaba). This is likely to weaken their own sales channels. On the other hand, an immense amount of effort – by specialized third party providers – is necessary to provide one’s own online shop with language ability. Furthermore, limitations on the number of available languages must be considered. Amazon Alexa speaks German, various English dialects and Japanese. Google offers six languages and has announced three additional languages. In contrast, third-party voice systems offer the widest range of languages and should therefore cover the countries relevant to you.

If the focus is currently on accumulating experience with voice control and nothing stands in the way of a distribution strategy via gatekeepers, voice control can be established on foreign marketplaces with limited effort. In particular, because it will take some time to achieve real market penetration for voice control. In the course of digitization efforts, the danger of winner-takes-all market is often evident. Thus, those who start late run the risk that they will lose further market share to Amazon.


How can retailers prepare for this?

In answering this question, we would like to focus deliberately on the use of software systems from various third-party providers and not on the integration of the dominant Gatekeepers (Amazon Alexa and Google Assistant). The implementation of an Amazon Alexa skill, for example, involves very different challenges than the selection and implementation of an independent solution provider for speech recognition/search.

When it comes to apparel retailers, it can be assumed that they do not have a large number of voice control specialists within their organization and that they will therefore rely on external experts when it comes to selecting and implementing solutions.

Basically, it can be stated that there are a multitude of integrable cloud solutions that are based on the technologies of artificial intelligence – especially Machine & Deep Learning – and that solve the problems of speech recognition. These can be integrated without the need for costly in-house development. The speech recognition rate of existing solutions is often very high, but the context comprehension (semantics) is sometimes still problematic. This leads to errors in the language representation and finding of suitable search results and is therefore an essential selection criterion.

Another crucial point is matching an interpreted search query to the design attributes of offered products. This not only requires the attribution of products (e.g. by product group, colour, size, etc.), but also, more complex search queries (e. g. “Show me an outfit similar to Paris Hilton’s outfit at the Grammy Awards”) require context-specific background information. Restricting voice search to a specific pre-defined vocabulary and having a lack of contextual information reduces the hit rate and thus the shopping experience. When selecting a solution, the aim should be to take into account a broad range of possible enquiry types as well as to learn from the experience of (non-successful) enquiries and optimize the system over time. Voice control is therefore subject to continuous further development and adaptation to changing circumstances (e.g. new products/assortments, an up-to-date vocabulary as well as changes in the technical integration with the existing online shop).


To what extent will language search also influence the search function in the online shop?

Fundamentally, the given linguistic diversity of requests from different users must be translated into accurate results of the desired or requested products. Only a good understanding of the language and appropriate product results / answers will motivate users to repeatedly use voice search.

Thus, retailers must clarify which range of queries the voice search should cover and whether voice search should go beyond a pure product selection (e. g. “Show me trousers in my size”). Particularly questions about the company’s services (e. g. “Where can I find the nearest store?”, “How long is the store opened?”, “To which countries can I have products delivered?”) can be expected. Limiting voice search to product search restricts the shopping experience and should be rethought.

Language queries enable a much greater degree of freedom than attribute-oriented product filter criteria. The language space is unlimited (users can ask any question). If there is no suitable interpretation, then the answer is disappointing (e.g. “Sorry, I can’t answer that unfortunately” or “There is no suitable product for your inquiry”).

The complexity of voice requests must be absorbed through the intelligence of the system and a conscious “voice search optimization” by means of long-tail+ keywords or rich snippets, for example. A simple keywording of products or the attribution of images is not sufficient.


“O.K. Voice Assistant, what do you recommend?”

“Dear fashion experts, voice control will be the central paradigm in the user interface design of the future and will penetrate many areas of life, including your business context. It offers you the chance to significantly change the interaction between users and your (fashion) brand. The apparel retail industry must not only rely on the current gatekeepers (Amazon, Google) of voice control systems, but must also take alternative solution providers into account so as not to cannibalize the success of its own sales channels. If you take this to heart, then you’re on the right track. See you at the exp37. Good luck and see you soon!”

Leave a Reply

Your email address will not be published. Required fields are marked *