Driving voice search

Machine learning can help marketers get deeper insights about who their customers are

Google has just announced voice search for eight more Indian languages — Kannada, Telugu, Tamil, Bengali, Marathi, Gujarati, Malayalam and Urdu. Prior to this, it only supported Hindi as an Indian language in voice search. It is interesting to see how would this move lead to innovation for digital marketing and brands?

India crossed one billion mobile phone subscribers in the middle of 2017 with teledensity moving upwards of 90 per cent. The Internet usage population is expected to hit 450 million in 2017. About 60 per cent of urban India uses Internet (269 million), but rural India with a population of 906 million has only 163 million Internet users. However, 78 million consumers of rural India are daily Internet users and there is a huge growth upside of close to 700-odd million with the potential to use the Internet. When we overlay this data with states having mobile population — UP (east) which has 8.87 per cent of the mobile phone subscribers of India, Maharashtra has a share of 8.13 per cent and Andhra has a share of 7.29 per cent — the need for local language-based search is only expected to grow significantly.

Majority of consumers already have a mobile device; if they need to access services through the net, voice search is the only way usage can be expanded given the economic strata they belong to. If one was a financial services institution, access to credit (business and loans or investments) will be at the touch of a voice search button. The language of the Internet will become intent-focused — “I want to know”, “I want a loan”, “I want to pay”, “I want a card” And all this will be done on the mobile. Now, the question marketers have to answer is how ready is their content for this kind of transformational marketing online. The need for cross-channel handover — online-to-offline, offline-to-online and seamless transition between them — will be the key to great customer experience. Google says 20-25 per cent of their searches are voice-based. comScore estimates over 50 per cent of all searches by 2020 will be voice-based. Therefore, algorithms in a brand’s website must be able to quickly identify the language and location of the customer to ensure satisfactory answers are provided and they don’t lose the buyer.

S Swaminathan

Marketers and brands should get ready for this kind of hyper-personalisation which will be driven by data. Consumers will go through a journey using voice search from a discovery phase to an informational and a transactional phase. During the discovery phase the need to implicitly understand and refine search engine marketing — not just using key words but recognising voice patterns will be key to driving a personalisation strategy and building contextual local content. In the subsequent phases, the need to understand behavioural patterns, expressed needs and unexpressed frustrations by understanding voice or word accentuation will lead to managing the customer experience using hybrid channels, both offline and online. The use of deep learning, AI and machine learning will rapidly rise across these phases for driving personalised experience. But the application of these techniques appropriately and also being realistic about their impact and outcomes is important to ensure appropriate intervention.

Voice search will make marketing more sophisticated but integrated. The silos that exists today will slowly begin to disappear and the need to use both the “left” brain quantitative thinking fused with “right” brained creative thinking will come together. Companies, brands and marketers who embrace this change will be the winners in this new ecosystem.

The author is co-founder & CEO, Hansa Cequity