Ketan Patel, CIO of tcc global

Voice technology is currently in a transitional state of limbo. It’s certainly not perfect. I myself use Google Home and for all the times I’ve vocally implored Google to “switch off the lights”, I still, more often than not, have to pull my finger out and do it the old-fashioned way. Yet you’d be foolish to laugh it off.

True, there hasn’t been a mass market uptake yet. It’s only in recent months that the biggest players have begun to play their full competitive hands in the space—Google announcing investment funds to support start-ups in the sector, for instance.

Retailers are currently dealing with bigger and more pressing issues, from trying to attract customers in store to working out how to stay competitive in the face of a spate of recent M&A activity in the sector. But this will change as the technology itself improves. The newest versions of Amazon and Google’s smart devices are set to be easier to interact with, incorporating important design features such as screens. In a retail context, this is significant: people have always wanted to see what they’re buying and interacting with.

We’re still probably around five years from the technology really reaching its potential, yet retailers would be wise to start investing in voice search now. With more people taking it up, companies need to have the systems in place that make the shopper experience as smooth as possible when mass adoption hits. So how can they prepare?

Cleanse your data

It’s important to consider voice technology on the same terms as artificial intelligence (AI). They are essentially one and the same—just as AI needs to consistently learn from data to keep improving, so does voice technology.

Too many retailers currently have little idea with what they want to do with their data, and they will struggle to capitalise when voice technology and AI reach their zenith. Figure out a solid data strategy, cleanse any data that detracts from achieving that, and reap the rewards in the future.

Fix product hierarchy and labelling

Retailers should be thinking about standardising their terms, training their IT architecture so it can deal with incoming data structures. This starts with correcting traditional search structures. For instance, a search for “women’s jumper” I made on the website of a well-known retailer returned 100 results, but a search for “women’s woollen jumper” returned 110. This simply doesn’t make sense, but is a common problem across retailers who have used traditional systems of product listing on their e-commerce platform.

Having consistent, accurate search structures will be crucial for voice search to be effective. For that to happen, product labels and descriptions need to be standardised across all mediums—both online and offline.

Escape legacy IT systems

Finally, the most important piece of preparation a retailer can do is to make sure its systems are ready for the huge processing power this new technology will require. Analysis from OpenAI has found that the compute power needed to run AI has doubled every 3.5 months since 2012. The power needed is already staggering and will only get bigger. It’s something that outdated legacy systems simply won’t be able to handle. Frankly, there’s no point even beginning to think of utilising voice search until you have the most up-to-date technology in place.

similar articles